[SimplifyLibCalls] Mark known arguments with nonnull

[xbolva00]

The patch shows the current idea so we can consult the design first, and then do boring marking work.

@efriedma Happy to hear your comments so I can address them now.

[xbolva00]

And I expect this will be noisy, so many tests would need to be updated :)

[efriedma]

The approach is fine.

lib/Transforms/Utils/SimplifyLibCalls.cpp

188 ↗

(On Diff #169897)

I think you're missing a check for null-pointer-is-valid metadata here.

[xbolva00]

• Updated

[xbolva00]

InstCombine does something weird for me.

define dso_local i32 @fff6(i8* nocapture readonly %s) local_unnamed_addr #5 {
entry:

%call = tail call i64 @strlen(i8* %s) #8
%conv = trunc i64 %call to i32
ret i32 %conv
}

I add nonnull for strlen so... -> call i64 @strlen(i8* nonnull %s) but after -instcombine we have undef here (@spatel ?)
define dso_local i32 @fff6(i8* nocapture readonly %s) local_unnamed_addr #5 {
entry:

ret i32 undef
}

I will look more closely to visitCallSite...

[xbolva00]

Ping @spatel whether he has any idea why instcombine turns
define dso_local i32 @fff6(i8* nocapture readonly %s) local_unnamed_addr #5 {
entry:
%call = tail call i64 @strlen(i8* nonnull %s) #8
%conv = trunc i64 %call to i32
ret i32 %conv
}

to undef

[spatel]

In D53342#1272785, @xbolva00 wrote:

Ping @spatel whether he has any idea why instcombine turns
define dso_local i32 @fff6(i8* nocapture readonly %s) local_unnamed_addr #5 {
entry:
%call = tail call i64 @strlen(i8* nonnull %s) #8
%conv = trunc i64 %call to i32
ret i32 %conv
}

to undef

I don't see this happening in the minimal case. Can you post the full IR and 'opt' command that you're using?

[xbolva00]

https://pastebin.com/8bxCuV5V + this patch
opt -instcombine p.ll -S

After I mark it, CI->dump() prints " %call = call i64 @strlen(i8* nonnull %s) #2" but later, some code turns this to undef.

[spatel]

In D53342#1272804, @xbolva00 wrote:

https://pastebin.com/8bxCuV5V + this patch
opt -instcombine p.ll -S

After I mark it, CI->dump() prints " %call = call i64 @strlen(i8* nonnull %s) #2" but later, some code turns this to undef.

You can't return the same value, or you'll invoke instcombine's RAUW function which does this:

// If we are replacing the instruction with itself, this must be in a
// segment of unreachable code, so just clobber the instruction.
if (&I == V)
  V = UndefValue::get(I.getType());

[xbolva00]

Thanks @spatel, I missed it :(

Updated patch with an alternative approach, is still @efriedma ok with it? if so, next step is update/add tests.

[efriedma]

Yes, it's fine.

[xbolva00]

• Tests updated

Clang side: https://reviews.llvm.org/D53666

[xbolva00]

• Avoid marking if address spaces

[xbolva00]

• updated test file

[xbolva00]

Maybe we can also drop nonnull attributes for certain functions with size argument, here, in instcombine?

Clang side was not so successful: https://reviews.llvm.org/D30806

[xbolva00]

• drop problematic nonnull attributes with unknown size arg

[xbolva00]

• formatting fixes

[xbolva00]

Side idea:
From C function callsites with size arg we could theoretically determine some pointer aliasing info..

[xbolva00]

Ping

[xbolva00]

Ping

[xbolva00]

Ping :)

[hfinkel]

In D53342#1299739, @xbolva00 wrote:

Ping :)

I'm not sure what you're pinging, exactly. Can you please update the patch summary to describe what this patch does.

Also, do we need D30806 to be fixed and recommitted first?

[jdoerfert]

Some comments and questions.

lib/Transforms/Utils/SimplifyLibCalls.cpp

183 ↗

(On Diff #171003)

Any reason you do not use isKnownNonZero ?

196 ↗

(On Diff #171003)

I don't get this check. Can you explain it to me (potentially in a comment)?

255 ↗

(On Diff #171003)

Why is non-null supposed to be "allowed" for the destination but not the source in case the length is 0?

416 ↗

(On Diff #171003)

It seems fragile to drop the non-null conditionally only. Shouldn't we always drop it and add it back if possible?

518 ↗

(On Diff #171003)

As above.

535 ↗

(On Diff #171003)

As above.

998 ↗

(On Diff #171003)

Here, and actually everywhere we currently look for a constant size, we could reuse the more powerful isKnownNonZero function.

[xbolva00]

I have no idea how to fix Clang side :/

[xbolva00]

lib/Transforms/Utils/SimplifyLibCalls.cpp

196 ↗

(On Diff #171003)

In some AS, dereferencing null ptr may be fine. See discussion: https://reviews.llvm.org/D53666

Maybe this is not a proper check, well, suggestions welcome

416 ↗

(On Diff #171003)

Ideally it should be on Clang side. They had tried but the solutions were reverted after miscompiles.

[hfinkel]

In D53342#1306028, @xbolva00 wrote:

I have no idea how to fix Clang side :/

Okay. Does that need to happen as a prerequisite to this patch?

[xbolva00]

In D53342#1306284, @hfinkel wrote:

In D53342#1306028, @xbolva00 wrote:

I have no idea how to fix Clang side :/

Okay. Does that need to happen as a prerequisite to this patch?

Not directly for this one, but we need it for the final goal (enable nonnull arg promotion in FunctionAttrs) so ISCCP or other passes can be more powerful.
If we fix it on Clang side, this patch would be smaller (without dropping nonnull attr)

Another thing we need to solve is a “call sinking” (if enabled, ISCCP misses many opportunities to remove null checks)

And we would be even more powerful with FunctionAttrs as a Module pass to use DT in arg attribute promotion logic (now we do it only for entry block).

[jdoerfert]

In D53342#1306336, @xbolva00 wrote:

And we would be even more powerful with FunctionAttrs as a Module pass to use DT in arg attribute promotion logic (now we do it only for entry block).

Just for some context:
I'm working on attribute interference to improve all IPOs right now.
In my experiments argument promotion is often limited by alias and dereferencability information.
Therefore I'd like this to happen so non-null information can be deduced by default which will
then improve dereferenceability information in the new interference pass.

lib/Transforms/Utils/SimplifyLibCalls.cpp

196 ↗

(On Diff #171003)

You do this check already 3 lines above ( F->nullPointerIsDefined())).

416 ↗

(On Diff #171003)

That is not a good answer to my question. Doing it in clang seems reasonable, if that is not what we want/can do we can do it here but properly. Thus, I'd argue it should not depend on the length as otherwise, we get some weird behaviors that cause us to drop it if we normalized the length to a constant before but not if we didn't.

[xbolva00]

lib/Transforms/Utils/SimplifyLibCalls.cpp

196 ↗

(On Diff #171003)

Not enough. Then we have e.g.

"call void @llvm.memset.p5i8.i64(i8 addrspace(5)* nonnull align 8 {{.*}}, i8 0, i64 40, i1 false"

As noted by Eli, "null pointers are potentially valid in non-zero address-spaces"

(test/CodeGenOpenCL/amdgpu-nullptr.cl)

416 ↗

(On Diff #171003)

In some problematic C functions, yes, we should drop this attribute always for certain parameters.

...and here we would tag it properly when we are sure that the size is > 0... we should not do it e.g. for:

strcmp(a, b, n), since n = 0 is valid and we could miscompile code due to null check removal transformation.

[jdoerfert]

lib/Transforms/Utils/SimplifyLibCalls.cpp

196 ↗

(On Diff #171003)

This is very very cryptic. First, the explicitly named check and then some condition without explanation. Maybe use NullPointerIsDefined instead as it combines both checks and it's clear what is happening.

416 ↗

(On Diff #171003)

strcmp(a, b, n), since n = 0 is valid and we could miscompile code due to null check removal transformation.

Afaik, null pointers are never "valid" but often assumed to be valid. That said, I don't see why you want to drop non-null only conditionally. Your very argument is that for a lenght of 0 we should not have non-null, however, your patch actually keeps it in this and other cases for which != 0 was not shown.

[xbolva00]

From some libc function we can get more info:

memcmp(p, d, 16)

Both p and d can be annotated with derefencable(16), right?

@efriedma

[jdoerfert]

In D53342#1604718, @xbolva00 wrote:

From some libc function we can get more info:

memcmp(p, d, 16)

Both p and d can be annotated with derefencable(16), right?

@efriedma

For functions like tihs, if the size is != null, we know the pointers are nonnull. And, yes, if the size is a constant, we know that is the dereferenceable number.

[jdoerfert]

lib/Transforms/Utils/SimplifyLibCalls.cpp

187 ↗

(On Diff #171003)

No need to check first.

[xbolva00]

lib/Transforms/Utils/SimplifyLibCalls.cpp

255 ↗

(On Diff #171003)

s = NULL;
strncat(d, s, 0) -> valid

The strncat() function is similar, except that

• it will use at most n bytes from src; and
• src does not need to be null-terminated if it contains n or more bytes.
  
  As with strcat(), the resulting string in dest is always null-termi‐ nated.

so "d" (dest) is always null terminated (and must be nonnull).

[xbolva00]

Updated implementation.
Updated/added tests ... uff :D

[xbolva00]

@jdoerfert please take a look again (yea, big diff :( )

[xbolva00]

update

[xbolva00]

lib/Transforms/Utils/SimplifyLibCalls.cpp

1149 ↗

(On Diff #215872)

@spatel @jdoerfert @lebedev.ri infinite loop here

https://github.com/llvm/llvm-project/blob/master/llvm/test/Transforms/InstCombine/memset-1.ll#L63
If I add return CI; -> memset is removed:

define float* @pr25892(i32 %size) #0 {
; CHECK-LABEL: @pr25892(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[CALL:%.*]] = tail call i8* @malloc(i32 [[SIZE:%.*]]) #0
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i8* [[CALL]], null
; CHECK-NEXT: br i1 [[CMP]], label [[CLEANUP:%.*]], label [[IF_END:%.*]]
; CHECK: if.end:
; CHECK-NEXT: [[BC:%.*]] = bitcast i8* [[CALL]] to float*
; CHECK-NEXT: br label [[CLEANUP]]
; CHECK: cleanup:
; CHECK-NEXT: [[RETVAL_0:%.*]] = phi float* [ [[BC]], [[IF_END]] ], [ null, [[ENTRY:%.*]] ]
; CHECK-NEXT: ret float* [[RETVAL_0]]
;

Maybe you can give me a hint what is wrong? Thanks

[xbolva00]

lib/Transforms/Utils/SimplifyLibCalls.cpp

1149 ↗

(On Diff #215872)

Anyway, we dont need to drop removeNonNull(CI, {0}); for llvm.memset.

It is only issue for libc's memset, and we catch it, e.g:

if (LenC && LenC->getZExtValue() > 0) {

  annotateDereferenceableBytes(CI, {0}, LenC->getZExtValue());
  annotateNonNull(CI, {0});
} else if (!isIntrinsic) {
   removeNonNull(CI, {0});

}

But I would like to know what is going here :)

[xbolva00]

Do not remove nonnull from intrinsics.

[xbolva00]

lib/Transforms/Utils/SimplifyLibCalls.cpp

1149 ↗

(On Diff #215872)

And maybe we can (after this patch) propose to annotate llvm.memset/memcpy/memmove with nonnull in td file? @jdoerfert

With -OO no opts run -> no problem even with nonnull attrs.
-O1 and higher:
InstCombine drops nonnull attrs when nonconstant size and then converts them to intrinsics -> seems fine and safe for me.

[xbolva00]

annotate snprintf if N is C > 0.

[jdoerfert]

I added initial comments, will have to go through again though.

lib/Transforms/Utils/SimplifyLibCalls.cpp

518 ↗

(On Diff #171003)

I'm confused, here and before, you annotate and later, under some condition, remove the annotation again. This seems risky.

Here for example, Len = 0 will not remove the attribute but an unknown Len will. I doubt that is intentional.

214 ↗

(On Diff #215916)

Maybe make it clear that this annotates nonnull only if it is not a valid pointer, e.g., name this annotateNonNullBasedOnAccess or sth.

217 ↗

(On Diff #215916)

Please do not use this method, it is super confusing. I'd argue you should use the getCallee anyway but if you think the surrounding function is the one you need, I'd prefer getCaller.

225 ↗

(On Diff #215916)

If you do not keep track (=statistics), there is probably no need to check first, as below, I think.

232 ↗

(On Diff #215916)

I don't think you need to check first.

518 ↗

(On Diff #215916)

What about this one?

678 ↗

(On Diff #215916)

I would argue we actually want to annotate deref all the time but do also do nonnull while we are at it. (The Attributor derives nonnull from deref if appropriate.) Long story short, why do we derive nonnull but not deref(1) here and in other places where we expect a \0 char?

(We could also add deref(N) if we already computed the static string size.)

2381 ↗

(On Diff #215916)

We could even use "isKnownNonZero" or other helper functions here.

[xbolva00]

lib/Transforms/Utils/SimplifyLibCalls.cpp

518 ↗

(On Diff #171003)

Len = 0 means we return Dst;

I need to reorder the code here a bit..

678 ↗

(On Diff #215916)

deref(N) for string constants?

• useless, no?

And, isn't it a attributor's work? get GEP with bound N and infer deref(N)

---

I agree with deref(1).

[xbolva00]

lib/Transforms/Utils/SimplifyLibCalls.cpp

217 ↗

(On Diff #215916)

Ok, getCaller.

CallInst has no getCallee.

[jdoerfert]

lib/Transforms/Utils/SimplifyLibCalls.cpp

518 ↗

(On Diff #171003)

Oh, this is all confusing (as I said).

217 ↗

(On Diff #215916)

getCallee -> getCalledFunction

I still think the callee is what is important here (conceptually), though I doubt it makes a difference in practice.

678 ↗

(On Diff #215916)

deref(N) for string constants?

Yes, or more general, for everything we know a minimal "length".

Useless?

I don't think so, but see below.

Strictly needed?

No. Though, we do already call GetStringLength all the time here, why not manifest the information?

And, isn't it a attributor's work? get GEP with bound N and infer deref(N)

It does, yes.

nonnull -> deref(1)

That doesn't happen in the Attributor right now but it should because deref(1) can be combined (e.g., in a phi) with null to deref_or_null(1) while nonnull cannot be combined with null to anything.

[xbolva00]

lib/Transforms/Utils/SimplifyLibCalls.cpp

2381 ↗

(On Diff #215916)

This is true but since we we need to compute “N” anyway, I think isKnownNonZero is a overkill a bit, since it computes what we already have - “N”. - should I use it even here?

But I could use it in some places, yes, (memcpy/memmove/memset)..

[jdoerfert]

lib/Transforms/Utils/SimplifyLibCalls.cpp

2381 ↗

(On Diff #215916)

For a lot of calls a nonnull size implies other stuff, especially, noalias, deref_or_null.
I think this is worth a separate patch but doing better nonnull checks makes sense because any value (not necessarily a constant) that is not zero will allow us to infer (1) nonnull, (2) noalias, (3) deref_or_null. If we want the constant size for some other transformations we should first look for a constant and only as a fallback call isknownnonzero.

[xbolva00]

Found a bug in LLVM.

We do not copy attributes, it seems.

define i8* @memset_size_select4(i1 %b, i8* %ptr) {
; CHECK-LABEL: @memset_size_select4(
; CHECK-NEXT: [[SIZE:%.*]] = select i1 [[B:%.*]], i32 10, i32 50
; CHECK-NEXT: call void @llvm.memset.p0i8.i32(i8* align 1 [[PTR:%.*]], i8 0, i32 [[SIZE]], i1 false)
; CHECK-NEXT: ret i8* [[PTR]]
;

%size = select i1 %b, i32 10, i32 50
%memset = tail call i8* @memset(i8* nonnull dereferenceable_or_null(40) %ptr, i32 0, i32 %size)
ret i8* %memset

}

Fixed with

CallInst *NewCI = B.CreateMemSet(CI->getArgOperand(0), Val, Size, 1);
NewCI->setAttributes(CI->getAttributes());

I will update this patch soon.

lib/Transforms/Utils/SimplifyLibCalls.cpp

2381 ↗

(On Diff #215916)

Until I start heavy mechanical work here I would like to know if this approch is acceptable.

Value *LibCallSimplifier::optimizeMemSet(CallInst *CI, IRBuilder<> &B,

                                       bool isIntrinsic) {
Value *Size = CI->getArgOperand(2);
if (ConstantInt *LenC = dyn_cast<ConstantInt>(Size)) {
  annotateDereferenceableBytes(CI, {0}, LenC->getZExtValue());
  annotateNonNull(CI, {0});
} else if (isKnownNonZero(Size, DL)) {
  annotateDereferenceableBytes(CI, {0}, 1);
  annotateNonNull(CI, {0});
} else if (!isIntrinsic) {
  removeNonNull(CI, {0});
}

Now we catch even this case :)
define i8* @memset_size_select(i1 %b, i8* %ptr) {
; CHECK-LABEL: @memset_size_select(
; CHECK-NEXT: [[SIZE:%.*]] = select i1 [[B:%.*]], i32 10, i32 50
; CHECK-NEXT: call void @llvm.memset.p0i8.i32(i8* nonnull align 1 dereferenceable(1) [[PTR:%.*]], i8 0, i32 [[SIZE]], i1 false)
; CHECK-NEXT: ret i8* [[PTR]]
;

%size = select i1 %b, i32 10, i32 50
%memset = tail call i8* @memset(i8* nonnull %ptr, i32 0, i32 %size) #1
ret i8* %memset

}

define i8* @memset_size_select_or(i1 %b, i8* %ptr, i32 %v) {
; CHECK-LABEL: @memset_size_select_or(
; CHECK-NEXT: [[SIZE:%.*]] = ashr i32 -2, [[V:%.*]]
; CHECK-NEXT: call void @llvm.memset.p0i8.i32(i8* nonnull align 1 dereferenceable(1) [[PTR:%.*]], i8 0, i32 [[SIZE]], i1 false)
; CHECK-NEXT: ret i8* [[PTR]]
;

%size = ashr i32 -2, %v
%memset = tail call i8* @memset(i8* nonnull %ptr, i32 0, i32 %size) #1
ret i8* %memset

}

for select of constants we can do better job than deref(1)

if (ConstantInt *LenC = dyn_cast<ConstantInt>(Size)) {

  annotateDereferenceableBytes(CI, {0}, LenC->getZExtValue());
  annotateNonNull(CI, {0});
} else if (isKnownNonZero(Size, DL)) {
  const  APInt *X, *Y;
  uint64_t DerefMin = 1;
  if (match(Size, m_Select(m_Value(), m_APInt(X), m_APInt(Y)))) {
    Min = std::min(X->getZExtValue(), Y->getZExtValue());
        annotateDereferenceableBytes(CI, {0}, DerefMin);

}

  annotateNonNull(CI, {0});
} else if (!isIntrinsic) {
  removeNonNull(CI, {0});
}

Ok?

[jdoerfert]

Can you use ` (three backticks) to enclose the code regions for better formatting?

[xbolva00]

New implementation.

[xbolva00]

For some cases I would like to copy just attributes for argno 0 - but I dont know to do it :(

I mean something like this:
NewCI->setAttrubutes(0, CI->getAttributes(0))

Maybe you can help me?

test/Transforms/InstCombine/mem-deref-bytes.ll

76 ↗

(On Diff #216194)

@jdoerfert please check this "side-effect" improvement while working on nonnull annotation..

if "null-pointer-is-valid"="true" and "nonnull dereferenceable_or_null(40)" -> nonnull dereferenceable(40) is ok?

if nonnull is missing, see memcmp_const_size_update_deref6,

See newly added "CI->paramHasAttr(ArgNo, Attribute::NonNull) check" in annotateDereferenceableBytes.

[jdoerfert]

In D53342#1637822, @xbolva00 wrote:

For some cases I would like to copy just attributes for argno 0 - but I dont know to do it :(

I mean something like this:
NewCI->setAttrubutes(0, CI->getAttributes(0))

Maybe you can help me?

Take the attribute list (getAttributes()) then getParamAttributes(ArgNo), then probably foreach addParamAttr(NewArgNo), and then set the attribute list. Make sure to always check that you update the list properly.

[jdoerfert]

test/Transforms/InstCombine/mem-deref-bytes.ll

76 ↗

(On Diff #216194)

This looks fine to me, also test6. nonnull + dereferenceable_or_null(X) is always dereferenceable(X) regardless of the validity of null.

[xbolva00]

Copy arg attributes.

[xbolva00]

Ok, thanks.

In terms of scope of annotations, I think it is good enough.

Now we have to check if all test changes are OK.

[xbolva00]

test/Transforms/InstCombine/mem-deref-bytes.ll

76 ↗

(On Diff #216194)

Thanks!

[xbolva00]

Fixed off by one bug.

[xbolva00]

Passes test suite.

[xbolva00]

I think I can tag many libc functions without size argument with nonnull args directly in “BuildLibCalls” - better approach.

New patch soon.

[xbolva00]

In D53342#1638343, @xbolva00 wrote:

I think I can tag many libc functions without size argument with nonnull args directly in “BuildLibCalls” - better approach.

New patch soon.

Or should we assume if NullPointerIsDefined = true that strlen could work with null pointer?

[xbolva00]

Anway if libc tags with nonnull, like int memcmp(void * nonnull dst, void * nonnull src, size_t n) this applies to function signature, not to specific callsites so this fix will not work, indeed. We need drop nonnull from signatures in BuildLibCalls.

I think we just doing shitty workarounds and gcc 4.9 optimizes it fully since ptr cannot be null even if size is 0.
And we have sanitizer check to do that:
https://reviews.llvm.org/D9673?id=25492

Jakub Jelinek at GCC Bugzilla:

C says e.g. for memchr:
"The memchr function locates the first occurrence of c (converted to an unsigned
char) in the initial n characters (each interpreted as unsigned char) of the object pointed to by s."
and elsewhere:
"If a null pointer constant is converted to a pointer type, the resulting pointer, called a null pointer, is guaranteed to compare unequal to a pointer to any object or function."
As memchr first argument must point to an object and NULL does not point to any object, NULL is not a valid value. It is similar to memcpy (NULL, NULL, 0) or memset (NULL, 0, 0) etc. being invalid.”

Sanitizer warns so I dont know why se should permit UB.

[xbolva00]

Maybe @aaron.ballman could help us

• whether Jakub Jelinek is right
• whether memcpy/memmove/memset with NULL dst/src pointers is UB (even for size = 0)

[xbolva00]

memcpy(0,0,0) is UB after all

https://stackoverflow.com/questions/5243012/is-it-guaranteed-to-be-safe-to-perform-memcpy0-0-0

So we can move nonnull annotation to BuildLibCalls.

[joerg]

Take a look at the list archives for the discussion about attribute nonnull and nullable a while ago. The short version is that the library interface specification might have a wildcard clause that NULL pointers are invalid unless explicitly allowed, but it works perfectly fine on any sensible implementation except the ds9k. When the combination of glibc annotations and aggressive gcc optimizations started, it pointlessly broke a lot of code for no good reason.

[xbolva00]

It is still a UB and UBSanitizer warns. If somebody uses (typical combo) clang and gcc, one may be very surprised why debug/dev (with clang) builds work and release (with gcc) builds not.

While in 2014 or so there was no way to find this UB in codebases, we have now UB Sanitizer. In 2019 there is no excuse "it is a UB but it still kinda work (except gcc, ds9k, ..)". Add simple "if" - problem solved. For general case, anybody can send a paper to specify this behaviour to C committee.

Now, because of "memcpy(NULL, NULL, 0) fans" we cannot optimize null checks (imagine you have safe wrapper (which always check if nonnull ptrs) above standard string functions..) - "no good reason". ?
I see a lot of opportunity from nonnull/dereferenceable annotation so I dont see why we should not to do it (ok, few codes may break - any sane codebase is checked by sanitizers anyway).

[joerg]

Please read the discussion. The general consensus was and likely is that this is at least somewhat bogus in the standard at best and the cost of not artificially breaking code is much higher than the benefit. Yes, GCC made different choices, but that's no excuse.

[xbolva00]

And nobody wrote a paper to fix “the bogus”.

@jdoerfert what do you think?

As I see, annotating callsites is atm pointless since FunctionAttrs/Attributor cannot propagate it - they do only if “function signature (decl)” contains nonnull. I would continue to work on this current “safe” approach if there is a small promise that Attributor will handle this kind of propagation too :)

[xbolva00]

I would like to hear @hfinkel’s opinion too :)

[xbolva00]

In D53342#1639205, @xbolva00 wrote:

And nobody wrote a paper to fix “the bogus”.

@jdoerfert what do you think?

As I see, annotating callsites is atm pointless since FunctionAttrs/Attributor cannot propagate it - they do only if “function signature (decl)” contains nonnull. I would continue to work on this current “safe” approach if there is a small promise that Attributor will handle this kind of propagation too :)

@uenoku informed me that Attributor will handle it with a WIP patch which is not yet commited. Good.

So this patch has a sense in terms of global scope. PTAL as I finished here all annotation work.

[hfinkel]

In D53342#1639210, @xbolva00 wrote:

I would like to hear @hfinkel’s opinion too :)

As I recall, there are two issues. One is: do we add the attributes based on the function names. The second is: What to do with the attributes that already exist in, e.g., some glibc header files.

My opinion is that:

1. We should definitely have this optimization capability
2. But we should have this after we filter out the annotations from memcpy/memset (etc.) in Clang (when the size is dynamic or zero). There should be a flag to turn off the filtering, but at least for now, it should be on by default.
3. We should have a function attribute reflecting the same filtering setting to appropriately control adding the attributes in the optimizer.

Yes, NULL pointers with memcpy/memset are UB from the standards perspective, but it's unclear to me that the value in exploiting this UB outweighs the potential problems introduced. While I generally agree with the viewpoint that says that people should use sanitizers, etc., and believe that they should in this case as well, because memcpy/memset are *so* common, because there's a wide-spread and reasonable disagreement as to whether this behavior should be undefined at all, and because there's no strong and broad case for performance improvements (unlike, for example, exploiting the lack of signed-integer overflow), I think that we should filter.

[xbolva00]

do we add the attributes based on the function names.

I dont see this as a problem - reserved names.

What to do with the attributes that already exist in, e.g., some glibc header files.

Can you point me on real example/OS/glibc version where this actually happens?

I have Ubuntu 18 LTS and glibc and I see no such attributes in IR.

[xbolva00]

But we should have this after we filter out the annotations from memcpy/memset (etc.) in Clang (when the size is dynamic or zero)

I dont think we want add/remove attributes in Clang based on size argument. It seems just bad to look at function names in Clang and go thru Clang AST to inspect size argument.

Either drop all and possibly add it back here in instcombine or leave it and drop it in eg. BuildLibCalls?

[xbolva00]

• these are libc functions and almost all C code is compiled by gcc, I think if the code triggered UB, gcc 4.9 and newer already exploited this UB -> UB releaved by “miscompile” or GCC UB sanitizer
• small percentange of C code which is compiled only by clang is hopefully tested with sanitizer -> UB revealed

On the other side the cost of patching Clang and llvm (+ maintenance) to keep this UB working in the user code (I would be very interested to see how many % of all code it could broken in 2019 - I believe very very small percentange) seems quite high.

[hfinkel]

In D53342#1640996, @xbolva00 wrote:

do we add the attributes based on the function names.

I dont see this as a problem - reserved names.

What to do with the attributes that already exist in, e.g., some glibc header files.

Can you point me on real example/OS/glibc version where this actually happens?

I have Ubuntu 18 LTS and glibc and I see no such attributes in IR.

The glibc header files have the attributes, AFAIKT: https://github.com/bminor/glibc/blob/master/string/string.h - and this looks like the header on the Fedora 27 system that I just checked.

/* Copy N bytes of SRC to DEST.  */
extern void *memcpy (void *__restrict __dest, const void *__restrict __src,
                     size_t __n) __THROW __nonnull ((1, 2));

If you're not seeing the attributes in the IR, and I'm not seeing them either, we might just not be adding them to avoid this issue. I do see them in Clang's AST:

|-FunctionDecl x </usr/include/string.h:42:1, /usr/include/sys/cdefs.h:289:63> /usr/include/string.h:42:14 used memcpy 'void *(void *restrict, const void *restrict, size_t)' extern
| |-ParmVarDecl x <col:22, col:39> col:39 __dest 'void *restrict'
| |-ParmVarDecl x <col:47, col:70> col:70 __src 'const void *restrict'
| |-ParmVarDecl x <line:43:8, col:15> col:15 __n 'size_t':'unsigned long'
| |-NoThrowAttr x </usr/include/sys/cdefs.h:55:35>
| `-NonNullAttr x <line:289:44, /usr/include/string.h:43:44> 1 2

In D53342#1641039, @xbolva00 wrote:

• these are libc functions and almost all C code is compiled by gcc, I think if the code triggered UB, gcc 4.9 and newer already exploited this UB -> UB releaved by “miscompile” or GCC UB sanitizer
• small percentange of C code which is compiled only by clang is hopefully tested with sanitizer -> UB revealed

On the other side the cost of patching Clang and llvm (+ maintenance) to keep this UB working in the user code (I would be very interested to see how many % of all code it could broken in 2019 - I believe very very small percentange) seems quite high.

You seem to be assuming that it would be obvious that a given program would be negatively affected by this, and I don't think this would be obvious at all. Sanitizers are great tools for debugging problems, but they're dynamic tools, so they only test the code paths and the parts of the configuration space that the instrumented program actually executes. Best practice may be to fuzz test all code with sanitizers, which can provide more confidence in finding problems like this, but only a very small percentage of the code that exists has been tested in this way (or has been subjected to any suitable, effective static analysis).

I look at this as something about which reasonable people can disagree. No one really has data on this, AFAIK, and so I just have my judgement. I suspect that the percentage of affected programs here is small, but not tiny. memcpy is a very-commonly-used function, and so I suspect we're in a situation where a small (but not tiny) percentage of a very large amount of code yields a large number of affected programs.

[xbolva00]

If Clang already drops it [0], that is good, then this patch is totally fine.

[0] @aaron.ballman, @rsmith, @rjmcall - can you tell us more what Clang actually does here?

[hfinkel]

If Clang already drops it [0], that is good, then this patch is totally fine.

But this patch would add the nonnull attributes back, no? I'm not sure that's fine.

[xbolva00]

Yes, but only when size is constant non zero (isKnownNonZero)

[hfinkel]

In D53342#1642707, @xbolva00 wrote:

Yes, but only when size is constant non zero (isKnownNonZero)

Ah. Indeed. I missed that part of the logic rearrangement. I agree, that's fine.

[xbolva00]

Rebased

[xbolva00]

Rebased, @jdoerfert please take a look :)

[xbolva00]

We should ditch “removeNonNull” handling since Clang never annotates callsites (Clang only annotates function signatures - currently nonnull is dropped, great!)

[xbolva00]

In D53342#1654084, @xbolva00 wrote:

We should ditch “removeNonNull” handling since Clang never annotates callsites (Clang only annotates function signatures - currently nonnull is dropped, great!)

What do you think, @jdoerfert ? removeNonNull sometimes caused instcombine’s infinite loop + i think we dont need removeNonNull at all.

I would like to move forward here..

[jdoerfert]

Sorry for the delay but it will take me a while to look through all the changes and make sure we did not overlook sth.

[jdoerfert]

Some minor nits, two questions that you could clarify or resolve, other than that I think this is fine.

lib/Transforms/Utils/SimplifyLibCalls.cpp

638 ↗

(On Diff #218260)

Somewhere above and somewhere below you copy all attributes, here only the param 0 attributes, is there a reason why?

1084 ↗

(On Diff #218260)

Leftover call.

2693 ↗

(On Diff #218260)

Move to the beginning of the function, puts has often no users (IMHO).

3111 ↗

(On Diff #218260)

Is there a reason this is not happening for the two functions below?

test/Transforms/InstCombine/strstr-1.ll

91 ↗

(On Diff #218260)

Add a newline at the end.

[xbolva00]

lib/Transforms/Utils/SimplifyLibCalls.cpp

638 ↗

(On Diff #218260)

I think it crashed for me. I will check it again.

3111 ↗

(On Diff #218260)

Oops, thanks! I will adjust those places too.

[xbolva00]

Btw

We should ditch “removeNonNull” handling since Clang never annotates callsites (Clang only annotates function signatures - currently nonnull is dropped, great!)

I will go this way and remove it. removeNonNull may pessimize things / create compiler infinite loops. Some other may infer that pointer X is nonnull and then InstCombine would remove it in strncpy(D, X, N); // non const N

Imagine
%add.ptr = getelementptr inbounds i8, i8* %string_literal, i64 1

%call2 = call i64 @strlen(i8* nonnull dereferenceable(1) %string_literal) #7
%sub3 = add i64 %call2, -2
%call4 = call i8* @strncpy(i8* nonnull %call1, i8* nonnull %add.ptr, i64 %sub3) #8

This is IR which cause instcombine infinite loop. InstCombine sees that sub3 is not "known zero" and will remove nonnull from add.ptr. As I noted, I am not worried if we remove "removeNonNull", since Clang does not annotate call sites, Clang just emit function signatures (and these do not contain "nonnulls").

[xbolva00]

Addressed review comments.

[xbolva00]

Removed left over call.

[xbolva00]

Hi @jdoerfert

Do you plan to adjust Attributor's place in opt pipeline? Attributor must run after InstCombine (too) to propagate nonull/deref attributes from callsites to function arguments.

llvm/trunk/include/llvm/Transforms/Utils/SimplifyLibCalls.h

private:
   Value *optimizeStrNCpy(CallInst *CI, IRBuilder<> &B);
   Value *optimizeStrLen(CallInst *CI, IRBuilder<> &B);
   Value *optimizeStrPBrk(CallInst *CI, IRBuilder<> &B);
   Value *optimizeStrTo(CallInst *CI, IRBuilder<> &B);
   Value *optimizeStrSpn(CallInst *CI, IRBuilder<> &B);
   Value *optimizeStrCSpn(CallInst *CI, IRBuilder<> &B);
   Value *optimizeStrStr(CallInst *CI, IRBuilder<> &B);
   Value *optimizeMemChr(CallInst *CI, IRBuilder<> &B);
+  Value *optimizeMemRChr(CallInst *CI, IRBuilder<> &B);
   Value *optimizeMemCmp(CallInst *CI, IRBuilder<> &B);
   Value *optimizeBCmp(CallInst *CI, IRBuilder<> &B);
   Value *optimizeMemCmpBCmpCommon(CallInst *CI, IRBuilder<> &B);
   Value *optimizeMemPCpy(CallInst *CI, IRBuilder<> &B);
-  Value *optimizeMemCpy(CallInst *CI, IRBuilder<> &B, bool isIntrinsic = false);
-  Value *optimizeMemMove(CallInst *CI, IRBuilder<> &B, bool isIntrinsic = false);
-  Value *optimizeMemSet(CallInst *CI, IRBuilder<> &B, bool isIntrinsic = false);
+  Value *optimizeMemCpy(CallInst *CI, IRBuilder<> &B);
+  Value *optimizeMemMove(CallInst *CI, IRBuilder<> &B);
+  Value *optimizeMemSet(CallInst *CI, IRBuilder<> &B);
   Value *optimizeRealloc(CallInst *CI, IRBuilder<> &B);
   Value *optimizeWcslen(CallInst *CI, IRBuilder<> &B);
   // Wrapper for all String/Memory Library Call Optimizations
   Value *optimizeStringMemoryLibCall(CallInst *CI, IRBuilder<> &B);
 
   // Math Library Optimizations
   Value *optimizeCAbs(CallInst *CI, IRBuilder<> &B);
   Value *optimizePow(CallInst *CI, IRBuilder<> &B);

llvm/trunk/lib/Transforms/Utils/SimplifyLibCalls.cpp

if (CI->getFunction()->hasFnAttribute(Attribute::SanitizeMemory))
     return false;
 
   return true;
 }
 
 static void annotateDereferenceableBytes(CallInst *CI,
                                          ArrayRef<unsigned> ArgNos,
                                          uint64_t DereferenceableBytes) {
-  const Function *F = CI->getFunction();
+  const Function *F = CI->getCaller();
   if (!F)
     return;
   for (unsigned ArgNo : ArgNos) {
     uint64_t DerefBytes = DereferenceableBytes;
     unsigned AS = CI->getArgOperand(ArgNo)->getType()->getPointerAddressSpace();
-    if (!llvm::NullPointerIsDefined(F, AS))
+    if (!llvm::NullPointerIsDefined(F, AS) ||
+        CI->paramHasAttr(ArgNo, Attribute::NonNull))
       DerefBytes = std::max(CI->getDereferenceableOrNullBytes(
                                 ArgNo + AttributeList::FirstArgIndex),
                             DereferenceableBytes);
   
     if (CI->getDereferenceableBytes(ArgNo + AttributeList::FirstArgIndex) <
         DerefBytes) {
       CI->removeParamAttr(ArgNo, Attribute::Dereferenceable);
-      if (!llvm::NullPointerIsDefined(F, AS))
+      if (!llvm::NullPointerIsDefined(F, AS) ||
+          CI->paramHasAttr(ArgNo, Attribute::NonNull))
         CI->removeParamAttr(ArgNo, Attribute::DereferenceableOrNull);
       CI->addParamAttr(ArgNo, Attribute::getWithDereferenceableBytes(
                                   CI->getContext(), DerefBytes));
     }
   }
 }
 
+static void annotateNonNullBasedOnAccess(CallInst *CI,
+                                         ArrayRef<unsigned> ArgNos) {
+  Function *F = CI->getCaller();
+  if (!F)
+    return;
+
+  for (unsigned ArgNo : ArgNos) {
+    if (CI->paramHasAttr(ArgNo, Attribute::NonNull))
+      continue;
+    unsigned AS = CI->getArgOperand(ArgNo)->getType()->getPointerAddressSpace();
+    if (llvm::NullPointerIsDefined(F, AS))
+      continue;
+
+    CI->addParamAttr(ArgNo, Attribute::NonNull);
+    annotateDereferenceableBytes(CI, ArgNo, 1);
+  }
+}
+
+static void annotateNonNullAndDereferenceable(CallInst *CI, ArrayRef<unsigned> ArgNos,
+                               Value *Size, const DataLayout &DL) {
+  if (ConstantInt *LenC = dyn_cast<ConstantInt>(Size)) {
+    annotateNonNullBasedOnAccess(CI, ArgNos);
+    annotateDereferenceableBytes(CI, ArgNos, LenC->getZExtValue());
+  } else if (isKnownNonZero(Size, DL)) {
+    annotateNonNullBasedOnAccess(CI, ArgNos);
+    const APInt *X, *Y;
+    uint64_t DerefMin = 1;
+    if (match(Size, m_Select(m_Value(), m_APInt(X), m_APInt(Y)))) {
+      DerefMin = std::min(X->getZExtValue(), Y->getZExtValue());
+      annotateDereferenceableBytes(CI, ArgNos, DerefMin);
+    }
+  }
+}
+
 //===----------------------------------------------------------------------===//
 // String and Memory Library Call Optimizations
 //===----------------------------------------------------------------------===//
 
 Value *LibCallSimplifier::optimizeStrCat(CallInst *CI, IRBuilder<> &B) {
   // Extract some information from the instruction
   Value *Dst = CI->getArgOperand(0);
   Value *Src = CI->getArgOperand(1);
+  annotateNonNullBasedOnAccess(CI, {0, 1});
 
   // See if we can get the length of the input string.
   uint64_t Len = GetStringLength(Src);
-  if (Len == 0)
+  if (Len)
+    annotateDereferenceableBytes(CI, 1, Len);
+  else
     return nullptr;
   --Len; // Unbias length.
 
   // Handle the simple, do-nothing case: strcat(x, "") -> x
   if (Len == 0)
     return Dst;
 
   return emitStrLenMemCpy(Src, Dst, Len, B);
B.CreateMemCpy(CpyDst, 1, Src, 1,
                  ConstantInt::get(DL.getIntPtrType(Src->getContext()), Len + 1));
   return Dst;
 }
 
 Value *LibCallSimplifier::optimizeStrNCat(CallInst *CI, IRBuilder<> &B) {
   // Extract some information from the instruction.
   Value *Dst = CI->getArgOperand(0);
   Value *Src = CI->getArgOperand(1);
+  Value *Size = CI->getArgOperand(2);
   uint64_t Len;
+  annotateNonNullBasedOnAccess(CI, 0);
+  if (isKnownNonZero(Size, DL))
+    annotateNonNullBasedOnAccess(CI, 1);
 
   // We don't do anything if length is not constant.
-  if (ConstantInt *LengthArg = dyn_cast<ConstantInt>(CI->getArgOperand(2)))
+  ConstantInt *LengthArg = dyn_cast<ConstantInt>(Size);
+  if (LengthArg) {
     Len = LengthArg->getZExtValue();
-  else
+    // strncat(x, c, 0) -> x
+    if (!Len)
+      return Dst;
+  } else {
     return nullptr;
+  }
 
   // See if we can get the length of the input string.
   uint64_t SrcLen = GetStringLength(Src);
-  if (SrcLen == 0)
-    return nullptr;
+  if (SrcLen) {
+    annotateDereferenceableBytes(CI, 1, SrcLen);
     --SrcLen; // Unbias length.
+  } else {
+    return nullptr;
+  }
 
-  // Handle the simple, do-nothing cases:
   // strncat(x, "", c) -> x
-  // strncat(x,  c, 0) -> x
-  if (SrcLen == 0 || Len == 0)
+  if (SrcLen == 0)
     return Dst;
 
   // We don't optimize this case.
   if (Len < SrcLen)
     return nullptr;
 
   // strncat(x, s, c) -> strcat(x, s)
   // s is constant so the strcat can be optimized further.
   return emitStrLenMemCpy(Src, Dst, SrcLen, B);
 }
 
 Value *LibCallSimplifier::optimizeStrChr(CallInst *CI, IRBuilder<> &B) {
   Function *Callee = CI->getCalledFunction();
   FunctionType *FT = Callee->getFunctionType();
   Value *SrcStr = CI->getArgOperand(0);
+  annotateNonNullBasedOnAccess(CI, 0);
 
   // If the second operand is non-constant, see if we can compute the length
   // of the input string and turn this into memchr.
   ConstantInt *CharC = dyn_cast<ConstantInt>(CI->getArgOperand(1));
   if (!CharC) {
     uint64_t Len = GetStringLength(SrcStr);
-    if (Len == 0 || !FT->getParamType(1)->isIntegerTy(32)) // memchr needs i32.
+    if (Len)
+      annotateDereferenceableBytes(CI, 0, Len);
+    else
+      return nullptr;
+    if (!FT->getParamType(1)->isIntegerTy(32)) // memchr needs i32.
       return nullptr;
 
     return emitMemChr(SrcStr, CI->getArgOperand(1), // include nul.
                       ConstantInt::get(DL.getIntPtrType(CI->getContext()), Len),
                       B, DL, TLI);
   }
 
   // Otherwise, the character is a constant, see if the first argument is
Value *LibCallSimplifier::optimizeStrChr(CallInst *CI, IRBuilder<> &B) {
 
   // strchr(s+n,c)  -> gep(s+n+i,c)
   return B.CreateGEP(B.getInt8Ty(), SrcStr, B.getInt64(I), "strchr");
 }
 
 Value *LibCallSimplifier::optimizeStrRChr(CallInst *CI, IRBuilder<> &B) {
   Value *SrcStr = CI->getArgOperand(0);
   ConstantInt *CharC = dyn_cast<ConstantInt>(CI->getArgOperand(1));
+  annotateNonNullBasedOnAccess(CI, 0);
 
   // Cannot fold anything if we're not looking for a constant.
   if (!CharC)
     return nullptr;
 
   StringRef Str;
   if (!getConstantStringInfo(SrcStr, Str)) {
     // strrchr(s, 0) -> strchr(s, 0)
return B.CreateNeg(B.CreateZExt(
         B.CreateLoad(B.getInt8Ty(), Str2P, "strcmpload"), CI->getType()));
 
   if (HasStr2 && Str2.empty()) // strcmp(x,"") -> *x
     return B.CreateZExt(B.CreateLoad(B.getInt8Ty(), Str1P, "strcmpload"),
                         CI->getType());
 
   // strcmp(P, "x") -> memcmp(P, "x", 2)
   uint64_t Len1 = GetStringLength(Str1P);
+  if (Len1)
+    annotateDereferenceableBytes(CI, 0, Len1);
   uint64_t Len2 = GetStringLength(Str2P);
+  if (Len2)
+    annotateDereferenceableBytes(CI, 1, Len2);
+
   if (Len1 && Len2) {
     return emitMemCmp(Str1P, Str2P,
                       ConstantInt::get(DL.getIntPtrType(CI->getContext()),
                                        std::min(Len1, Len2)),
                       B, DL, TLI);
   }
 
   // strcmp to memcmp
   if (!HasStr1 && HasStr2) {
     if (canTransformToMemCmp(CI, Str1P, Len2, DL))
       return emitMemCmp(
           Str1P, Str2P,
           ConstantInt::get(DL.getIntPtrType(CI->getContext()), Len2), B, DL,
           TLI);
   } else if (HasStr1 && !HasStr2) {
     if (canTransformToMemCmp(CI, Str2P, Len1, DL))
       return emitMemCmp(
           Str1P, Str2P,
           ConstantInt::get(DL.getIntPtrType(CI->getContext()), Len1), B, DL,
           TLI);
   }
 
+  annotateNonNullBasedOnAccess(CI, {0, 1});
   return nullptr;
 }
 
 Value *LibCallSimplifier::optimizeStrNCmp(CallInst *CI, IRBuilder<> &B) {
-  Value *Str1P = CI->getArgOperand(0), *Str2P = CI->getArgOperand(1);
+  Value *Str1P = CI->getArgOperand(0);
+  Value *Str2P = CI->getArgOperand(1);
+  Value *Size = CI->getArgOperand(2);
   if (Str1P == Str2P) // strncmp(x,x,n)  -> 0
     return ConstantInt::get(CI->getType(), 0);
 
+  if (isKnownNonZero(Size, DL))
+    annotateNonNullBasedOnAccess(CI, {0, 1});
   // Get the length argument if it is constant.
   uint64_t Length;
-  if (ConstantInt *LengthArg = dyn_cast<ConstantInt>(CI->getArgOperand(2)))
+  if (ConstantInt *LengthArg = dyn_cast<ConstantInt>(Size))
     Length = LengthArg->getZExtValue();
   else
     return nullptr;
 
   if (Length == 0) // strncmp(x,y,0)   -> 0
     return ConstantInt::get(CI->getType(), 0);
 
   if (Length == 1) // strncmp(x,y,1) -> memcmp(x,y,1)
-    return emitMemCmp(Str1P, Str2P, CI->getArgOperand(2), B, DL, TLI);
+    return emitMemCmp(Str1P, Str2P, Size, B, DL, TLI);
 
   StringRef Str1, Str2;
   bool HasStr1 = getConstantStringInfo(Str1P, Str1);
   bool HasStr2 = getConstantStringInfo(Str2P, Str2);
 
   // strncmp(x, y)  -> cnst  (if both x and y are constant strings)
   if (HasStr1 && HasStr2) {
     StringRef SubStr1 = Str1.substr(0, Length);
     StringRef SubStr2 = Str2.substr(0, Length);
     return ConstantInt::get(CI->getType(), SubStr1.compare(SubStr2));
   }
 
   if (HasStr1 && Str1.empty()) // strncmp("", x, n) -> -*x
     return B.CreateNeg(B.CreateZExt(
         B.CreateLoad(B.getInt8Ty(), Str2P, "strcmpload"), CI->getType()));
 
   if (HasStr2 && Str2.empty()) // strncmp(x, "", n) -> *x
     return B.CreateZExt(B.CreateLoad(B.getInt8Ty(), Str1P, "strcmpload"),
                         CI->getType());
 
   uint64_t Len1 = GetStringLength(Str1P);
+  if (Len1)
+    annotateDereferenceableBytes(CI, 0, Len1);
   uint64_t Len2 = GetStringLength(Str2P);
+  if (Len2)
+    annotateDereferenceableBytes(CI, 1, Len2);
 
   // strncmp to memcmp
   if (!HasStr1 && HasStr2) {
     Len2 = std::min(Len2, Length);
     if (canTransformToMemCmp(CI, Str1P, Len2, DL))
       return emitMemCmp(
           Str1P, Str2P,
           ConstantInt::get(DL.getIntPtrType(CI->getContext()), Len2), B, DL,
Value *LibCallSimplifier::optimizeStrNCmp(CallInst *CI, IRBuilder<> &B) {
 
   return nullptr;
 }
 
 Value *LibCallSimplifier::optimizeStrCpy(CallInst *CI, IRBuilder<> &B) {
   Value *Dst = CI->getArgOperand(0), *Src = CI->getArgOperand(1);
   if (Dst == Src) // strcpy(x,x)  -> x
     return Src;
   
+  annotateNonNullBasedOnAccess(CI, {0, 1});
   // See if we can get the length of the input string.
   uint64_t Len = GetStringLength(Src);
-  if (Len == 0)
+  if (Len)
+    annotateDereferenceableBytes(CI, 1, Len);
+  else
     return nullptr;
 
   // We have enough information to now generate the memcpy call to do the
   // copy for us.  Make a memcpy to copy the nul byte with align = 1.
+  CallInst *NewCI =
       B.CreateMemCpy(Dst, 1, Src, 1,
                      ConstantInt::get(DL.getIntPtrType(CI->getContext()), Len));
+  NewCI->setAttributes(CI->getAttributes());
   return Dst;
 }
 
 Value *LibCallSimplifier::optimizeStpCpy(CallInst *CI, IRBuilder<> &B) {
   Function *Callee = CI->getCalledFunction();
   Value *Dst = CI->getArgOperand(0), *Src = CI->getArgOperand(1);
   if (Dst == Src) { // stpcpy(x,x)  -> x+strlen(x)
     Value *StrLen = emitStrLen(Src, B, DL, TLI);
     return StrLen ? B.CreateInBoundsGEP(B.getInt8Ty(), Dst, StrLen) : nullptr;
   }
 
   // See if we can get the length of the input string.
   uint64_t Len = GetStringLength(Src);
-  if (Len == 0)
+  if (Len)
+    annotateDereferenceableBytes(CI, 1, Len);
+  else
     return nullptr;
 
   Type *PT = Callee->getFunctionType()->getParamType(0);
   Value *LenV = ConstantInt::get(DL.getIntPtrType(PT), Len);
   Value *DstEnd = B.CreateGEP(B.getInt8Ty(), Dst,
                               ConstantInt::get(DL.getIntPtrType(PT), Len - 1));
 
   // We have enough information to now generate the memcpy call to do the
   // copy for us.  Make a memcpy to copy the nul byte with align = 1.
-  B.CreateMemCpy(Dst, 1, Src, 1, LenV);
+  CallInst *NewCI = B.CreateMemCpy(Dst, 1, Src, 1, LenV);
+  NewCI->setAttributes(CI->getAttributes());
   return DstEnd;
 }
 
 Value *LibCallSimplifier::optimizeStrNCpy(CallInst *CI, IRBuilder<> &B) {
   Function *Callee = CI->getCalledFunction();
   Value *Dst = CI->getArgOperand(0);
   Value *Src = CI->getArgOperand(1);
-  Value *LenOp = CI->getArgOperand(2);
+  Value *Size = CI->getArgOperand(2);
+  annotateNonNullBasedOnAccess(CI, 0);
+  if (isKnownNonZero(Size, DL))
+    annotateNonNullBasedOnAccess(CI, 1);
+
+  uint64_t Len;
+  if (ConstantInt *LengthArg = dyn_cast<ConstantInt>(Size))
+    Len = LengthArg->getZExtValue();
+  else
+    return nullptr;
+
+  // strncpy(x, y, 0) -> x
+  if (Len == 0)
+    return Dst;
 
   // See if we can get the length of the input string.
   uint64_t SrcLen = GetStringLength(Src);
-  if (SrcLen == 0)
+  if (SrcLen) {
+    annotateDereferenceableBytes(CI, 1, SrcLen);
+    --SrcLen; // Unbias length.
+  } else {
     return nullptr;
-  --SrcLen;
+  }
 
   if (SrcLen == 0) {
     // strncpy(x, "", y) -> memset(align 1 x, '\0', y)
-    B.CreateMemSet(Dst, B.getInt8('\0'), LenOp, 1);
+    CallInst *NewCI = B.CreateMemSet(Dst, B.getInt8('\0'), Size, 1);
+    AttrBuilder ArgAttrs(CI->getAttributes().getParamAttributes(0));
+    NewCI->getAttributes().addParamAttributes(CI->getContext(), 0, ArgAttrs);
     return Dst;
   }
 
-  uint64_t Len;
-  if (ConstantInt *LengthArg = dyn_cast<ConstantInt>(LenOp))
-    Len = LengthArg->getZExtValue();
-  else
-    return nullptr;
-
-  if (Len == 0)
-    return Dst; // strncpy(x, y, 0) -> x
-
   // Let strncpy handle the zero padding
   if (Len > SrcLen + 1)
     return nullptr;
 
   Type *PT = Callee->getFunctionType()->getParamType(0);
   // strncpy(x, s, c) -> memcpy(align 1 x, align 1 s, c) [s and c are constant]
-  B.CreateMemCpy(Dst, 1, Src, 1, ConstantInt::get(DL.getIntPtrType(PT), Len));
-
+  CallInst *NewCI = B.CreateMemCpy(Dst, 1, Src, 1, ConstantInt::get(DL.getIntPtrType(PT), Len));
+ // AttrBuilder ArgAttrs(CI->getAttributes().getParamAttributes(0));
+//  NewCI->getAttributes().addParamAttributes(CI->getContext(), 0, ArgAttrs);
+  NewCI->setAttributes(CI->getAttributes());
   return Dst;
 }
 
 Value *LibCallSimplifier::optimizeStringLength(CallInst *CI, IRBuilder<> &B,
                                                unsigned CharSize) {
   Value *Src = CI->getArgOperand(0);
 
   // Constant folding: strlen("xyz") -> 3
Value *LibCallSimplifier::optimizeStringLength(CallInst *CI, IRBuilder<> &B,
   if (isOnlyUsedInZeroEqualityComparison(CI))
     return B.CreateZExt(B.CreateLoad(B.getIntNTy(CharSize), Src, "strlenfirst"),
                         CI->getType());
 
   return nullptr;
 }
 
 Value *LibCallSimplifier::optimizeStrLen(CallInst *CI, IRBuilder<> &B) {
-  return optimizeStringLength(CI, B, 8);
+  if (Value *V = optimizeStringLength(CI, B, 8))
+    return V;
+  annotateNonNullBasedOnAccess(CI, 0);
+  return nullptr;
 }
 
 Value *LibCallSimplifier::optimizeWcslen(CallInst *CI, IRBuilder<> &B) {
   Module &M = *CI->getModule();
   unsigned WCharSize = TLI->getWCharSize(M) * 8;
   // We cannot perform this optimization without wchar_size metadata.
   if (WCharSize == 0)
     return nullptr;
if (HasStr1 && HasStr2) {
     return B.CreateBitCast(Result, CI->getType());
   }
 
   // fold strstr(x, "y") -> strchr(x, 'y').
   if (HasStr2 && ToFindStr.size() == 1) {
     Value *StrChr = emitStrChr(CI->getArgOperand(0), ToFindStr[0], B, TLI);
     return StrChr ? B.CreateBitCast(StrChr, CI->getType()) : nullptr;
   }
+
+  annotateNonNullBasedOnAccess(CI, {0, 1});
+  return nullptr;
+}
+
+Value *LibCallSimplifier::optimizeMemRChr(CallInst *CI, IRBuilder<> &B) {
+  if (isKnownNonZero(CI->getOperand(2), DL))
+    annotateNonNullBasedOnAccess(CI, 0);
   return nullptr;
 }
 
 Value *LibCallSimplifier::optimizeMemChr(CallInst *CI, IRBuilder<> &B) {
   Value *SrcStr = CI->getArgOperand(0);
+  Value *Size = CI->getArgOperand(2);
+  annotateNonNullAndDereferenceable(CI, 0, Size, DL);
   ConstantInt *CharC = dyn_cast<ConstantInt>(CI->getArgOperand(1));
-  ConstantInt *LenC = dyn_cast<ConstantInt>(CI->getArgOperand(2));
+  ConstantInt *LenC = dyn_cast<ConstantInt>(Size);
 
   // memchr(x, y, 0) -> null
   if (LenC) {
     if (LenC->isZero())
       return Constant::getNullValue(CI->getType());
-    annotateDereferenceableBytes(CI, {0}, LenC->getZExtValue());
-  }
+  } else {
     // From now on we need at least constant length and string.
+    return nullptr;
+  }
+
   StringRef Str;
-  if (!LenC || !getConstantStringInfo(SrcStr, Str, 0, /*TrimAtNul=*/false))
+  if (!getConstantStringInfo(SrcStr, Str, 0, /*TrimAtNul=*/false))
     return nullptr;
 
   // Truncate the string to LenC. If Str is smaller than LenC we will still only
   // scan the string, as reading past the end of it is undefined and we can just
   // return null if we don't find the char.
   Str = Str.substr(0, LenC->getZExtValue());
 
   // If the char is variable but the input str and length are not we can turn
if (getConstantStringInfo(LHS, LHSStr) &&
     uint64_t Ret = 0;
     int Cmp = memcmp(LHSStr.data(), RHSStr.data(), Len);
     if (Cmp < 0)
       Ret = -1;
     else if (Cmp > 0)
       Ret = 1;
     return ConstantInt::get(CI->getType(), Ret);
   }
+
   return nullptr;
 }
 
 // Most simplifications for memcmp also apply to bcmp.
 Value *LibCallSimplifier::optimizeMemCmpBCmpCommon(CallInst *CI,
                                                    IRBuilder<> &B) {
   Value *LHS = CI->getArgOperand(0), *RHS = CI->getArgOperand(1);
   Value *Size = CI->getArgOperand(2);
 
   if (LHS == RHS) // memcmp(s,s,x) -> 0
     return Constant::getNullValue(CI->getType());
 
+  annotateNonNullAndDereferenceable(CI, {0, 1}, Size, DL);
   // Handle constant lengths.
-  if (ConstantInt *LenC = dyn_cast<ConstantInt>(Size)) {
-    if (Value *Res = optimizeMemCmpConstantSize(CI, LHS, RHS,
-                                                LenC->getZExtValue(), B, DL))
-      return Res;
-    annotateDereferenceableBytes(CI, {0, 1}, LenC->getZExtValue());
-  }
+  ConstantInt *LenC = dyn_cast<ConstantInt>(Size);
+  if (!LenC)
+    return nullptr;
+
+  // memcmp(d,s,0) -> 0
+  if (LenC->getZExtValue() == 0)
+    return Constant::getNullValue(CI->getType());
 
+  if (Value *Res =
+          optimizeMemCmpConstantSize(CI, LHS, RHS, LenC->getZExtValue(), B, DL))
+    return Res;
   return nullptr;
 }
 
 Value *LibCallSimplifier::optimizeMemCmp(CallInst *CI, IRBuilder<> &B) {
   if (Value *V = optimizeMemCmpBCmpCommon(CI, B))
     return V;
 
   // memcmp(x, y, Len) == 0 -> bcmp(x, y, Len) == 0
   // bcmp can be more efficient than memcmp because it only has to know that
   // there is a difference, not how different one is to the other.
   if (TLI->has(LibFunc_bcmp) && isOnlyUsedInZeroEqualityComparison(CI)) {
     Value *LHS = CI->getArgOperand(0);
     Value *RHS = CI->getArgOperand(1);
     Value *Size = CI->getArgOperand(2);
     return emitBCmp(LHS, RHS, Size, B, DL, TLI);
   }
 
   return nullptr;
 }
 
 Value *LibCallSimplifier::optimizeBCmp(CallInst *CI, IRBuilder<> &B) {
   return optimizeMemCmpBCmpCommon(CI, B);
 }
 
-Value *LibCallSimplifier::optimizeMemCpy(CallInst *CI, IRBuilder<> &B,
-                                         bool isIntrinsic) {
+Value *LibCallSimplifier::optimizeMemCpy(CallInst *CI, IRBuilder<> &B) {
   Value *Size = CI->getArgOperand(2);
-  if (ConstantInt *LenC = dyn_cast<ConstantInt>(Size))
-    annotateDereferenceableBytes(CI, {0, 1}, LenC->getZExtValue());
-
-  if (isIntrinsic)
+  annotateNonNullAndDereferenceable(CI, {0, 1}, Size, DL);
+  if (isa<IntrinsicInst>(CI))
     return nullptr;
 
   // memcpy(x, y, n) -> llvm.memcpy(align 1 x, align 1 y, n)
+  CallInst *NewCI =
       B.CreateMemCpy(CI->getArgOperand(0), 1, CI->getArgOperand(1), 1, Size);
+  NewCI->setAttributes(CI->getAttributes());
   return CI->getArgOperand(0);
 }
 
 Value *LibCallSimplifier::optimizeMemPCpy(CallInst *CI, IRBuilder<> &B) {
   Value *Dst = CI->getArgOperand(0);
   Value *N = CI->getArgOperand(2);
   // mempcpy(x, y, n) -> llvm.memcpy(align 1 x, align 1 y, n), x + n
   CallInst *NewCI = B.CreateMemCpy(Dst, 1, CI->getArgOperand(1), 1, N);
   NewCI->setAttributes(CI->getAttributes());
   return B.CreateInBoundsGEP(B.getInt8Ty(), Dst, N);
 }
 
-Value *LibCallSimplifier::optimizeMemMove(CallInst *CI, IRBuilder<> &B, bool isIntrinsic) {
+Value *LibCallSimplifier::optimizeMemMove(CallInst *CI, IRBuilder<> &B) {
   Value *Size = CI->getArgOperand(2);
-  if (ConstantInt *LenC = dyn_cast<ConstantInt>(Size))
-    annotateDereferenceableBytes(CI, {0, 1}, LenC->getZExtValue());
-
-  if (isIntrinsic)
+  annotateNonNullAndDereferenceable(CI, {0, 1}, Size, DL);
+  if (isa<IntrinsicInst>(CI))
     return nullptr;
 
   // memmove(x, y, n) -> llvm.memmove(align 1 x, align 1 y, n)
-  B.CreateMemMove( CI->getArgOperand(0), 1, CI->getArgOperand(1), 1, Size);
-  return  CI->getArgOperand(0);
+  CallInst *NewCI =
+      B.CreateMemMove(CI->getArgOperand(0), 1, CI->getArgOperand(1), 1, Size);
+  NewCI->setAttributes(CI->getAttributes());
+  return CI->getArgOperand(0);
 }
 
 /// Fold memset[_chk](malloc(n), 0, n) --> calloc(1, n).
 Value *LibCallSimplifier::foldMallocMemset(CallInst *Memset, IRBuilder<> &B) {
   // This has to be a memset of zeros (bzero).
   auto *FillValue = dyn_cast<ConstantInt>(Memset->getArgOperand(1));
   if (!FillValue || FillValue->getZExtValue() != 0)
     return nullptr;
if (Value *Calloc = emitCalloc(ConstantInt::get(SizeType, 1),
                                  Malloc->getAttributes(), B, *TLI)) {
     substituteInParent(Malloc, Calloc);
     return Calloc;
   }
 
   return nullptr;
 }
 
-Value *LibCallSimplifier::optimizeMemSet(CallInst *CI, IRBuilder<> &B,
-                                         bool isIntrinsic) {
+Value *LibCallSimplifier::optimizeMemSet(CallInst *CI, IRBuilder<> &B) {
   Value *Size = CI->getArgOperand(2);
-  if (ConstantInt *LenC = dyn_cast<ConstantInt>(Size))
-    annotateDereferenceableBytes(CI, {0}, LenC->getZExtValue());
-
-  if (isIntrinsic)
+  annotateNonNullAndDereferenceable(CI, 0, Size, DL);
+  if (isa<IntrinsicInst>(CI))
     return nullptr;
 
   if (auto *Calloc = foldMallocMemset(CI, B))
     return Calloc;
 
   // memset(p, v, n) -> llvm.memset(align 1 p, v, n)
   Value *Val = B.CreateIntCast(CI->getArgOperand(1), B.getInt8Ty(), false);
-  B.CreateMemSet(CI->getArgOperand(0), Val, Size, 1);
+  CallInst *NewCI = B.CreateMemSet(CI->getArgOperand(0), Val, Size, 1);
+  NewCI->setAttributes(CI->getAttributes());
   return CI->getArgOperand(0);
 }
 
 Value *LibCallSimplifier::optimizeRealloc(CallInst *CI, IRBuilder<> &B) {
   if (isa<ConstantPointerNull>(CI->getArgOperand(0)))
     return emitMalloc(CI->getArgOperand(1), B, DL, TLI);
 
   return nullptr;
auto SmallPrintFFn =
         M->getOrInsertFunction(TLI->getName(LibFunc_small_printf),
                                FT, Callee->getAttributes());
     CallInst *New = cast<CallInst>(CI->clone());
     New->setCalledFunction(SmallPrintFFn);
     B.Insert(New);
     return New;
   }
 
+  annotateNonNullBasedOnAccess(CI, 0);
   return nullptr;
 }
 
 Value *LibCallSimplifier::optimizeSPrintFString(CallInst *CI, IRBuilder<> &B) {
   // Check for a fixed format string.
   StringRef FormatStr;
   if (!getConstantStringInfo(CI->getArgOperand(1), FormatStr))
     return nullptr;
auto SmallSPrintFFn =
         M->getOrInsertFunction(TLI->getName(LibFunc_small_sprintf),
                                FT, Callee->getAttributes());
     CallInst *New = cast<CallInst>(CI->clone());
     New->setCalledFunction(SmallSPrintFFn);
     B.Insert(New);
     return New;
   }
 
+  annotateNonNullBasedOnAccess(CI, {0, 1});
   return nullptr;
 }
 
 Value *LibCallSimplifier::optimizeSnPrintFString(CallInst *CI, IRBuilder<> &B) {
-  // Check for a fixed format string.
-  StringRef FormatStr;
-  if (!getConstantStringInfo(CI->getArgOperand(2), FormatStr))
-    return nullptr;
-
   // Check for size
   ConstantInt *Size = dyn_cast<ConstantInt>(CI->getArgOperand(1));
   if (!Size)
     return nullptr;
 
   uint64_t N = Size->getZExtValue();
+  // Check for a fixed format string.
+  StringRef FormatStr;
+  if (!getConstantStringInfo(CI->getArgOperand(2), FormatStr))
+    return nullptr;
 
   // If we just have a format string (nothing else crazy) transform it.
   if (CI->getNumArgOperands() == 3) {
     // Make sure there's no % in the constant array.  We could try to handle
     // %% -> % in the future if we cared.
     if (FormatStr.find('%') != StringRef::npos)
       return nullptr; // we found a format specifier, bail out.
 
Value *LibCallSimplifier::optimizeSnPrintFString(CallInst *CI, IRBuilder<> &B) {
   return nullptr;
 }
 
 Value *LibCallSimplifier::optimizeSnPrintF(CallInst *CI, IRBuilder<> &B) {
   if (Value *V = optimizeSnPrintFString(CI, B)) {
     return V;
   }
 
+  if (isKnownNonZero(CI->getOperand(1), DL))
+    annotateNonNullBasedOnAccess(CI, 0);
   return nullptr;
 }
 
 Value *LibCallSimplifier::optimizeFPrintFString(CallInst *CI, IRBuilder<> &B) {
   optimizeErrorReporting(CI, B, 0);
 
   // All the optimizations depend on the format string.
   StringRef FormatStr;
if (isLocallyOpenedFile(CI->getArgOperand(3), CI, B, TLI))
     return emitFReadUnlocked(CI->getArgOperand(0), CI->getArgOperand(1),
                              CI->getArgOperand(2), CI->getArgOperand(3), B, DL,
                              TLI);
 
   return nullptr;
 }
 
 Value *LibCallSimplifier::optimizePuts(CallInst *CI, IRBuilder<> &B) {
+  annotateNonNullBasedOnAccess(CI, 0);
   if (!CI->use_empty())
     return nullptr;
 
   // Check for a constant string.
   // puts("") -> putchar('\n')
   StringRef Str;
   if (getConstantStringInfo(CI->getArgOperand(0), Str) && Str.empty())
     return emitPutChar(B.getInt32('\n'), B, TLI);
if (TLI->getLibFunc(*Callee, Func) && TLI->has(Func)) {
     case LibFunc_strspn:
       return optimizeStrSpn(CI, Builder);
     case LibFunc_strcspn:
       return optimizeStrCSpn(CI, Builder);
     case LibFunc_strstr:
       return optimizeStrStr(CI, Builder);
     case LibFunc_memchr:
       return optimizeMemChr(CI, Builder);
+    case LibFunc_memrchr:
+      return optimizeMemRChr(CI, Builder);
     case LibFunc_bcmp:
       return optimizeBCmp(CI, Builder);
     case LibFunc_memcmp:
       return optimizeMemCmp(CI, Builder);
     case LibFunc_memcpy:
       return optimizeMemCpy(CI, Builder);
     case LibFunc_mempcpy:
       return optimizeMemPCpy(CI, Builder);
if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(CI)) {
     case Intrinsic::exp2:
       return optimizeExp2(CI, Builder);
     case Intrinsic::log:
       return optimizeLog(CI, Builder);
     case Intrinsic::sqrt:
       return optimizeSqrt(CI, Builder);
     // TODO: Use foldMallocMemset() with memset intrinsic.
     case Intrinsic::memset:
-      return optimizeMemSet(CI, Builder, true);
+      return optimizeMemSet(CI, Builder);
     case Intrinsic::memcpy:
-      return optimizeMemCpy(CI, Builder, true);
+      return optimizeMemCpy(CI, Builder);
     case Intrinsic::memmove:
-      return optimizeMemMove(CI, Builder, true);
+      return optimizeMemMove(CI, Builder);
     default:
       return nullptr;
     }
   }
 
   // Also try to simplify calls to fortified library functions.
   if (Value *SimplifiedFortifiedCI = FortifiedSimplifier.optimizeCall(CI)) {
     // Try to further simplify the result.
if (ObjSizeCI->isMinusOne())
       return true;
     // If the object size wasn't -1 (unknown), bail out if we were asked to.
     if (OnlyLowerUnknownSize)
       return false;
     if (StrOp) {
       uint64_t Len = GetStringLength(CI->getArgOperand(*StrOp));
       // If the length is 0 we don't know how long it is and so we can't
       // remove the check.
-      if (Len == 0)
+      if (Len)
+        annotateDereferenceableBytes(CI, *StrOp, Len);
+      else
         return false;
       return ObjSizeCI->getZExtValue() >= Len;
     }
 
     if (SizeOp) {
       if (ConstantInt *SizeCI =
               dyn_cast<ConstantInt>(CI->getArgOperand(*SizeOp)))
         return ObjSizeCI->getZExtValue() >= SizeCI->getZExtValue();
     }
   }
   return false;
 }
 
 Value *FortifiedLibCallSimplifier::optimizeMemCpyChk(CallInst *CI,
                                                      IRBuilder<> &B) {
   if (isFortifiedCallFoldable(CI, 3, 2)) {
-    B.CreateMemCpy(CI->getArgOperand(0), 1, CI->getArgOperand(1), 1,
-                   CI->getArgOperand(2));
+    CallInst *NewCI = B.CreateMemCpy(
+        CI->getArgOperand(0), 1, CI->getArgOperand(1), 1, CI->getArgOperand(2));
+    NewCI->setAttributes(CI->getAttributes());
     return CI->getArgOperand(0);
   }
   return nullptr;
 }
 
 Value *FortifiedLibCallSimplifier::optimizeMemMoveChk(CallInst *CI,
                                                       IRBuilder<> &B) {
   if (isFortifiedCallFoldable(CI, 3, 2)) {
-    B.CreateMemMove(CI->getArgOperand(0), 1, CI->getArgOperand(1), 1,
-                    CI->getArgOperand(2));
+    CallInst *NewCI = B.CreateMemMove(
+        CI->getArgOperand(0), 1, CI->getArgOperand(1), 1, CI->getArgOperand(2));
+    NewCI->setAttributes(CI->getAttributes());
     return CI->getArgOperand(0);
   }
   return nullptr;
 }
 
 Value *FortifiedLibCallSimplifier::optimizeMemSetChk(CallInst *CI,
                                                      IRBuilder<> &B) {
   // TODO: Try foldMallocMemset() here.
 
   if (isFortifiedCallFoldable(CI, 3, 2)) {
     Value *Val = B.CreateIntCast(CI->getArgOperand(1), B.getInt8Ty(), false);
+    CallInst *NewCI =
         B.CreateMemSet(CI->getArgOperand(0), Val, CI->getArgOperand(2), 1);
+    NewCI->setAttributes(CI->getAttributes());
     return CI->getArgOperand(0);
   }
   return nullptr;
 }
 
 Value *FortifiedLibCallSimplifier::optimizeStrpCpyChk(CallInst *CI,
                                                       IRBuilder<> &B,
                                                       LibFunc Func) {
else
       return emitStpCpy(Dst, Src, B, TLI);
   }
 
   if (OnlyLowerUnknownSize)
     return nullptr;
 
   // Maybe we can stil fold __st[rp]cpy_chk to __memcpy_chk.
   uint64_t Len = GetStringLength(Src);
-  if (Len == 0)
+  if (Len)
+    annotateDereferenceableBytes(CI, 1, Len);
+  else
     return nullptr;
 
   Type *SizeTTy = DL.getIntPtrType(CI->getContext());
   Value *LenV = ConstantInt::get(SizeTTy, Len);
   Value *Ret = emitMemCpyChk(Dst, Src, LenV, ObjSize, B, DL, TLI);
   // If the function was an __stpcpy_chk, and we were able to fold it into
   // a __memcpy_chk, we still need to return the correct end pointer.
   if (Ret && Func == LibFunc_stpcpy_chk)

llvm/trunk/test/Analysis/BasicAA/gep-alias.ll

 ; CHECK: store i8 99,
 ; Find the loads and make sure they are used in the arguments to the printf.
 ; CHECK: [[T0ADDR:%[a-zA-Z0-9_]+]] = getelementptr inbounds [3 x i8], [3 x i8]* %t, i32 0, i32 0
 ; CHECK: [[T0:%[a-zA-Z0-9_]+]] = load i8, i8* [[T0ADDR]], align 1
 ; CHECK: [[T0ARG:%[a-zA-Z0-9_]+]] = zext i8 [[T0]] to i32
 ; CHECK: [[U0ADDR:%[a-zA-Z0-9_]+]] = getelementptr inbounds [3 x i8], [3 x i8]* %u, i32 0, i32 0
 ; CHECK: [[U0:%[a-zA-Z0-9_]+]] = load i8, i8* [[U0ADDR]], align 1
 ; CHECK: [[U0ARG:%[a-zA-Z0-9_]+]] = zext i8 [[U0]] to i32
-; CHECK: call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([7 x i8], [7 x i8]* @.str, i32 0, i32 0), i32 [[T0ARG]], i32 [[U0ARG]])
+; CHECK: call i32 (i8*, ...) @printf(i8* nonnull dereferenceable(1) getelementptr inbounds ([7 x i8], [7 x i8]* @.str, i32 0, i32 0), i32 [[T0ARG]], i32 [[U0ARG]])
 ; CHECK: ret
 define void @test13() {
 entry:
   %t = alloca [3 x i8], align 1
   %u = alloca [3 x i8], align 1
   %tmp = load i32, i32* @P, align 4
   %tmp1 = mul i32 %tmp, 3
   %mul = add i32 %tmp1, -2147483646

llvm/trunk/test/Analysis/TypeBasedAliasAnalysis/memcpyopt.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; RUN: opt -S -tbaa -basicaa -memcpyopt -instcombine < %s | FileCheck %s
 
 target datalayout = "e-p:64:64:64"
 
 ; The second memcpy is redundant and can be deleted. There's an intervening store, but
 ; it has a TBAA tag which declares that it is unrelated.
 
 define void @foo(i8* nocapture %p, i8* nocapture %q, i8* nocapture %s) nounwind {
 ; CHECK: @foo
-; CHECK-NEXT: tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 1 dereferenceable(16) %p, i8* align 1 dereferenceable(16) %q, i64 16, i1 false), !tbaa !0
+; CHECK-NEXT: tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 1 dereferenceable(16) %p, i8* nonnull align 1 dereferenceable(16) %q, i64 16, i1 false), !tbaa !0
 ; CHECK-NEXT: store i8 2, i8* %s, align 1, !tbaa [[TAGA:!.*]]
 ; CHECK-NEXT: ret void
   tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* %p, i8* %q, i64 16, i1 false), !tbaa !2
   store i8 2, i8* %s, align 1, !tbaa !1
   tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* %q, i8* %p, i64 16, i1 false), !tbaa !2
   ret void
 }
 

llvm/trunk/test/CodeGen/X86/no-plt-libcalls.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Check if "RtLibUseGOT" works correctly when lib calls are simplified.
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 @percent_s = constant [4 x i8] c"%s\0A\00"
 @hello_world = constant [13 x i8] c"hello world\0A\00"
 declare i32 @printf(i8*, ...)
 define void @printf_call() {
+; CHECK-LABEL: @printf_call(
+; CHECK-NEXT:    [[PUTS:%.*]] = call i32 @puts(i8* nonnull dereferenceable(1) getelementptr inbounds ([13 x i8], [13 x i8]* @hello_world, i64 0, i64 0))
+; CHECK-NEXT:    ret void
+;
   %fmt = getelementptr [4 x i8], [4 x i8]* @percent_s, i32 0, i32 0
   %str = getelementptr [13 x i8], [13 x i8]* @hello_world, i32 0, i32 0
   call i32 (i8*, ...) @printf(i8* %fmt, i8* %str)
-; CHECK:  call i32 @puts(i8* getelementptr inbounds ([13 x i8], [13 x i8]* @hello_world, i64 0, i64 0))
   ret void
 }
 
 ; CHECK: Function Attrs: nofree nounwind nonlazybind
 ; CHECK-NEXT: declare i32 @puts(i8* nocapture readonly)
 
 !llvm.module.flags = !{!0}
 !0 = !{i32 7, !"RtLibUseGOT", i32 1}

llvm/trunk/test/Other/cgscc-libcall-update.ll

 ; Make sure that the CGSCC pass manager can handle when instcombine simplifies
 ; one libcall into an unrelated libcall and update the call graph accordingly.
 ;
 ; Also check that it can handle inlining *removing* a libcall entirely.
 ;
 ; Finally, we include some recursive patterns and forced analysis invaliadtion
 ; that can trigger infinite CGSCC refinement if not handled correctly.
 ;
 ; RUN: opt -passes='cgscc(inline,function(instcombine,invalidate<all>))' -S < %s | FileCheck %s
 
 define i8* @wibble(i8* %arg1, i8* %arg2) {
 ; CHECK-LABEL: define i8* @wibble(
 bb:
   %tmp = alloca [1024 x i8], align 16
   %tmp2 = getelementptr inbounds [1024 x i8], [1024 x i8]* %tmp, i64 0, i64 0
   call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp2, i8* %arg1, i64 1024, i1 false)
-; CHECK:         call void @llvm.memcpy
+; CHECK:         call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 16 dereferenceable(1024) 
   %tmp3 = call i64 @llvm.objectsize.i64.p0i8(i8* %tmp2, i1 false, i1 true, i1 false)
   %tmp4 = call i8* @__strncpy_chk(i8* %arg2, i8* %tmp2, i64 1023, i64 %tmp3)
 ; CHECK-NOT:     call
-; CHECK:         call i8* @strncpy(i8* %arg2, i8* nonnull %tmp2, i64 1023)
+; CHECK:         call i8* @strncpy(i8* nonnull dereferenceable(1) %arg2, i8* nonnull dereferenceable(1) %tmp2, i64 1023)
+
 ; CHECK-NOT:     call
 
   ret i8* %tmp4
 ; CHECK:         ret
 }
 
 define i8* @strncpy(i8* %arg1, i8* %arg2, i64 %size) noinline {
 bb:
+; CHECK:         call i8* @my_special_strncpy(i8* %arg1, i8* %arg2, i64 %size)
   %result = call i8* @my_special_strncpy(i8* %arg1, i8* %arg2, i64 %size)
   ret i8* %result
 }
 
 declare i8* @my_special_strncpy(i8* %arg1, i8* %arg2, i64 %size)
 
 declare i64 @llvm.objectsize.i64.p0i8(i8*, i1, i1, i1)
 

llvm/trunk/test/Transforms/InstCombine/ARM/strcmp.ll

;
   ret i32 %temp1
 }
 
 ; strcmp(x, y)   -> memcmp(x, y, <known length>)
 ; (This transform is rather difficult to trigger in a useful manner)
 define arm_aapcscc i32 @test5(i1 %b) {
 ; CHECK-LABEL: @test5(
 ; CHECK-NEXT:    [[STR2:%.*]] = select i1 [[B:%.*]], i8* getelementptr inbounds ([5 x i8], [5 x i8]* @hell, i32 0, i32 0), i8* getelementptr inbounds ([5 x i8], [5 x i8]* @bell, i32 0, i32 0)
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* dereferenceable(5) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i8* dereferenceable(5) [[STR2]], i32 5)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(5) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i8* nonnull dereferenceable(5) [[STR2]], i32 5)
 ; CHECK-NEXT:    ret i32 [[MEMCMP]]
 ;
 
   %str1 = getelementptr inbounds [6 x i8], [6 x i8]* @hello, i32 0, i32 0
   %temp1 = getelementptr inbounds [5 x i8], [5 x i8]* @hell, i32 0, i32 0
   %temp2 = getelementptr inbounds [5 x i8], [5 x i8]* @bell, i32 0, i32 0
   %str2 = select i1 %b, i8* %temp1, i8* %temp2
   %temp3 = call arm_aapcscc i32 @strcmp(i8* %str1, i8* %str2)
;
   ret i32 %temp1
 }
 
 ; strcmp(x, y)   -> memcmp(x, y, <known length>)
 ; (This transform is rather difficult to trigger in a useful manner)
 define arm_aapcs_vfpcc i32 @test5_vfp(i1 %b) {
 ; CHECK-LABEL: @test5_vfp(
 ; CHECK-NEXT:    [[STR2:%.*]] = select i1 [[B:%.*]], i8* getelementptr inbounds ([5 x i8], [5 x i8]* @hell, i32 0, i32 0), i8* getelementptr inbounds ([5 x i8], [5 x i8]* @bell, i32 0, i32 0)
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* dereferenceable(5) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i8* dereferenceable(5) [[STR2]], i32 5)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(5) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i8* nonnull dereferenceable(5) [[STR2]], i32 5)
 ; CHECK-NEXT:    ret i32 [[MEMCMP]]
 ;
 
   %str1 = getelementptr inbounds [6 x i8], [6 x i8]* @hello, i32 0, i32 0
   %temp1 = getelementptr inbounds [5 x i8], [5 x i8]* @hell, i32 0, i32 0
   %temp2 = getelementptr inbounds [5 x i8], [5 x i8]* @bell, i32 0, i32 0
   %str2 = select i1 %b, i8* %temp1, i8* %temp2
   %temp3 = call arm_aapcs_vfpcc i32 @strcmp(i8* %str1, i8* %str2)

llvm/trunk/test/Transforms/InstCombine/align-addr.ll

 declare void @use(i8*)
 
 %struct.s = type { i32, i32, i32, i32 }
 
 define void @test3(%struct.s* sret %a4) {
 ; Check that the alignment is bumped up the alignment of the sret type.
 ; CHECK-LABEL: @test3(
 ; CHECK-NEXT:    [[A4_CAST:%.*]] = bitcast %struct.s* [[A4:%.*]] to i8*
-; CHECK-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 4 dereferenceable(16) [[A4_CAST]], i8 0, i64 16, i1 false)
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i64(i8* nonnull align 4 dereferenceable(16) [[A4_CAST]], i8 0, i64 16, i1 false)
 ; CHECK-NEXT:    call void @use(i8* [[A4_CAST]])
 ; CHECK-NEXT:    ret void
 ;
   %a4.cast = bitcast %struct.s* %a4 to i8*
   call void @llvm.memset.p0i8.i64(i8* %a4.cast, i8 0, i64 16, i1 false)
   call void @use(i8* %a4.cast)
   ret void
 }

llvm/trunk/test/Transforms/InstCombine/fortify-folding.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; RUN: opt < %s -instcombine -S | FileCheck %s --dump-input-on-failure
 
 target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"
 
 @a = common global [60 x i8] zeroinitializer, align 1
 @b = common global [60 x i8] zeroinitializer, align 1
 @.str = private constant [12 x i8] c"abcdefghijk\00"
 
 %struct.__va_list_tag = type { i32, i32, i8*, i8* }
 
 define i8* @test_memccpy() {
-  ; CHECK-LABEL: define i8* @test_memccpy
-  ; CHECK-NEXT: call i8* @memccpy(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), i32 0, i64 60)
-  ; CHECK-NEXT: ret i8*
+; CHECK-LABEL: @test_memccpy(
+; CHECK-NEXT:    [[MEMCCPY:%.*]] = call i8* @memccpy(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), i32 0, i64 60)
+; CHECK-NEXT:    ret i8* [[MEMCCPY]]
+;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i8* @__memccpy_chk(i8* %dst, i8* %src, i32 0, i64 60, i64 -1)
   ret i8* %ret
 }
 
 define i8* @test_not_memccpy() {
-  ; CHECK-LABEL: define i8* @test_not_memccpy
-  ; CHECK-NEXT: call i8* @__memccpy_chk
-  ; CHECK-NEXT: ret i8*
+; CHECK-LABEL: @test_not_memccpy(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @__memccpy_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), i32 0, i64 60, i64 59)
+; CHECK-NEXT:    ret i8* [[RET]]
+;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i8* @__memccpy_chk(i8* %dst, i8* %src, i32 0, i64 60, i64 59)
   ret i8* %ret
 }
 
 define i32 @test_snprintf() {
-  ; CHECK-LABEL: define i32 @test_snprintf
-  ; CHECK-NEXT: call i32 (i8*, i64, i8*, ...) @snprintf(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i64 60, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0))
-  ; CHECK-NEXT: ret i32
+; CHECK-LABEL: @test_snprintf(
+; CHECK-NEXT:    [[SNPRINTF:%.*]] = call i32 (i8*, i64, i8*, ...) @snprintf(i8* nonnull dereferenceable(1) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i64 60, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0))
+; CHECK-NEXT:    ret i32 [[SNPRINTF]]
+;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %fmt = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i32 (i8*, i64, i32, i64, i8*, ...) @__snprintf_chk(i8* %dst, i64 60, i32 0, i64 -1, i8* %fmt)
   ret i32 %ret
 }
 
 define i32 @test_not_snprintf() {
-  ; CHECK-LABEL: define i32 @test_not_snprintf
-  ; CHECK-NEXT: call i32 (i8*, i64, i32, i64, i8*, ...) @__snprintf_chk
-  ; CHECK-NEXT: call i32 (i8*, i64, i32, i64, i8*, ...) @__snprintf_chk
-  ; CHECK-NEXT: ret i32
+; CHECK-LABEL: @test_not_snprintf(
+; CHECK-NEXT:    [[RET:%.*]] = call i32 (i8*, i64, i32, i64, i8*, ...) @__snprintf_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i64 60, i32 0, i64 59, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0))
+; CHECK-NEXT:    [[IGN:%.*]] = call i32 (i8*, i64, i32, i64, i8*, ...) @__snprintf_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i64 60, i32 1, i64 -1, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0))
+; CHECK-NEXT:    ret i32 [[RET]]
+;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %fmt = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i32 (i8*, i64, i32, i64, i8*, ...) @__snprintf_chk(i8* %dst, i64 60, i32 0, i64 59, i8* %fmt)
   %ign = call i32 (i8*, i64, i32, i64, i8*, ...) @__snprintf_chk(i8* %dst, i64 60, i32 1, i64 -1, i8* %fmt)
   ret i32 %ret
 }
 
 define i32 @test_sprintf() {
-  ; CHECK-LABEL: define i32 @test_sprintf
-  ; CHECK-NEXT: call i32 (i8*, i8*, ...) @sprintf(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0))
-  ; CHECK-NEXT: ret i32
+; CHECK-LABEL: @test_sprintf(
+; CHECK-NEXT:    [[SPRINTF:%.*]] = call i32 (i8*, i8*, ...) @sprintf(i8* nonnull dereferenceable(1) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* nonnull dereferenceable(1) getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0))
+; CHECK-NEXT:    ret i32 [[SPRINTF]]
+;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %fmt = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i32 (i8*, i32, i64, i8*, ...) @__sprintf_chk(i8* %dst, i32 0, i64 -1, i8* %fmt)
   ret i32 %ret
 }
 
 define i32 @test_not_sprintf() {
-  ; CHECK-LABEL: define i32 @test_not_sprintf
-  ; CHECK-NEXT: call i32 (i8*, i32, i64, i8*, ...) @__sprintf_chk
-  ; CHECK-NEXT: call i32 (i8*, i32, i64, i8*, ...) @__sprintf_chk
-  ; CHECK-NEXT: ret i32
+; CHECK-LABEL: @test_not_sprintf(
+; CHECK-NEXT:    [[RET:%.*]] = call i32 (i8*, i32, i64, i8*, ...) @__sprintf_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i32 0, i64 59, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0))
+; CHECK-NEXT:    [[IGNORED:%.*]] = call i32 (i8*, i32, i64, i8*, ...) @__sprintf_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i32 1, i64 -1, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0))
+; CHECK-NEXT:    ret i32 [[RET]]
+;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %fmt = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i32 (i8*, i32, i64, i8*, ...) @__sprintf_chk(i8* %dst, i32 0, i64 59, i8* %fmt)
   %ignored = call i32 (i8*, i32, i64, i8*, ...) @__sprintf_chk(i8* %dst, i32 1, i64 -1, i8* %fmt)
   ret i32 %ret
 }
 
 define i8* @test_strcat() {
-  ; CHECK-LABEL: define i8* @test_strcat
-  ; CHECK-NEXT: call i8* @strcat(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0))
-  ; CHECK-NEXT: ret i8*
+; CHECK-LABEL: @test_strcat(
+; CHECK-NEXT:    [[STRCAT:%.*]] = call i8* @strcat(i8* nonnull dereferenceable(1) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* nonnull dereferenceable(1) getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0))
+; CHECK-NEXT:    ret i8* [[STRCAT]]
+;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i8* @__strcat_chk(i8* %dst, i8* %src, i64 -1)
   ret i8* %ret
 }
 
 define i8* @test_not_strcat() {
-  ; CHECK-LABEL: define i8* @test_not_strcat
-  ; CHECK-NEXT: call i8* @__strcat_chk
-  ; CHECK-NEXT: ret i8*
+; CHECK-LABEL: @test_not_strcat(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @__strcat_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), i64 0)
+; CHECK-NEXT:    ret i8* [[RET]]
+;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i8* @__strcat_chk(i8* %dst, i8* %src, i64 0)
   ret i8* %ret
 }
 
 define i64 @test_strlcat() {
-  ; CHECK-LABEL: define i64 @test_strlcat
-  ; CHECK-NEXT: call i64 @strlcat(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), i64 22)
-  ; CHECK-NEXT: ret i64
+; CHECK-LABEL: @test_strlcat(
+; CHECK-NEXT:    [[STRLCAT:%.*]] = call i64 @strlcat(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), i64 22)
+; CHECK-NEXT:    ret i64 [[STRLCAT]]
+;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i64 @__strlcat_chk(i8* %dst, i8* %src, i64 22, i64 -1)
   ret i64 %ret
 }
 
 define i64 @test_not_strlcat() {
-  ; CHECK-LABEL: define i64 @test_not_strlcat
-  ; CHECK-NEXT: call i64 @__strlcat_chk
-  ; CHECK-NEXT: ret i64
+; CHECK-LABEL: @test_not_strlcat(
+; CHECK-NEXT:    [[RET:%.*]] = call i64 @__strlcat_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), i64 22, i64 0)
+; CHECK-NEXT:    ret i64 [[RET]]
+;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i64 @__strlcat_chk(i8* %dst, i8* %src, i64 22, i64 0)
   ret i64 %ret
 }
 
 define i8* @test_strncat() {
-  ; CHECK-LABEL: define i8* @test_strncat
-  ; CHECK-NEXT: call i8* @strncat(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), i64 22)
-  ; CHECK-NEXT: ret i8*
+; CHECK-LABEL: @test_strncat(
+; CHECK-NEXT:    [[STRNCAT:%.*]] = call i8* @strncat(i8* nonnull dereferenceable(1) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* nonnull dereferenceable(1) getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), i64 22)
+; CHECK-NEXT:    ret i8* [[STRNCAT]]
+;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i8* @__strncat_chk(i8* %dst, i8* %src, i64 22, i64 -1)
   ret i8* %ret
 }
 
 define i8* @test_not_strncat() {
-  ; CHECK-LABEL: define i8* @test_not_strncat
-  ; CHECK-NEXT: call i8* @__strncat_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), i64 22, i64 3)
-  ; CHECK-NEXT: ret i8*
+; CHECK-LABEL: @test_not_strncat(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @__strncat_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), i64 22, i64 3)
+; CHECK-NEXT:    ret i8* [[RET]]
+;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i8* @__strncat_chk(i8* %dst, i8* %src, i64 22, i64 3)
   ret i8* %ret
 }
 
 define i64 @test_strlcpy() {
-  ; CHECK-LABEL: define i64 @test_strlcpy
-  ; CHECK-NEXT: call i64 @strlcpy(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), i64 22)
-  ; CHECK-NEXT: ret i64
+; CHECK-LABEL: @test_strlcpy(
+; CHECK-NEXT:    [[STRLCPY:%.*]] = call i64 @strlcpy(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), i64 22)
+; CHECK-NEXT:    ret i64 [[STRLCPY]]
+;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i64 @__strlcpy_chk(i8* %dst, i8* %src, i64 22, i64 -1)
   ret i64 %ret
 }
 
 define i64 @test_not_strlcpy() {
-  ; CHECK-LABEL: define i64 @test_not_strlcpy
-  ; CHECK-NEXT: call i64 @__strlcpy_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), i64 22, i64 2)
-  ; CHECK-NEXT: ret i64
+; CHECK-LABEL: @test_not_strlcpy(
+; CHECK-NEXT:    [[RET:%.*]] = call i64 @__strlcpy_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), i64 22, i64 2)
+; CHECK-NEXT:    ret i64 [[RET]]
+;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i64 @__strlcpy_chk(i8* %dst, i8* %src, i64 22, i64 2)
   ret i64 %ret
 }
 
 define i32 @test_vsnprintf() {
-  ; CHECK-LABEL: define i32 @test_vsnprintf
-  ; CHECK-NEXT: call i32 @vsnprintf(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i64 4, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), %struct.__va_list_tag* null)
+; CHECK-LABEL: @test_vsnprintf(
+; CHECK-NEXT:    [[VSNPRINTF:%.*]] = call i32 @vsnprintf(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i64 4, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), %struct.__va_list_tag* null)
+; CHECK-NEXT:    ret i32 [[VSNPRINTF]]
+;
   ; ret i32
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i32 @__vsnprintf_chk(i8* %dst, i64 4, i32 0, i64 -1, i8* %src, %struct.__va_list_tag* null)
   ret i32 %ret
 }
 
 define i32 @test_not_vsnprintf() {
-  ; CHECK-LABEL: define i32 @test_not_vsnprintf
-  ; CHECK-NEXT: call i32 @__vsnprintf_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i64 4, i32 0, i64 3, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), %struct.__va_list_tag* null)
-  ; CHECK-NEXT: call i32 @__vsnprintf_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i64 4, i32 1, i64 -1, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), %struct.__va_list_tag* null)
+; CHECK-LABEL: @test_not_vsnprintf(
+; CHECK-NEXT:    [[RET:%.*]] = call i32 @__vsnprintf_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i64 4, i32 0, i64 3, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), %struct.__va_list_tag* null)
+; CHECK-NEXT:    [[IGN:%.*]] = call i32 @__vsnprintf_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i64 4, i32 1, i64 -1, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), %struct.__va_list_tag* null)
+; CHECK-NEXT:    ret i32 [[RET]]
+;
   ; ret i32
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i32 @__vsnprintf_chk(i8* %dst, i64 4, i32 0, i64 3, i8* %src, %struct.__va_list_tag* null)
   %ign = call i32 @__vsnprintf_chk(i8* %dst, i64 4, i32 1, i64 -1, i8* %src, %struct.__va_list_tag* null)
   ret i32 %ret
 }
 
 define i32 @test_vsprintf() {
-  ; CHECK-LABEL: define i32 @test_vsprintf
-  ; CHECK-NEXT: call i32 @vsprintf(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), %struct.__va_list_tag* null)
+; CHECK-LABEL: @test_vsprintf(
+; CHECK-NEXT:    [[VSPRINTF:%.*]] = call i32 @vsprintf(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), %struct.__va_list_tag* null)
+; CHECK-NEXT:    ret i32 [[VSPRINTF]]
+;
   ; ret i32
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i32 @__vsprintf_chk(i8* %dst, i32 0, i64 -1, i8* %src, %struct.__va_list_tag* null)
   ret i32 %ret
 }
 
 define i32 @test_not_vsprintf() {
-  ; CHECK-LABEL: define i32 @test_not_vsprintf
-  ; CHECK-NEXT: call i32 @__vsprintf_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i32 0, i64 3, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), %struct.__va_list_tag* null)
-  ; CHECK-NEXT: call i32 @__vsprintf_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i32 1, i64 -1, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), %struct.__va_list_tag* null)
+; CHECK-LABEL: @test_not_vsprintf(
+; CHECK-NEXT:    [[RET:%.*]] = call i32 @__vsprintf_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i32 0, i64 3, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), %struct.__va_list_tag* null)
+; CHECK-NEXT:    [[IGN:%.*]] = call i32 @__vsprintf_chk(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i64 0, i64 0), i32 1, i64 -1, i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i64 0, i64 0), %struct.__va_list_tag* null)
+; CHECK-NEXT:    ret i32 [[RET]]
+;
   ; ret i32
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
   %ret = call i32 @__vsprintf_chk(i8* %dst, i32 0, i64 3, i8* %src, %struct.__va_list_tag* null)
   %ign = call i32 @__vsprintf_chk(i8* %dst, i32 1, i64 -1, i8* %src, %struct.__va_list_tag* null)
   ret i32 %ret
 }
 

llvm/trunk/test/Transforms/InstCombine/getelementptr.ll

entry:
   %orientations62 = getelementptr [1 x [1 x %struct.x]], [1 x [1 x %struct.x]]* %orientations, i32 0, i32 0, i32 0
   br label %bb10
 
 bb10:
   %indvar = phi i32 [ 0, %entry ], [ %indvar.next, %bb10 ]
   %tmp.0.reg2mem.0.rec = mul i32 %indvar, -1
   %tmp12.rec = add i32 %tmp.0.reg2mem.0.rec, -1
   %tmp12 = getelementptr inbounds %struct.x, %struct.x* %tmp45, i32 %tmp12.rec
-  %tmp16 = call i32 (i8*, ...) @printf( i8* getelementptr ([12 x i8], [12 x i8]* @.str1, i32 0, i32 0), %struct.x* %tmp12 ) nounwind
+  %tmp16 = call i32 (i8*, ...) @printf( i8* nonnull dereferenceable(1) getelementptr ([12 x i8], [12 x i8]* @.str1, i32 0, i32 0), %struct.x* %tmp12 ) nounwind
   %tmp84 = icmp eq %struct.x* %tmp12, %orientations62
   %indvar.next = add i32 %indvar, 1
   br i1 %tmp84, label %bb17, label %bb10
 
 bb17:
   ret i32 0
 ; CHECK-LABEL: @test28(
 ; CHECK: icmp eq i32 %indvar, 0

 
 ; Instcombine should be able to fold this getelementptr.
 
 define i32 @test35() nounwind {
   call i32 (i8*, ...) @printf(i8* getelementptr ([17 x i8], [17 x i8]* @"\01LC8", i32 0, i32 0),
              i8* getelementptr (%t1, %t1* bitcast (%t0* @s to %t1*), i32 0, i32 1, i32 0)) nounwind
   ret i32 0
 ; CHECK-LABEL: @test35(
-; CHECK: call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([17 x i8], [17 x i8]* @"\01LC8", i64 0, i64 0), i8* getelementptr inbounds (%t0, %t0* @s, i64 0, i32 1, i64 0)) [[$NUW:#[0-9]+]]
+; CHECK: call i32 (i8*, ...) @printf(i8*  nonnull dereferenceable(1) getelementptr inbounds ([17 x i8], [17 x i8]* @"\01LC8", i64 0, i64 0), i8* getelementptr inbounds (%t0, %t0* @s, i64 0, i32 1, i64 0)) [[$NUW:#[0-9]+]]
 }
 
 ; Don't treat signed offsets as unsigned.
 define i8* @test36() nounwind {
   ret i8* getelementptr ([11 x i8], [11 x i8]* @array, i32 0, i64 -1)
 ; CHECK-LABEL: @test36(
 ; CHECK: ret i8* getelementptr ([11 x i8], [11 x i8]* @array, i64 0, i64 -1)
 }

llvm/trunk/test/Transforms/InstCombine/malloc-free-delete.ll

 ; CHECK-LABEL: @test5(
 ; CHECK-NEXT:    [[A:%.*]] = call dereferenceable_or_null(700) i8* @malloc(i32 700)
 ; CHECK-NEXT:    [[B:%.*]] = call dereferenceable_or_null(700) i8* @malloc(i32 700)
 ; CHECK-NEXT:    [[C:%.*]] = call dereferenceable_or_null(700) i8* @malloc(i32 700)
 ; CHECK-NEXT:    [[D:%.*]] = call dereferenceable_or_null(700) i8* @malloc(i32 700)
 ; CHECK-NEXT:    [[E:%.*]] = call dereferenceable_or_null(700) i8* @malloc(i32 700)
 ; CHECK-NEXT:    [[F:%.*]] = call dereferenceable_or_null(700) i8* @malloc(i32 700)
 ; CHECK-NEXT:    [[G:%.*]] = call dereferenceable_or_null(700) i8* @malloc(i32 700)
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(32) [[PTR:%.*]], i8* align 1 dereferenceable(32) [[A]], i32 32, i1 false)
-; CHECK-NEXT:    call void @llvm.memmove.p0i8.p0i8.i32(i8* align 1 dereferenceable(32) [[PTR]], i8* align 1 dereferenceable(32) [[B]], i32 32, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(32) [[PTR:%.*]], i8* nonnull align 1 dereferenceable(32) [[A]], i32 32, i1 false)
+; CHECK-NEXT:    call void @llvm.memmove.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(32) [[PTR]], i8* nonnull align 1 dereferenceable(32) [[B]], i32 32, i1 false)
 ; CHECK-NEXT:    store i8* [[C]], i8** [[ESC:%.*]], align 8
 ; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* [[D]], i8* [[PTR]], i32 32, i1 true)
 ; CHECK-NEXT:    call void @llvm.memmove.p0i8.p0i8.i32(i8* [[E]], i8* [[PTR]], i32 32, i1 true)
 ; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* [[F]], i8 5, i32 32, i1 true)
 ; CHECK-NEXT:    store volatile i8 4, i8* [[G]], align 1
 ; CHECK-NEXT:    ret void
 ;
   %a = call i8* @malloc(i32 700)

llvm/trunk/test/Transforms/InstCombine/mem-deref-bytes-addrspaces.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; RUN: opt -instcombine  -S < %s | FileCheck %s
 
 declare i32 @memcmp(i8 addrspace(1)* nocapture, i8* nocapture, i64)
 
-define i32 @memcmp_const_size_update_deref8(i8 addrspace(1)* nocapture readonly %d, i8* nocapture readonly %s) {
-; CHECK-LABEL: @memcmp_const_size_update_deref8(
-; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8 addrspace(1)* dereferenceable(16) dereferenceable_or_null(40) [[D:%.*]], i8* dereferenceable(16) [[S:%.*]], i64 16)
+define i32 @memcmp_const_size_update_deref(i8 addrspace(1)* nocapture readonly %d, i8* nocapture readonly %s) {
+; CHECK-LABEL: @memcmp_const_size_update_deref(
+; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8 addrspace(1)* dereferenceable(16) dereferenceable_or_null(40) [[D:%.*]], i8* nonnull dereferenceable(16) [[S:%.*]], i64 16)
 ; CHECK-NEXT:    ret i32 [[CALL]]
 ;
   %call = tail call i32 @memcmp(i8 addrspace(1)* dereferenceable_or_null(40) %d, i8* %s, i64 16)
   ret i32 %call
 }
+
+define i32 @memcmp_nonconst_size_nonnnull(i8 addrspace(1)* nocapture readonly %d, i8* nocapture readonly %s, i64 %n) {
+; CHECK-LABEL: @memcmp_nonconst_size_nonnnull(
+; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8 addrspace(1)* nonnull dereferenceable_or_null(40) [[D:%.*]], i8* nonnull [[S:%.*]], i64 [[N:%.*]])
+; CHECK-NEXT:    ret i32 [[CALL]]
+;
+  %call = tail call i32 @memcmp(i8 addrspace(1)* nonnull dereferenceable_or_null(40) %d, i8* nonnull %s, i64 %n)
+  ret i32 %call
+}

llvm/trunk/test/Transforms/InstCombine/mem-deref-bytes.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; RUN: opt -instcombine  -S < %s | FileCheck %s
 
 declare i32 @memcmp(i8* nocapture, i8* nocapture, i64)
 declare i8* @memcpy(i8* nocapture, i8* nocapture, i64)
 declare i8* @memmove(i8* nocapture, i8* nocapture, i64)
 declare i8* @memset(i8* nocapture, i8, i64)
 declare i8* @memchr(i8* nocapture, i32, i64)
 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i1)
 declare void @llvm.memmove.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i1)
 declare void @llvm.memset.p0i8.i64(i8* nocapture, i8, i64, i1)
 
 define i32 @memcmp_const_size_set_deref(i8* nocapture readonly %d, i8* nocapture readonly %s) {
 ; CHECK-LABEL: @memcmp_const_size_set_deref(
-; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* dereferenceable(16) [[D:%.*]], i8* dereferenceable(16) [[S:%.*]], i64 16)
+; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* nonnull dereferenceable(16) [[D:%.*]], i8* nonnull dereferenceable(16) [[S:%.*]], i64 16)
 ; CHECK-NEXT:    ret i32 [[CALL]]
 ;
   %call = tail call i32 @memcmp(i8* %d, i8* %s, i64 16)
   ret i32 %call
 }
 
 define i32 @memcmp_const_size_update_deref(i8* nocapture readonly %d, i8* nocapture readonly %s) {
 ; CHECK-LABEL: @memcmp_const_size_update_deref(
-; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* dereferenceable(16) [[D:%.*]], i8* dereferenceable(16) [[S:%.*]], i64 16)
+; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* nonnull dereferenceable(16) [[D:%.*]], i8* nonnull dereferenceable(16) [[S:%.*]], i64 16)
 ; CHECK-NEXT:    ret i32 [[CALL]]
 ;
   %call = tail call i32 @memcmp(i8* dereferenceable(4) %d, i8* dereferenceable(8) %s, i64 16)
   ret i32 %call
 }
 
 define i32 @memcmp_const_size_update_deref2(i8* nocapture readonly %d, i8* nocapture readonly %s) {
 ; CHECK-LABEL: @memcmp_const_size_update_deref2(
-; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* dereferenceable(16) [[D:%.*]], i8* dereferenceable(16) [[S:%.*]], i64 16)
+; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* nonnull dereferenceable(16) [[D:%.*]], i8* nonnull dereferenceable(16) [[S:%.*]], i64 16)
 ; CHECK-NEXT:    ret i32 [[CALL]]
 ;
   %call = tail call i32 @memcmp(i8* %d, i8* dereferenceable_or_null(8) %s, i64 16)
   ret i32 %call
 }
 
 define i32 @memcmp_const_size_update_deref3(i8* nocapture readonly %d, i8* nocapture readonly %s) {
 ; CHECK-LABEL: @memcmp_const_size_update_deref3(
-; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* dereferenceable(40) [[D:%.*]], i8* dereferenceable(16) [[S:%.*]], i64 16)
+; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* nonnull dereferenceable(40) [[D:%.*]], i8* nonnull dereferenceable(16) [[S:%.*]], i64 16)
 ; CHECK-NEXT:    ret i32 [[CALL]]
 ;
   %call = tail call i32 @memcmp(i8* dereferenceable(40) %d, i8* %s, i64 16)
   ret i32 %call
 }
 
 define i32 @memcmp_const_size_update_deref4(i8* nocapture readonly %d, i8* nocapture readonly %s) {
 ; CHECK-LABEL: @memcmp_const_size_update_deref4(
-; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* dereferenceable(16) [[D:%.*]], i8* dereferenceable(16) [[S:%.*]], i64 16)
+; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* nonnull dereferenceable(16) [[D:%.*]], i8* nonnull dereferenceable(16) [[S:%.*]], i64 16)
 ; CHECK-NEXT:    ret i32 [[CALL]]
 ;
   %call = tail call i32 @memcmp(i8* dereferenceable_or_null(16) %d, i8* %s, i64 16)
   ret i32 %call
 }
 
 define i32 @memcmp_const_size_update_deref5(i8* nocapture readonly %d, i8* nocapture readonly %s) "null-pointer-is-valid"="false" {
 ; CHECK-LABEL: @memcmp_const_size_update_deref5(
-; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* dereferenceable(40) [[D:%.*]], i8* dereferenceable(16) [[S:%.*]], i64 16)
+; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* nonnull dereferenceable(40) [[D:%.*]], i8* nonnull dereferenceable(16) [[S:%.*]], i64 16)
 ; CHECK-NEXT:    ret i32 [[CALL]]
 ;
   %call = tail call i32 @memcmp(i8* dereferenceable_or_null(40) %d, i8* %s, i64 16)
   ret i32 %call
 }
 
 define i32 @memcmp_const_size_update_deref6(i8* nocapture readonly %d, i8* nocapture readonly %s) "null-pointer-is-valid"="true" {
 ; CHECK-LABEL: @memcmp_const_size_update_deref6(
 ; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* dereferenceable(16) dereferenceable_or_null(40) [[D:%.*]], i8* dereferenceable(16) [[S:%.*]], i64 16)
 ; CHECK-NEXT:    ret i32 [[CALL]]
 ;
   %call = tail call i32 @memcmp(i8* dereferenceable_or_null(40) %d, i8* %s, i64 16)
   ret i32 %call
 }
 
+define i32 @memcmp_const_size_update_deref7(i8* nocapture readonly %d, i8* nocapture readonly %s) "null-pointer-is-valid"="true" {
+; CHECK-LABEL: @memcmp_const_size_update_deref7(
+; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* nonnull dereferenceable(40) [[D:%.*]], i8* dereferenceable(16) [[S:%.*]], i64 16)
+; CHECK-NEXT:    ret i32 [[CALL]]
+;
+  %call = tail call i32 @memcmp(i8* nonnull dereferenceable_or_null(40) %d, i8* %s, i64 16)
+  ret i32 %call
+}
+
 define i32 @memcmp_const_size_no_update_deref(i8* nocapture readonly %d, i8* nocapture readonly %s) {
 ; CHECK-LABEL: @memcmp_const_size_no_update_deref(
-; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* dereferenceable(40) [[D:%.*]], i8* dereferenceable(20) [[S:%.*]], i64 16)
+; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* nonnull dereferenceable(40) [[D:%.*]], i8* nonnull dereferenceable(20) [[S:%.*]], i64 16)
 ; CHECK-NEXT:    ret i32 [[CALL]]
 ;
   %call = tail call i32 @memcmp(i8* dereferenceable(40) %d, i8* dereferenceable(20) %s, i64 16)
   ret i32 %call
 }
 
 define i32 @memcmp_nonconst_size(i8* nocapture readonly %d, i8* nocapture readonly %s, i64 %n) {
 ; CHECK-LABEL: @memcmp_nonconst_size(
 ; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* [[D:%.*]], i8* [[S:%.*]], i64 [[N:%.*]])
 ; CHECK-NEXT:    ret i32 [[CALL]]
 ;
   %call = tail call i32 @memcmp(i8* %d, i8* %s, i64 %n)
   ret i32 %call
 }
 
 define i8* @memcpy_const_size_set_deref(i8* nocapture readonly %d, i8* nocapture readonly %s) {
 ; CHECK-LABEL: @memcpy_const_size_set_deref(
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 1 dereferenceable(64) [[D:%.*]], i8* align 1 dereferenceable(64) [[S:%.*]], i64 64, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 1 dereferenceable(64) [[D:%.*]], i8* nonnull align 1 dereferenceable(64) [[S:%.*]], i64 64, i1 false)
 ; CHECK-NEXT:    ret i8* [[D]]
 ;
   %call = tail call i8* @memcpy(i8* %d, i8* %s, i64 64)
   ret i8* %call
 }
 
 define i8* @memmove_const_size_set_deref(i8* nocapture readonly %d, i8* nocapture readonly %s) {
 ; CHECK-LABEL: @memmove_const_size_set_deref(
-; CHECK-NEXT:    call void @llvm.memmove.p0i8.p0i8.i64(i8* align 1 dereferenceable(64) [[D:%.*]], i8* align 1 dereferenceable(64) [[S:%.*]], i64 64, i1 false)
+; CHECK-NEXT:    call void @llvm.memmove.p0i8.p0i8.i64(i8* nonnull align 1 dereferenceable(64) [[D:%.*]], i8* nonnull align 1 dereferenceable(64) [[S:%.*]], i64 64, i1 false)
 ; CHECK-NEXT:    ret i8* [[D]]
 ;
   %call = tail call i8* @memmove(i8* %d, i8* %s, i64 64)
   ret i8* %call
 }
 
 define i8* @memset_const_size_set_deref(i8* nocapture readonly %s, i8 %c) {
 ; CHECK-LABEL: @memset_const_size_set_deref(
-; CHECK-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 1 dereferenceable(64) [[S:%.*]], i8 [[C:%.*]], i64 64, i1 false)
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i64(i8* nonnull align 1 dereferenceable(64) [[S:%.*]], i8 [[C:%.*]], i64 64, i1 false)
 ; CHECK-NEXT:    ret i8* [[S]]
 ;
   %call = tail call i8* @memset(i8* %s, i8 %c, i64 64)
   ret i8* %call
 }
 
 define i8* @memchr_const_size_set_deref(i8* nocapture readonly %s, i32 %c) {
 ; CHECK-LABEL: @memchr_const_size_set_deref(
-; CHECK-NEXT:    [[CALL:%.*]] = tail call i8* @memchr(i8* dereferenceable(64) [[S:%.*]], i32 [[C:%.*]], i64 64)
+; CHECK-NEXT:    [[CALL:%.*]] = tail call i8* @memchr(i8* nonnull dereferenceable(64) [[S:%.*]], i32 [[C:%.*]], i64 64)
 ; CHECK-NEXT:    ret i8* [[CALL]]
 ;
   %call = tail call i8* @memchr(i8* %s, i32 %c, i64 64)
   ret i8* %call
 }
 
 define i8* @llvm_memcpy_const_size_set_deref(i8* nocapture readonly %d, i8* nocapture readonly %s) {
 ; CHECK-LABEL: @llvm_memcpy_const_size_set_deref(
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 1 dereferenceable(16) [[D:%.*]], i8* align 1 dereferenceable(16) [[S:%.*]], i64 16, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 1 dereferenceable(16) [[D:%.*]], i8* nonnull align 1 dereferenceable(16) [[S:%.*]], i64 16, i1 false)
 ; CHECK-NEXT:    ret i8* [[D]]
 ;
   call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 1 %d, i8* align 1 %s, i64 16, i1 false)
   ret i8* %d
 }
 
 define i8* @llvm_memmove_const_size_set_deref(i8* nocapture readonly %d, i8* nocapture readonly %s) {
 ; CHECK-LABEL: @llvm_memmove_const_size_set_deref(
-; CHECK-NEXT:    call void @llvm.memmove.p0i8.p0i8.i64(i8* align 1 dereferenceable(16) [[D:%.*]], i8* align 1 dereferenceable(16) [[S:%.*]], i64 16, i1 false)
+; CHECK-NEXT:    call void @llvm.memmove.p0i8.p0i8.i64(i8* nonnull align 1 dereferenceable(16) [[D:%.*]], i8* nonnull align 1 dereferenceable(16) [[S:%.*]], i64 16, i1 false)
 ; CHECK-NEXT:    ret i8* [[D]]
 ;
   call void @llvm.memmove.p0i8.p0i8.i64(i8* align 1 %d, i8* align 1 %s, i64 16, i1 false)
   ret i8* %d
 }
 define i8* @llvm_memset_const_size_set_deref(i8* nocapture readonly %s, i8 %c) {
 ; CHECK-LABEL: @llvm_memset_const_size_set_deref(
-; CHECK-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 1 dereferenceable(16) [[S:%.*]], i8 [[C:%.*]], i64 16, i1 false)
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i64(i8* nonnull align 1 dereferenceable(16) [[S:%.*]], i8 [[C:%.*]], i64 16, i1 false)
 ; CHECK-NEXT:    ret i8* [[S]]
 ;
   call void @llvm.memset.p0i8.i64(i8* align 1 %s, i8 %c, i64 16, i1 false)
   ret i8* %s
 }

llvm/trunk/test/Transforms/InstCombine/memchr.ll

;
   %src = getelementptr [14 x i8], [14 x i8]* @hello, i32 0, i32 0
   %dst = call i8* @memchr(i8* %src, i32 0, i32 14)
   store i8* %dst, i8** @chp
   ret void
 }
 
 define void @test4(i32 %chr) {
 ; CHECK-LABEL: @test4(
-; CHECK-NEXT:    [[DST:%.*]] = call i8* @memchr(i8* dereferenceable(14) getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 0), i32 [[CHR:%.*]], i32 14)
+; CHECK-NEXT:    [[DST:%.*]] = call i8* @memchr(i8* nonnull dereferenceable(14) getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 0), i32 [[CHR:%.*]], i32 14)
 ; CHECK-NEXT:    store i8* [[DST]], i8** @chp, align 4
 ; CHECK-NEXT:    ret void
 ;
   %src = getelementptr [14 x i8], [14 x i8]* @hello, i32 0, i32 0
   %dst = call i8* @memchr(i8* %src, i32 %chr, i32 14)
   store i8* %dst, i8** @chp
   ret void
 }
;
   %dst = call i8* @memchr(i8* getelementptr inbounds ([3 x i8], [3 x i8]* @newlines, i64 0, i64 0), i32 %C, i32 2)
   %cmp = icmp ne i8* %dst, null
   ret i1 %cmp
 }
 
 ; No 64 bits here
 define i1 @test12(i32 %C) {
 ; CHECK-LABEL: @test12(
-; CHECK-NEXT:    [[DST:%.*]] = call i8* @memchr(i8* dereferenceable(3) getelementptr inbounds ([4 x i8], [4 x i8]* @spaces, i32 0, i32 0), i32 [[C:%.*]], i32 3)
+; CHECK-NEXT:    [[DST:%.*]] = call i8* @memchr(i8* nonnull dereferenceable(3) getelementptr inbounds ([4 x i8], [4 x i8]* @spaces, i32 0, i32 0), i32 [[C:%.*]], i32 3)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp ne i8* [[DST]], null
 ; CHECK-NEXT:    ret i1 [[CMP]]
 ;
   %dst = call i8* @memchr(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @spaces, i64 0, i64 0), i32 %C, i32 3)
   %cmp = icmp ne i8* %dst, null
   ret i1 %cmp
 }
 

 ;
   %dst = call i8* @memchr(i8* getelementptr inbounds ([2 x i8], [2 x i8]* @single, i64 0, i64 0), i32 %C, i32 1)
   %cmp = icmp ne i8* %dst, null
   ret i1 %cmp
 }
 
 define i1 @test15(i32 %C) {
 ; CHECK-LABEL: @test15(
-; CHECK-NEXT:    [[DST:%.*]] = call i8* @memchr(i8* dereferenceable(3) getelementptr inbounds ([3 x i8], [3 x i8]* @negative, i32 0, i32 0), i32 [[C:%.*]], i32 3)
+; CHECK-NEXT:    [[DST:%.*]] = call i8* @memchr(i8* nonnull dereferenceable(3) getelementptr inbounds ([3 x i8], [3 x i8]* @negative, i32 0, i32 0), i32 [[C:%.*]], i32 3)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp ne i8* [[DST]], null
 ; CHECK-NEXT:    ret i1 [[CMP]]
 ;
   %dst = call i8* @memchr(i8* getelementptr inbounds ([3 x i8], [3 x i8]* @negative, i64 0, i64 0), i32 %C, i32 3)
   %cmp = icmp ne i8* %dst, null
   ret i1 %cmp
 }
 
 @s = internal constant [1 x i8] [i8 0], align 1
 define i8* @pr32124() {
 ; CHECK-LABEL: @pr32124(
 ; CHECK-NEXT:    ret i8* getelementptr inbounds ([1 x i8], [1 x i8]* @s, i32 0, i32 0)
 ;
   %res = tail call i8* @memchr(i8* getelementptr ([1 x i8], [1 x i8]* @s, i64 0, i64 0), i32 0, i32 1)
   ret i8* %res
 }
+
+define i8* @test16(i8* %str, i32 %c, i32 %n) {
+; CHECK-LABEL: @test16(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @memchr(i8* [[STR:%.*]], i32 [[C:%.*]], i32 [[N:%.*]])
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+
+  %ret = call i8* @memchr(i8* %str, i32 %c, i32 %n)
+  ret i8* %ret
+}
+
+define i8* @test17(i8* %str, i32 %c, i32 %n) {
+; CHECK-LABEL: @test17(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @memchr(i8* nonnull [[STR:%.*]], i32 [[C:%.*]], i32 [[N:%.*]])
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+
+  %ret = call i8* @memchr(i8* nonnull %str, i32 %c, i32 %n)
+  ret i8* %ret
+}
+
+define i8* @test18(i8* %str, i32 %c) {
+; CHECK-LABEL: @test18(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @memchr(i8* nonnull dereferenceable(5) [[STR:%.*]], i32 [[C:%.*]], i32 5)
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+
+  %ret = call i8* @memchr(i8* %str, i32 %c, i32 5)
+  ret i8* %ret
+}
+
+define i8* @test19(i8* %str, i32 %c) "null-pointer-is-valid"="true" {
+; CHECK-LABEL: @test19(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @memchr(i8* dereferenceable(5) [[STR:%.*]], i32 [[C:%.*]], i32 5)
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+
+  %ret = call i8* @memchr(i8* %str, i32 %c, i32 5)
+  ret i8* %ret
+}

llvm/trunk/test/Transforms/InstCombine/memcmp-constant-fold.ll

 }
 
 ; TODO: Any memcmp where all arguments are constants should be constant folded. Currently, we only handle i8 array constants.
 
 @intbuf = private unnamed_addr constant [2 x i32] [i32 0, i32 1], align 4
 
 define i1 @memcmp_3bytes_aligned_constant_i32(i8* align 4 %x) {
 ; ALL-LABEL: @memcmp_3bytes_aligned_constant_i32(
-; ALL-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* dereferenceable(3) bitcast (i32* getelementptr inbounds ([2 x i32], [2 x i32]* @intbuf, i64 0, i64 1) to i8*), i8* dereferenceable(3) bitcast ([2 x i32]* @intbuf to i8*), i64 3)
+; ALL-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* nonnull dereferenceable(3) bitcast (i32* getelementptr inbounds ([2 x i32], [2 x i32]* @intbuf, i64 0, i64 1) to i8*), i8* nonnull dereferenceable(3) bitcast ([2 x i32]* @intbuf to i8*), i64 3)
 ; ALL-NEXT:    [[CMPEQ0:%.*]] = icmp eq i32 [[CALL]], 0
 ; ALL-NEXT:    ret i1 [[CMPEQ0]]
 ;
   %call = tail call i32 @memcmp(i8* bitcast (i32* getelementptr inbounds ([2 x i32], [2 x i32]* @intbuf, i64 0, i64 1) to i8*), i8* bitcast (i32* getelementptr inbounds ([2 x i32], [2 x i32]* @intbuf, i64 0, i64 0) to i8*), i64 3)
   %cmpeq0 = icmp eq i32 %call, 0
   ret i1 %cmpeq0
 }
 
 ; A sloppy implementation would infinite loop by recreating the unused instructions.
 
 define i1 @memcmp_4bytes_one_unaligned_i8(i8* align 4 %x, i8* align 1 %y) {
 ; ALL-LABEL: @memcmp_4bytes_one_unaligned_i8(
-; ALL-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* dereferenceable(4) [[X:%.*]], i8* dereferenceable(4) [[Y:%.*]], i64 4)
+; ALL-NEXT:    [[CALL:%.*]] = tail call i32 @memcmp(i8* nonnull dereferenceable(4) [[X:%.*]], i8* nonnull dereferenceable(4) [[Y:%.*]], i64 4)
 ; ALL-NEXT:    [[CMPEQ0:%.*]] = icmp eq i32 [[CALL]], 0
 ; ALL-NEXT:    ret i1 [[CMPEQ0]]
 ;
   %bc = bitcast i8* %x to i32*
   %lhsv = load i32, i32* %bc
   %call = tail call i32 @memcmp(i8* %x, i8* %y, i64 4)
   %cmpeq0 = icmp eq i32 %call, 0
   ret i1 %cmpeq0
 }
 

llvm/trunk/test/Transforms/InstCombine/memcpy-from-global.ll

 
 @G = constant %T {i8 1, [123 x i8] zeroinitializer }
 @H = constant [2 x %U] zeroinitializer, align 16
 
 define void @test2() {
 ; CHECK-LABEL: @test2(
 ; CHECK-NEXT:    [[B1:%.*]] = alloca [124 x i8], align 8
 ; CHECK-NEXT:    [[B1_SUB:%.*]] = getelementptr inbounds [124 x i8], [124 x i8]* [[B1]], i64 0, i64 0
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 8 dereferenceable(124) [[B1_SUB]], i8* align 16 dereferenceable(124) getelementptr inbounds (%T, %T* @G, i64 0, i32 0), i64 124, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 8 dereferenceable(124) [[B1_SUB]], i8* nonnull align 16 dereferenceable(124) getelementptr inbounds (%T, %T* @G, i64 0, i32 0), i64 124, i1 false)
 ; CHECK-NEXT:    call void @bar(i8* nonnull [[B1_SUB]])
 ; CHECK-NEXT:    ret void
 ;
   %A = alloca %T
   %B = alloca %T
   %a = bitcast %T* %A to i8*
   %b = bitcast %T* %B to i8*
 
;
   call void @bar(i8* %a) readonly
   ret void
 }
 
 define void @test8() {
 ; CHECK-LABEL: @test8(
 ; CHECK-NEXT:    [[AL:%.*]] = alloca [[U:%.*]], align 16
 ; CHECK-NEXT:    [[A:%.*]] = bitcast %U* [[AL]] to i8*
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 16 dereferenceable(20) [[A]], i8* align 4 dereferenceable(20) bitcast (%U* getelementptr inbounds ([2 x %U], [2 x %U]* @H, i64 0, i64 1) to i8*), i64 20, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 16 dereferenceable(20) [[A]], i8* nonnull align 4 dereferenceable(20) bitcast (%U* getelementptr inbounds ([2 x %U], [2 x %U]* @H, i64 0, i64 1) to i8*), i64 20, i1 false)
 ; CHECK-NEXT:    call void @bar(i8* nonnull [[A]]) #2
 ; CHECK-NEXT:    ret void
 ;
   %al = alloca %U, align 16
   %a = bitcast %U* %al to i8*
   call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%U* getelementptr ([2 x %U], [2 x %U]* @H, i64 0, i32 1) to i8*), i64 20, i1 false)
   call void @bar(i8* %a) readonly
   ret void

 @_ZL3KKK = internal unnamed_addr constant [3 x i8] c"\01\01\02", align 1
 
 ; Should not replace alloca with global because of size mismatch.
 define void @test9_small_global() {
 ; CHECK-LABEL: @test9_small_global(
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[CC:%.*]] = alloca [1000000 x i8], align 16
 ; CHECK-NEXT:    [[ARRAYDECAY:%.*]] = getelementptr inbounds [1000000 x i8], [1000000 x i8]* [[CC]], i64 0, i64 0
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 16 dereferenceable(3) [[ARRAYDECAY]], i8* align 16 dereferenceable(3) getelementptr inbounds ([3 x i8], [3 x i8]* @_ZL3KKK, i64 0, i64 0), i64 3, i1 false)
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 dereferenceable(1000000) getelementptr inbounds ([1000000 x i8], [1000000 x i8]* @bbb, i64 0, i64 0), i8* nonnull align 16 dereferenceable(1000000) [[ARRAYDECAY]], i64 1000000, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 16 dereferenceable(3) [[ARRAYDECAY]], i8* nonnull align 16 dereferenceable(3) getelementptr inbounds ([3 x i8], [3 x i8]* @_ZL3KKK, i64 0, i64 0), i64 3, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 16 dereferenceable(1000000) getelementptr inbounds ([1000000 x i8], [1000000 x i8]* @bbb, i64 0, i64 0), i8* nonnull align 16 dereferenceable(1000000) [[ARRAYDECAY]], i64 1000000, i1 false)
 ; CHECK-NEXT:    ret void
 ;
 entry:
   %cc = alloca [1000000 x i8], align 16
   %cc.0..sroa_idx = getelementptr inbounds [1000000 x i8], [1000000 x i8]* %cc, i64 0, i64 0
   %arraydecay = getelementptr inbounds [1000000 x i8], [1000000 x i8]* %cc, i32 0, i32 0
   call void @llvm.memcpy.p0i8.p0i8.i64(i8* %arraydecay, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @_ZL3KKK, i32 0, i32 0), i64 3, i1 false)
   call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 getelementptr inbounds ([1000000 x i8], [1000000 x i8]* @bbb, i32 0, i32 0), i8* align 16 %arraydecay, i64 1000000, i1 false)
   ret void
 }
 
 ; Should replace alloca with global as they have exactly the same size.
 define void @test10_same_global() {
 ; CHECK-LABEL: @test10_same_global(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 dereferenceable(3) getelementptr inbounds ([1000000 x i8], [1000000 x i8]* @bbb, i64 0, i64 0), i8* align 16 dereferenceable(3) getelementptr inbounds ([3 x i8], [3 x i8]* @_ZL3KKK, i64 0, i64 0), i64 3, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 16 dereferenceable(3) getelementptr inbounds ([1000000 x i8], [1000000 x i8]* @bbb, i64 0, i64 0), i8* nonnull align 16 dereferenceable(3) getelementptr inbounds ([3 x i8], [3 x i8]* @_ZL3KKK, i64 0, i64 0), i64 3, i1 false)
 ; CHECK-NEXT:    ret void
 ;
 entry:
   %cc = alloca [3 x i8], align 1
   %cc.0..sroa_idx = getelementptr inbounds [3 x i8], [3 x i8]* %cc, i64 0, i64 0
   %arraydecay = getelementptr inbounds [3 x i8], [3 x i8]* %cc, i32 0, i32 0
   call void @llvm.memcpy.p0i8.p0i8.i64(i8* %arraydecay, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @_ZL3KKK, i32 0, i32 0), i64 3, i1 false)
   call void @llvm.memcpy.p0i8.p0i8.i64(i8* getelementptr inbounds ([1000000 x i8], [1000000 x i8]* @bbb, i32 0, i32 0), i8* %arraydecay, i64 3, i1 false)
   ret void
 }
 
 attributes #0 = { "null-pointer-is-valid"="true" }

llvm/trunk/test/Transforms/InstCombine/memcpy-to-load.ll

;
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %d, i8* %s, i32 2, i1 false)
   ret void
 }
 
 ; We don't expand small non-power-of-2. Should we? Might be a target-dependent choice.
 
 define void @copy_3_bytes(i8* %d, i8* %s) {
 ; ALL-LABEL: @copy_3_bytes(
-; ALL-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(3) [[D:%.*]], i8* align 1 dereferenceable(3) [[S:%.*]], i32 3, i1 false)
+; ALL-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(3) [[D:%.*]], i8* nonnull align 1 dereferenceable(3) [[S:%.*]], i32 3, i1 false)
 ; ALL-NEXT:    ret void
 ;
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %d, i8* %s, i32 3, i1 false)
   ret void
 }
 
 define void @copy_4_bytes(i8* %d, i8* %s) {
 ; ALL-LABEL: @copy_4_bytes(
 ; ALL-NEXT:    [[TMP1:%.*]] = bitcast i8* [[S:%.*]] to i32*
 ; ALL-NEXT:    [[TMP2:%.*]] = bitcast i8* [[D:%.*]] to i32*
 ; ALL-NEXT:    [[TMP3:%.*]] = load i32, i32* [[TMP1]], align 1
 ; ALL-NEXT:    store i32 [[TMP3]], i32* [[TMP2]], align 1
 ; ALL-NEXT:    ret void
 ;
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %d, i8* %s, i32 4, i1 false)
   ret void
 }
 
 ; We don't expand small non-power-of-2. Should we? Might be a target-dependent choice.
 
 define void @copy_5_bytes(i8* %d, i8* %s) {
 ; ALL-LABEL: @copy_5_bytes(
-; ALL-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(5) [[D:%.*]], i8* align 1 dereferenceable(5) [[S:%.*]], i32 5, i1 false)
+; ALL-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(5) [[D:%.*]], i8* nonnull align 1 dereferenceable(5) [[S:%.*]], i32 5, i1 false)
 ; ALL-NEXT:    ret void
 ;
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %d, i8* %s, i32 5, i1 false)
   ret void
 }
 
 define void @copy_8_bytes(i8* %d, i8* %s) {
 ; ALL-LABEL: @copy_8_bytes(
 ; ALL-NEXT:    [[TMP1:%.*]] = bitcast i8* [[S:%.*]] to i64*
 ; ALL-NEXT:    [[TMP2:%.*]] = bitcast i8* [[D:%.*]] to i64*
 ; ALL-NEXT:    [[TMP3:%.*]] = load i64, i64* [[TMP1]], align 1
 ; ALL-NEXT:    store i64 [[TMP3]], i64* [[TMP2]], align 1
 ; ALL-NEXT:    ret void
 ;
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %d, i8* %s, i32 8, i1 false)
   ret void
 }
 
 define void @copy_16_bytes(i8* %d, i8* %s) {
 ; ALL-LABEL: @copy_16_bytes(
-; ALL-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(16) [[D:%.*]], i8* align 1 dereferenceable(16) [[S:%.*]], i32 16, i1 false)
+; ALL-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(16) [[D:%.*]], i8* nonnull align 1 dereferenceable(16) [[S:%.*]], i32 16, i1 false)
 ; ALL-NEXT:    ret void
 ;
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %d, i8* %s, i32 16, i1 false)
   ret void
 }
 

llvm/trunk/test/Transforms/InstCombine/memcpy.ll

;
   tail call void @llvm.memcpy.p0i8.p0i8.i32(i8* %a, i8* %a, i32 100, i1 true)
   ret void
 }
 
 ; 17179869184 == 0x400000000 - make sure that doesn't get truncated to 32-bit.
 
 define void @test3(i8* %d, i8* %s) {
 ; CHECK-LABEL: @test3(
-; CHECK-NEXT:    tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 dereferenceable(17179869184) [[D:%.*]], i8* align 4 dereferenceable(17179869184) [[S:%.*]], i64 17179869184, i1 false)
+; CHECK-NEXT:    tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 4 dereferenceable(17179869184) [[D:%.*]], i8* nonnull align 4 dereferenceable(17179869184) [[S:%.*]], i64 17179869184, i1 false)
 ; CHECK-NEXT:    ret void
 ;
   tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %d, i8* align 4 %s, i64 17179869184, i1 false)
   ret void
 }
 
 @UnknownConstant = external constant i128
 
 define void @memcpy_to_constant(i8* %src) {
 ; CHECK-LABEL: @memcpy_to_constant(
 ; CHECK-NEXT:    ret void
 ;
   %dest = bitcast i128* @UnknownConstant to i8*
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 16, i1 false)
   ret void
 }

llvm/trunk/test/Transforms/InstCombine/memcpy_chk-1.ll

 @t1 = common global %struct.T1 zeroinitializer
 @t2 = common global %struct.T2 zeroinitializer
 @t3 = common global %struct.T3 zeroinitializer
 
 ; Check cases where dstlen >= len.
 
 define i8* @test_simplify1() {
 ; CHECK-LABEL: @test_simplify1(
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 dereferenceable(1824) bitcast (%struct.T1* @t1 to i8*), i8* align 4 dereferenceable(1824) bitcast (%struct.T2* @t2 to i8*), i64 1824, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 4 dereferenceable(1824) bitcast (%struct.T1* @t1 to i8*), i8* nonnull align 4 dereferenceable(1824) bitcast (%struct.T2* @t2 to i8*), i64 1824, i1 false)
 ; CHECK-NEXT:    ret i8* bitcast (%struct.T1* @t1 to i8*)
 ;
   %dst = bitcast %struct.T1* @t1 to i8*
   %src = bitcast %struct.T2* @t2 to i8*
   %ret = call i8* @__memcpy_chk(i8* %dst, i8* %src, i64 1824, i64 1824)
   ret i8* %ret
 }
 
 define i8* @test_simplify2() {
 ; CHECK-LABEL: @test_simplify2(
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 dereferenceable(1824) bitcast (%struct.T1* @t1 to i8*), i8* align 4 dereferenceable(1824) bitcast (%struct.T3* @t3 to i8*), i64 1824, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 4 dereferenceable(1824) bitcast (%struct.T1* @t1 to i8*), i8* nonnull align 4 dereferenceable(1824) bitcast (%struct.T3* @t3 to i8*), i64 1824, i1 false)
 ; CHECK-NEXT:    ret i8* bitcast (%struct.T1* @t1 to i8*)
 ;
   %dst = bitcast %struct.T1* @t1 to i8*
   %src = bitcast %struct.T3* @t3 to i8*
   %ret = call i8* @__memcpy_chk(i8* %dst, i8* %src, i64 1824, i64 2848)
   ret i8* %ret
 }
 
;
   %src = bitcast %struct.T2* @t2 to i8*
   %ret = call i8* @__memcpy_chk(i8* %dst, i8* %src, i64 1024, i64 0)
   ret i8* %ret
 }
 
 define i8* @test_simplify_return_indcall(i8* ()* %alloc) {
 ; CHECK-LABEL: @test_simplify_return_indcall(
 ; CHECK-NEXT:    [[DST:%.*]] = call i8* [[ALLOC:%.*]]()
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 1 dereferenceable(1824) [[DST]], i8* align 4 dereferenceable(1824) bitcast (%struct.T2* @t2 to i8*), i64 1824, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 1 dereferenceable(1824) [[DST]], i8* nonnull align 4 dereferenceable(1824) bitcast (%struct.T2* @t2 to i8*), i64 1824, i1 false)
 ; CHECK-NEXT:    ret i8* [[DST]]
 ;
   %src = bitcast %struct.T2* @t2 to i8*
   %dst = call i8* %alloc()
   %ret = call i8* @__memcpy_chk(i8* %dst, i8* %src, i64 1824, i64 1824)
   ret i8* %ret
 }
 
 declare i8* @__memcpy_chk(i8*, i8*, i64, i64)

llvm/trunk/test/Transforms/InstCombine/memmove_chk-1.ll

 @t1 = common global %struct.T1 zeroinitializer
 @t2 = common global %struct.T2 zeroinitializer
 @t3 = common global %struct.T3 zeroinitializer
 
 ; Check cases where dstlen >= len.
 
 define i8* @test_simplify1() {
 ; CHECK-LABEL: @test_simplify1(
-; CHECK-NEXT:    call void @llvm.memmove.p0i8.p0i8.i64(i8* align 4 dereferenceable(1824) bitcast (%struct.T1* @t1 to i8*), i8* align 4 dereferenceable(1824) bitcast (%struct.T2* @t2 to i8*), i64 1824, i1 false)
+; CHECK-NEXT:    call void @llvm.memmove.p0i8.p0i8.i64(i8* nonnull align 4 dereferenceable(1824) bitcast (%struct.T1* @t1 to i8*), i8* nonnull align 4 dereferenceable(1824) bitcast (%struct.T2* @t2 to i8*), i64 1824, i1 false)
 ; CHECK-NEXT:    ret i8* bitcast (%struct.T1* @t1 to i8*)
 ;
   %dst = bitcast %struct.T1* @t1 to i8*
   %src = bitcast %struct.T2* @t2 to i8*
 
   %ret = call i8* @__memmove_chk(i8* %dst, i8* %src, i64 1824, i64 1824)
   ret i8* %ret
 }
 
 define i8* @test_simplify2() {
 ; CHECK-LABEL: @test_simplify2(
-; CHECK-NEXT:    call void @llvm.memmove.p0i8.p0i8.i64(i8* align 4 dereferenceable(1824) bitcast (%struct.T1* @t1 to i8*), i8* align 4 dereferenceable(1824) bitcast (%struct.T3* @t3 to i8*), i64 1824, i1 false)
+; CHECK-NEXT:    call void @llvm.memmove.p0i8.p0i8.i64(i8* nonnull align 4 dereferenceable(1824) bitcast (%struct.T1* @t1 to i8*), i8* nonnull align 4 dereferenceable(1824) bitcast (%struct.T3* @t3 to i8*), i64 1824, i1 false)
 ; CHECK-NEXT:    ret i8* bitcast (%struct.T1* @t1 to i8*)
 ;
   %dst = bitcast %struct.T1* @t1 to i8*
   %src = bitcast %struct.T3* @t3 to i8*
 
   %ret = call i8* @__memmove_chk(i8* %dst, i8* %src, i64 1824, i64 2848)
   ret i8* %ret
 }

llvm/trunk/test/Transforms/InstCombine/mempcpy.ll

 ; CHECK-NEXT:    ret i8* [[TMP4]]
 ;
   %r = tail call i8* @mempcpy(i8* %d, i8* %s, i64 8)
   ret i8* %r
 }
 
 define i8* @memcpy_big_const_n(i8* %d, i8* nocapture readonly %s) {
 ; CHECK-LABEL: @memcpy_big_const_n(
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 1 dereferenceable(1024) [[D:%.*]], i8* align 1 dereferenceable(1024) [[S:%.*]], i64 1024, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 1 dereferenceable(1024) [[D:%.*]], i8* nonnull align 1 dereferenceable(1024) [[S:%.*]], i64 1024, i1 false)
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i8, i8* [[D]], i64 1024
 ; CHECK-NEXT:    ret i8* [[TMP1]]
 ;
   %r = tail call i8* @mempcpy(i8* %d, i8* %s, i64 1024)
   ret i8* %r
 }
 
 declare i8* @mempcpy(i8*, i8* nocapture readonly, i64)

llvm/trunk/test/Transforms/InstCombine/memrchr.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -instcombine -S | FileCheck %s
+
+
+declare i8* @memrchr(i8*, i32, i32)
+
+define i8* @test1(i8* %str, i32 %c, i32 %n) {
+; CHECK-LABEL: @test1(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @memrchr(i8* [[STR:%.*]], i32 [[C:%.*]], i32 [[N:%.*]])
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+
+  %ret = call i8* @memrchr(i8* %str, i32 %c, i32 %n)
+  ret i8* %ret
+}
+
+define i8* @test2(i8* %str, i32 %c, i32 %n) {
+; CHECK-LABEL: @test2(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @memrchr(i8* nonnull [[STR:%.*]], i32 [[C:%.*]], i32 [[N:%.*]])
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+
+  %ret = call i8* @memrchr(i8* nonnull %str, i32 %c, i32 %n)
+  ret i8* %ret
+}
+
+define i8* @test3(i8* %str, i32 %c) {
+; CHECK-LABEL: @test3(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @memrchr(i8* [[STR:%.*]], i32 [[C:%.*]], i32 5)
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+
+  %ret = call i8* @memrchr(i8* %str, i32 %c, i32 5)
+  ret i8* %ret
+}
+
+define i8* @test4(i8* %str, i32 %c) "null-pointer-is-valid"="true" {
+; CHECK-LABEL: @test4(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @memrchr(i8* [[STR:%.*]], i32 [[C:%.*]], i32 5)
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+
+  %ret = call i8* @memrchr(i8* %str, i32 %c, i32 5)
+  ret i8* %ret
+}
+
+define i8* @test5(i8* %str, i32 %c) {
+; CHECK-LABEL: @test5(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @memrchr(i8* [[STR:%.*]], i32 [[C:%.*]], i32 0)
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+
+  %ret = call i8* @memrchr(i8* %str, i32 %c, i32 0)
+  ret i8* %ret
+}

llvm/trunk/test/Transforms/InstCombine/memset-1.ll

;
   ret i8* %call
 }
 
 ; This should not create a calloc and should not crash the compiler.
 
 define i8* @notmalloc_memset(i32 %size, i8*(i32)* %notmalloc) {
 ; CHECK-LABEL: @notmalloc_memset(
 ; CHECK-NEXT:    [[CALL1:%.*]] = call i8* [[NOTMALLOC:%.*]](i32 [[SIZE:%.*]]) #0
-; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* align 1 [[CALL1]], i8 0, i32 [[SIZE]], i1 false)
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* align 1 [[CALL1]], i8 0, i32 [[SIZE]], i1 false) #0
 ; CHECK-NEXT:    ret i8* [[CALL1]]
 ;
   %call1 = call i8* %notmalloc(i32 %size) #1
   %call2 = call i8* @memset(i8* %call1, i32 0, i32 %size) #1
   ret i8* %call2
 }
 
 ; FIXME: memset(malloc(x), 0, x) -> calloc(1, x)
 ; This doesn't fire currently because the malloc has more than one use.
 
 define float* @pr25892(i32 %size) #0 {
 ; CHECK-LABEL: @pr25892(
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[CALL:%.*]] = tail call i8* @malloc(i32 [[SIZE:%.*]]) #0
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i8* [[CALL]], null
 ; CHECK-NEXT:    br i1 [[CMP]], label [[CLEANUP:%.*]], label [[IF_END:%.*]]
 ; CHECK:       if.end:
 ; CHECK-NEXT:    [[BC:%.*]] = bitcast i8* [[CALL]] to float*
-; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* nonnull align 1 [[CALL]], i8 0, i32 [[SIZE]], i1 false)
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* nonnull align 1 [[CALL]], i8 0, i32 [[SIZE]], i1 false) #0
 ; CHECK-NEXT:    br label [[CLEANUP]]
 ; CHECK:       cleanup:
 ; CHECK-NEXT:    [[RETVAL_0:%.*]] = phi float* [ [[BC]], [[IF_END]] ], [ null, [[ENTRY:%.*]] ]
 ; CHECK-NEXT:    ret float* [[RETVAL_0]]
 ;
 entry:
   %call = tail call i8* @malloc(i32 %size) #1
   %cmp = icmp eq i8* %call, null
   br i1 %cmp, label %cleanup, label %if.end
 if.end:
   %bc = bitcast i8* %call to float*
   %call2 = tail call i8* @memset(i8* nonnull %call, i32 0, i32 %size) #1
   br label %cleanup
 cleanup:
   %retval.0 = phi float* [ %bc, %if.end ], [ null, %entry ]
   ret float* %retval.0
 }
 
 ; If there's a calloc transform, the store must also be eliminated.
 
 define i8* @buffer_is_modified_then_memset(i32 %size) {
 ; CHECK-LABEL: @buffer_is_modified_then_memset(
 ; CHECK-NEXT:    [[PTR:%.*]] = tail call i8* @malloc(i32 [[SIZE:%.*]]) #0
 ; CHECK-NEXT:    store i8 1, i8* [[PTR]], align 1
-; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* align 1 [[PTR]], i8 0, i32 [[SIZE]], i1 false)
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* nonnull align 1 [[PTR]], i8 0, i32 [[SIZE]], i1 false) #0
 ; CHECK-NEXT:    ret i8* [[PTR]]
 ;
   %ptr = tail call i8* @malloc(i32 %size) #1
   store i8 1, i8* %ptr           ;; fdata[0] = 1;
   %memset = tail call i8* @memset(i8* nonnull %ptr, i32 0, i32 %size) #1
   ret i8* %memset
 }
 
+define i8* @memset_size_select(i1 %b, i8* %ptr) {
+; CHECK-LABEL: @memset_size_select(
+; CHECK-NEXT:    [[SIZE:%.*]] = select i1 [[B:%.*]], i32 10, i32 50
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* nonnull align 1 dereferenceable(10) [[PTR:%.*]], i8 0, i32 [[SIZE]], i1 false) #0
+; CHECK-NEXT:    ret i8* [[PTR]]
+;
+  %size = select i1 %b, i32 10, i32 50
+  %memset = tail call i8* @memset(i8* nonnull %ptr, i32 0, i32 %size) #1
+  ret i8* %memset
+}
+
+
+define i8* @memset_size_select2(i1 %b, i8* %ptr) {
+; CHECK-LABEL: @memset_size_select2(
+; CHECK-NEXT:    [[SIZE:%.*]] = select i1 [[B:%.*]], i32 10, i32 50
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* nonnull align 1 dereferenceable(80) [[PTR:%.*]], i8 0, i32 [[SIZE]], i1 false) #0
+; CHECK-NEXT:    ret i8* [[PTR]]
+;
+  %size = select i1 %b, i32 10, i32 50
+  %memset = tail call i8* @memset(i8* nonnull dereferenceable(80) %ptr, i32 0, i32 %size) #1
+  ret i8* %memset
+}
+
+define i8* @memset_size_select3(i1 %b, i8* %ptr) {
+; CHECK-LABEL: @memset_size_select3(
+; CHECK-NEXT:    [[SIZE:%.*]] = select i1 [[B:%.*]], i32 10, i32 50
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* nonnull align 1 dereferenceable(40) [[PTR:%.*]], i8 0, i32 [[SIZE]], i1 false)
+; CHECK-NEXT:    ret i8* [[PTR]]
+;
+  %size = select i1 %b, i32 10, i32 50
+  %memset = tail call i8* @memset(i8* dereferenceable_or_null(40) %ptr, i32 0, i32 %size)
+  ret i8* %memset
+}
+
+define i8* @memset_size_select4(i1 %b, i8* %ptr) {
+; CHECK-LABEL: @memset_size_select4(
+; CHECK-NEXT:    [[SIZE:%.*]] = select i1 [[B:%.*]], i32 10, i32 50
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* nonnull align 1 dereferenceable(40) [[PTR:%.*]], i8 0, i32 [[SIZE]], i1 false) #0
+; CHECK-NEXT:    ret i8* [[PTR]]
+;
+  %size = select i1 %b, i32 10, i32 50
+  %memset = tail call i8* @memset(i8* nonnull dereferenceable_or_null(40) %ptr, i32 0, i32 %size) #1
+  ret i8* %memset
+}
+
+define i8* @memset_size_ashr(i1 %b, i8* %ptr, i32 %v) {
+; CHECK-LABEL: @memset_size_ashr(
+; CHECK-NEXT:    [[SIZE:%.*]] = ashr i32 -2, [[V:%.*]]
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* nonnull align 1 [[PTR:%.*]], i8 0, i32 [[SIZE]], i1 false) #0
+; CHECK-NEXT:    ret i8* [[PTR]]
+;
+  %size = ashr i32 -2, %v
+  %memset = tail call i8* @memset(i8* nonnull %ptr, i32 0, i32 %size) #1
+  ret i8* %memset
+}
+
+define i8* @memset_attrs1(i1 %b, i8* %ptr, i32 %size) {
+; CHECK-LABEL: @memset_attrs1(
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* align 1 dereferenceable_or_null(40) [[PTR:%.*]], i8 0, i32 [[SIZE:%.*]], i1 false) #0
+; CHECK-NEXT:    ret i8* [[PTR]]
+;
+  %memset = tail call i8* @memset(i8* dereferenceable_or_null(40) %ptr, i32 0, i32 %size) #1
+  ret i8* %memset
+}
+
+; be sure to drop nonnull since size is unknown and can be 0
+; do not change dereferenceable attribute
+define i8* @memset_attrs2(i1 %b, i8* %ptr, i32 %size) {
+; CHECK-LABEL: @memset_attrs2(
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* nonnull align 1 dereferenceable(40) [[PTR:%.*]], i8 0, i32 [[SIZE:%.*]], i1 false) #0
+; CHECK-NEXT:    ret i8* [[PTR]]
+;
+  %memset = tail call i8* @memset(i8* nonnull dereferenceable(40) %ptr, i32 0, i32 %size) #1
+  ret i8* %memset
+}
+
+; size is unknown, just copy attrs, no changes in attrs
+define i8* @memset_attrs3(i1 %b, i8* %ptr, i32 %size) {
+; CHECK-LABEL: @memset_attrs3(
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* nonnull align 1 dereferenceable_or_null(40) [[PTR:%.*]], i8 0, i32 [[SIZE:%.*]], i1 false) #0
+; CHECK-NEXT:    ret i8* [[PTR]]
+;
+  %memset = tail call i8* @memset(i8* nonnull dereferenceable_or_null(40) %ptr, i32 0, i32 %size) #1
+  ret i8* %memset
+}
+
+; be sure to drop nonnull since size is unknown and can be 0
+define i8* @memset_attrs4(i1 %b, i8* %ptr, i32 %size) {
+; CHECK-LABEL: @memset_attrs4(
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* nonnull align 1 [[PTR:%.*]], i8 0, i32 [[SIZE:%.*]], i1 false) #0
+; CHECK-NEXT:    ret i8* [[PTR]]
+;
+  %memset = tail call i8* @memset(i8* nonnull %ptr, i32 0, i32 %size) #1
+  ret i8* %memset
+}
+
+
 attributes #0 = { nounwind ssp uwtable }
 attributes #1 = { nounwind }
 attributes #2 = { nounwind readnone }
 

llvm/trunk/test/Transforms/InstCombine/memset_chk-1.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test lib call simplification of __memset_chk calls with various values
 ; for dstlen and len.
 ;
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 ; rdar://7719085
 
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
 
 %struct.T = type { [100 x i32], [100 x i32], [1024 x i8] }
 @t = common global %struct.T zeroinitializer
 
 ; Check cases where dstlen >= len.
 
 define i8* @test_simplify1() {
 ; CHECK-LABEL: @test_simplify1(
-; CHECK-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 4 dereferenceable(1824) bitcast (%struct.T* @t to i8*), i8 0, i64 1824, i1 false)
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i64(i8* nonnull align 4 dereferenceable(1824) bitcast (%struct.T* @t to i8*), i8 0, i64 1824, i1 false)
 ; CHECK-NEXT:    ret i8* bitcast (%struct.T* @t to i8*)
 ;
   %dst = bitcast %struct.T* @t to i8*
 
   %ret = call i8* @__memset_chk(i8* %dst, i32 0, i64 1824, i64 1824)
   ret i8* %ret
 }
 
 define i8* @test_simplify2() {
 ; CHECK-LABEL: @test_simplify2(
-; CHECK-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 4 dereferenceable(1824) bitcast (%struct.T* @t to i8*), i8 0, i64 1824, i1 false)
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i64(i8* nonnull align 4 dereferenceable(1824) bitcast (%struct.T* @t to i8*), i8 0, i64 1824, i1 false)
 ; CHECK-NEXT:    ret i8* bitcast (%struct.T* @t to i8*)
 ;
   %dst = bitcast %struct.T* @t to i8*
 
   %ret = call i8* @__memset_chk(i8* %dst, i32 0, i64 1824, i64 3648)
   ret i8* %ret
 }
 
 define i8* @test_simplify3() {
 ; CHECK-LABEL: @test_simplify3(
-; CHECK-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 4 dereferenceable(1824) bitcast (%struct.T* @t to i8*), i8 0, i64 1824, i1 false)
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i64(i8* nonnull align 4 dereferenceable(1824) bitcast (%struct.T* @t to i8*), i8 0, i64 1824, i1 false)
 ; CHECK-NEXT:    ret i8* bitcast (%struct.T* @t to i8*)
 ;
   %dst = bitcast %struct.T* @t to i8*
 
   %ret = call i8* @__memset_chk(i8* %dst, i32 0, i64 1824, i64 -1)
   ret i8* %ret
 }
 
;
   %ret = call i8* @__memset_chk(i8* %dst, i32 0, i64 1824, i64 0)
   ret i8* %ret
 }
 
 ; Test that RAUW in SimplifyLibCalls for __memset_chk generates valid IR
 define i32 @test_rauw(i8* %a, i8* %b, i8** %c) {
 ; CHECK-LABEL: @test_rauw(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[CALL49:%.*]] = call i64 @strlen(i8* [[A:%.*]])
+; CHECK-NEXT:    [[CALL49:%.*]] = call i64 @strlen(i8* nonnull dereferenceable(1) [[A:%.*]])
 ; CHECK-NEXT:    [[ADD180:%.*]] = add i64 [[CALL49]], 1
 ; CHECK-NEXT:    [[YO107:%.*]] = call i64 @llvm.objectsize.i64.p0i8(i8* [[B:%.*]], i1 false, i1 false, i1 false)
 ; CHECK-NEXT:    [[CALL50:%.*]] = call i8* @__memmove_chk(i8* [[B]], i8* [[A]], i64 [[ADD180]], i64 [[YO107]])
-; CHECK-NEXT:    [[STRLEN:%.*]] = call i64 @strlen(i8* [[B]])
+; CHECK-NEXT:    [[STRLEN:%.*]] = call i64 @strlen(i8* nonnull dereferenceable(1) [[B]])
 ; CHECK-NEXT:    [[STRCHR2:%.*]] = getelementptr i8, i8* [[B]], i64 [[STRLEN]]
 ; CHECK-NEXT:    [[TMP0:%.*]] = bitcast i8** [[C:%.*]] to i64*
 ; CHECK-NEXT:    [[D1:%.*]] = load i64, i64* [[TMP0]], align 8
 ; CHECK-NEXT:    [[SUB183:%.*]] = ptrtoint i8* [[B]] to i64
 ; CHECK-NEXT:    [[SUB184:%.*]] = sub i64 [[D1]], [[SUB183]]
 ; CHECK-NEXT:    [[ADD52_I_I:%.*]] = add nsw i64 [[SUB184]], 1
 ; CHECK-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 1 [[STRCHR2]], i8 0, i64 [[ADD52_I_I]], i1 false)
 ; CHECK-NEXT:    ret i32 4

llvm/trunk/test/Transforms/InstCombine/objsize.ll

 ; rdar://7782496
 @s = external global i8*
 
 define i8* @test5(i32 %n) nounwind ssp {
 ; CHECK-LABEL: @test5(
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = tail call noalias dereferenceable_or_null(20) i8* @malloc(i32 20) #0
 ; CHECK-NEXT:    [[TMP1:%.*]] = load i8*, i8** @s, align 8
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(10) [[TMP0]], i8* align 1 dereferenceable(10) [[TMP1]], i32 10, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(10) [[TMP0]], i8* nonnull align 1 dereferenceable(10) [[TMP1]], i32 10, i1 false)
 ; CHECK-NEXT:    ret i8* [[TMP0]]
 ;
 entry:
   %0 = tail call noalias i8* @malloc(i32 20) nounwind
   %1 = tail call i32 @llvm.objectsize.i32.p0i8(i8* %0, i1 false, i1 false, i1 false)
   %2 = load i8*, i8** @s, align 8
   %3 = tail call i8* @__memcpy_chk(i8* %0, i8* %2, i32 10, i32 %1) nounwind
   ret i8* %0

llvm/trunk/test/Transforms/InstCombine/printf-1.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the printf library call simplifier works correctly.
 ;
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 ; RUN: opt < %s -mtriple xcore-xmos-elf -instcombine -S | FileCheck %s -check-prefix=CHECK-IPRINTF
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 
 @hello_world = constant [13 x i8] c"hello world\0A\00"
 @h = constant [2 x i8] c"h\00"
 @h2 = constant [3 x i8] c"%%\00"
 @percent = constant [2 x i8] c"%\00"
 @percent_c = constant [3 x i8] c"%c\00"
 @percent_d = constant [3 x i8] c"%d\00"
 @percent_f = constant [3 x i8] c"%f\00"
 @percent_s = constant [4 x i8] c"%s\0A\00"
 @empty = constant [1 x i8] c"\00"
 ; CHECK: [[$STR:@[a-z0-9]+]] = private unnamed_addr constant [12 x i8] c"hello world\00", align 1
 
 declare i32 @printf(i8*, ...)
 
 ; Check printf("") -> noop.
 
 define void @test_simplify1() {
 ; CHECK-LABEL: @test_simplify1(
+; CHECK-NEXT:    ret void
+;
+; CHECK-IPRINTF-LABEL: @test_simplify1(
+; CHECK-IPRINTF-NEXT:    ret void
+;
   %fmt = getelementptr [1 x i8], [1 x i8]* @empty, i32 0, i32 0
   call i32 (i8*, ...) @printf(i8* %fmt)
   ret void
-; CHECK-NEXT: ret void
 }
 
 ; Check printf("x") -> putchar('x'), even for '%'.
 
 define void @test_simplify2() {
 ; CHECK-LABEL: @test_simplify2(
+; CHECK-NEXT:    [[PUTCHAR:%.*]] = call i32 @putchar(i32 104)
+; CHECK-NEXT:    ret void
+;
+; CHECK-IPRINTF-LABEL: @test_simplify2(
+; CHECK-IPRINTF-NEXT:    [[PUTCHAR:%.*]] = call i32 @putchar(i32 104)
+; CHECK-IPRINTF-NEXT:    ret void
+;
   %fmt = getelementptr [2 x i8], [2 x i8]* @h, i32 0, i32 0
   call i32 (i8*, ...) @printf(i8* %fmt)
-; CHECK-NEXT: call i32 @putchar(i32 104)
   ret void
-; CHECK-NEXT: ret void
 }
 
 ; Special case: printf("%%") -> putchar('%').
 
 define void @test_simplify2b() {
 ; CHECK-LABEL: @test_simplify2b(
+; CHECK-NEXT:    [[PUTCHAR:%.*]] = call i32 @putchar(i32 37)
+; CHECK-NEXT:    ret void
+;
+; CHECK-IPRINTF-LABEL: @test_simplify2b(
+; CHECK-IPRINTF-NEXT:    [[PUTCHAR:%.*]] = call i32 @putchar(i32 37)
+; CHECK-IPRINTF-NEXT:    ret void
+;
   %fmt = getelementptr [3 x i8], [3 x i8]* @h2, i32 0, i32 0
   call i32 (i8*, ...) @printf(i8* %fmt)
-; CHECK-NEXT: call i32 @putchar(i32 37)
   ret void
-; CHECK-NEXT: ret void
 }
 
 define void @test_simplify3() {
 ; CHECK-LABEL: @test_simplify3(
+; CHECK-NEXT:    [[PUTCHAR:%.*]] = call i32 @putchar(i32 37)
+; CHECK-NEXT:    ret void
+;
+; CHECK-IPRINTF-LABEL: @test_simplify3(
+; CHECK-IPRINTF-NEXT:    [[PUTCHAR:%.*]] = call i32 @putchar(i32 37)
+; CHECK-IPRINTF-NEXT:    ret void
+;
   %fmt = getelementptr [2 x i8], [2 x i8]* @percent, i32 0, i32 0
   call i32 (i8*, ...) @printf(i8* %fmt)
-; CHECK-NEXT: call i32 @putchar(i32 37)
   ret void
-; CHECK-NEXT: ret void
 }
 
 ; Check printf("foo\n") -> puts("foo").
 
 define void @test_simplify4() {
 ; CHECK-LABEL: @test_simplify4(
+; CHECK-NEXT:    [[PUTS:%.*]] = call i32 @puts(i8* nonnull dereferenceable(1) getelementptr inbounds ([12 x i8], [12 x i8]* @str, i32 0, i32 0))
+; CHECK-NEXT:    ret void
+;
+; CHECK-IPRINTF-LABEL: @test_simplify4(
+; CHECK-IPRINTF-NEXT:    [[PUTS:%.*]] = call i32 @puts(i8* nonnull dereferenceable(1) getelementptr inbounds ([12 x i8], [12 x i8]* @str, i32 0, i32 0))
+; CHECK-IPRINTF-NEXT:    ret void
+;
   %fmt = getelementptr [13 x i8], [13 x i8]* @hello_world, i32 0, i32 0
   call i32 (i8*, ...) @printf(i8* %fmt)
-; CHECK-NEXT: call i32 @puts(i8* getelementptr inbounds ([12 x i8], [12 x i8]* [[$STR]], i32 0, i32 0))
   ret void
-; CHECK-NEXT: ret void
 }
 
 ; Check printf("%c", chr) -> putchar(chr).
 
 define void @test_simplify5() {
 ; CHECK-LABEL: @test_simplify5(
+; CHECK-NEXT:    [[PUTCHAR:%.*]] = call i32 @putchar(i32 104)
+; CHECK-NEXT:    ret void
+;
+; CHECK-IPRINTF-LABEL: @test_simplify5(
+; CHECK-IPRINTF-NEXT:    [[PUTCHAR:%.*]] = call i32 @putchar(i32 104)
+; CHECK-IPRINTF-NEXT:    ret void
+;
   %fmt = getelementptr [3 x i8], [3 x i8]* @percent_c, i32 0, i32 0
   call i32 (i8*, ...) @printf(i8* %fmt, i8 104)
-; CHECK-NEXT: call i32 @putchar(i32 104)
   ret void
-; CHECK-NEXT: ret void
 }
 
 ; Check printf("%s\n", str) -> puts(str).
 
 define void @test_simplify6() {
 ; CHECK-LABEL: @test_simplify6(
+; CHECK-NEXT:    [[PUTS:%.*]] = call i32 @puts(i8* nonnull dereferenceable(1) getelementptr inbounds ([13 x i8], [13 x i8]* @hello_world, i32 0, i32 0))
+; CHECK-NEXT:    ret void
+;
+; CHECK-IPRINTF-LABEL: @test_simplify6(
+; CHECK-IPRINTF-NEXT:    [[PUTS:%.*]] = call i32 @puts(i8* nonnull dereferenceable(1) getelementptr inbounds ([13 x i8], [13 x i8]* @hello_world, i32 0, i32 0))
+; CHECK-IPRINTF-NEXT:    ret void
+;
   %fmt = getelementptr [4 x i8], [4 x i8]* @percent_s, i32 0, i32 0
   %str = getelementptr [13 x i8], [13 x i8]* @hello_world, i32 0, i32 0
   call i32 (i8*, ...) @printf(i8* %fmt, i8* %str)
-; CHECK-NEXT: call i32 @puts(i8* getelementptr inbounds ([13 x i8], [13 x i8]* @hello_world, i32 0, i32 0))
   ret void
-; CHECK-NEXT: ret void
 }
 
 ; Check printf(format, ...) -> iprintf(format, ...) if no floating point.
 
 define void @test_simplify7() {
+; CHECK-LABEL: @test_simplify7(
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 (i8*, ...) @printf(i8* nonnull dereferenceable(1) getelementptr inbounds ([3 x i8], [3 x i8]* @percent_d, i32 0, i32 0), i32 187)
+; CHECK-NEXT:    ret void
+;
 ; CHECK-IPRINTF-LABEL: @test_simplify7(
+; CHECK-IPRINTF-NEXT:    [[TMP1:%.*]] = call i32 (i8*, ...) @iprintf(i8* getelementptr inbounds ([3 x i8], [3 x i8]* @percent_d, i32 0, i32 0), i32 187)
+; CHECK-IPRINTF-NEXT:    ret void
+;
   %fmt = getelementptr [3 x i8], [3 x i8]* @percent_d, i32 0, i32 0
   call i32 (i8*, ...) @printf(i8* %fmt, i32 187)
-; CHECK-IPRINTF-NEXT: call i32 (i8*, ...) @iprintf(i8* getelementptr inbounds ([3 x i8], [3 x i8]* @percent_d, i32 0, i32 0), i32 187)
   ret void
-; CHECK-IPRINTF-NEXT: ret void
 }
 
 define void @test_no_simplify1() {
+; CHECK-LABEL: @test_no_simplify1(
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 (i8*, ...) @printf(i8* nonnull dereferenceable(1) getelementptr inbounds ([3 x i8], [3 x i8]* @percent_f, i32 0, i32 0), double 1.870000e+00)
+; CHECK-NEXT:    ret void
+;
 ; CHECK-IPRINTF-LABEL: @test_no_simplify1(
+; CHECK-IPRINTF-NEXT:    [[TMP1:%.*]] = call i32 (i8*, ...) @printf(i8* nonnull dereferenceable(1) getelementptr inbounds ([3 x i8], [3 x i8]* @percent_f, i32 0, i32 0), double 1.870000e+00)
+; CHECK-IPRINTF-NEXT:    ret void
+;
   %fmt = getelementptr [3 x i8], [3 x i8]* @percent_f, i32 0, i32 0
   call i32 (i8*, ...) @printf(i8* %fmt, double 1.87)
-; CHECK-IPRINTF-NEXT: call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([3 x i8], [3 x i8]* @percent_f, i32 0, i32 0), double 1.870000e+00)
   ret void
-; CHECK-IPRINTF-NEXT: ret void
 }
 
 define void @test_no_simplify2(i8* %fmt, double %d) {
 ; CHECK-LABEL: @test_no_simplify2(
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 (i8*, ...) @printf(i8* nonnull dereferenceable(1) [[FMT:%.*]], double [[D:%.*]])
+; CHECK-NEXT:    ret void
+;
+; CHECK-IPRINTF-LABEL: @test_no_simplify2(
+; CHECK-IPRINTF-NEXT:    [[TMP1:%.*]] = call i32 (i8*, ...) @printf(i8* nonnull dereferenceable(1) [[FMT:%.*]], double [[D:%.*]])
+; CHECK-IPRINTF-NEXT:    ret void
+;
   call i32 (i8*, ...) @printf(i8* %fmt, double %d)
-; CHECK-NEXT: call i32 (i8*, ...) @printf(i8* %fmt, double %d)
   ret void
-; CHECK-NEXT: ret void
 }
 
 define i32 @test_no_simplify3() {
 ; CHECK-LABEL: @test_no_simplify3(
+; CHECK-NEXT:    [[RET:%.*]] = call i32 (i8*, ...) @printf(i8* nonnull dereferenceable(1) getelementptr inbounds ([2 x i8], [2 x i8]* @h, i32 0, i32 0))
+; CHECK-NEXT:    ret i32 [[RET]]
+;
+; CHECK-IPRINTF-LABEL: @test_no_simplify3(
+; CHECK-IPRINTF-NEXT:    [[TMP1:%.*]] = call i32 (i8*, ...) @iprintf(i8* getelementptr inbounds ([2 x i8], [2 x i8]* @h, i32 0, i32 0))
+; CHECK-IPRINTF-NEXT:    ret i32 [[TMP1]]
+;
   %fmt = getelementptr [2 x i8], [2 x i8]* @h, i32 0, i32 0
   %ret = call i32 (i8*, ...) @printf(i8* %fmt)
-; CHECK-NEXT: call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([2 x i8], [2 x i8]* @h, i32 0, i32 0))
   ret i32 %ret
-; CHECK-NEXT: ret i32 %ret
 }

llvm/trunk/test/Transforms/InstCombine/printf-2.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the printf library call simplifier works correctly.
 ;
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 
 @hello_world = constant [13 x i8] c"hello world\0A\00"
 @h = constant [2 x i8] c"h\00"
 @percent_s = constant [4 x i8] c"%s\0A\00"
 @format_str = constant [3 x i8] c"%s\00"
 @charstr = constant [2 x i8] c"a\00"
 
 declare void @printf(i8*, ...)
 
 ; Check simplification of printf with void return type.
 
 define void @test_simplify1() {
 ; CHECK-LABEL: @test_simplify1(
+; CHECK-NEXT:    [[PUTCHAR:%.*]] = call i32 @putchar(i32 104)
+; CHECK-NEXT:    ret void
+;
   %fmt = getelementptr [2 x i8], [2 x i8]* @h, i32 0, i32 0
   call void (i8*, ...) @printf(i8* %fmt)
-; CHECK-NEXT: call i32 @putchar(i32 104)
   ret void
-; CHECK-NEXT: ret void
 }
 
 define void @test_simplify2() {
 ; CHECK-LABEL: @test_simplify2(
+; CHECK-NEXT:    [[PUTS:%.*]] = call i32 @puts(i8* nonnull dereferenceable(1) getelementptr inbounds ([12 x i8], [12 x i8]* @str, i32 0, i32 0))
+; CHECK-NEXT:    ret void
+;
   %fmt = getelementptr [13 x i8], [13 x i8]* @hello_world, i32 0, i32 0
   call void (i8*, ...) @printf(i8* %fmt)
-; CHECK-NEXT: call i32 @puts(i8* getelementptr inbounds ([12 x i8], [12 x i8]* @str, i32 0, i32 0))
   ret void
-; CHECK-NEXT: ret void
 }
 
 define void @test_simplify6() {
 ; CHECK-LABEL: @test_simplify6(
+; CHECK-NEXT:    [[PUTS:%.*]] = call i32 @puts(i8* nonnull dereferenceable(1) getelementptr inbounds ([13 x i8], [13 x i8]* @hello_world, i32 0, i32 0))
+; CHECK-NEXT:    ret void
+;
   %fmt = getelementptr [4 x i8], [4 x i8]* @percent_s, i32 0, i32 0
   %str = getelementptr [13 x i8], [13 x i8]* @hello_world, i32 0, i32 0
   call void (i8*, ...) @printf(i8* %fmt, i8* %str)
-; CHECK-NEXT: call i32 @puts(i8* getelementptr inbounds ([13 x i8], [13 x i8]* @hello_world, i32 0, i32 0))
   ret void
-; CHECK-NEXT: ret void
 }
 
 define void @test_simplify7() {
 ; CHECK-LABEL: @test_simplify7(
+; CHECK-NEXT:    [[PUTCHAR:%.*]] = call i32 @putchar(i32 97)
+; CHECK-NEXT:    ret void
+;
   %fmt = getelementptr [3 x i8], [3 x i8]* @format_str, i32 0, i32 0
   %str = getelementptr [2 x i8], [2 x i8]* @charstr, i32 0, i32 0
   call void (i8*, ...) @printf(i8* %fmt, i8* %str)
-; CHECK-NEXT: call i32 @putchar(i32 97)
   ret void
-; CHECK-NEXT: ret void
 }

llvm/trunk/test/Transforms/InstCombine/puts-1.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the puts library call simplifier works correctly.
 ;
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 
 @empty = constant [1 x i8] zeroinitializer
 
 declare i32 @puts(i8*)
 
 ; Check puts("") -> putchar('\n').
 
 define void @test_simplify1() {
 ; CHECK-LABEL: @test_simplify1(
+; CHECK-NEXT:    [[PUTCHAR:%.*]] = call i32 @putchar(i32 10)
+; CHECK-NEXT:    ret void
+;
   %str = getelementptr [1 x i8], [1 x i8]* @empty, i32 0, i32 0
   call i32 @puts(i8* %str)
-; CHECK-NEXT: call i32 @putchar(i32 10)
   ret void
-; CHECK-NEXT: ret void
 }
 
 ; Don't simplify if the return value is used.
 
 define i32 @test_no_simplify1() {
 ; CHECK-LABEL: @test_no_simplify1(
+; CHECK-NEXT:    [[RET:%.*]] = call i32 @puts(i8* nonnull dereferenceable(1) getelementptr inbounds ([1 x i8], [1 x i8]* @empty, i32 0, i32 0))
+; CHECK-NEXT:    ret i32 [[RET]]
+;
   %str = getelementptr [1 x i8], [1 x i8]* @empty, i32 0, i32 0
   %ret = call i32 @puts(i8* %str)
-; CHECK-NEXT: call i32 @puts(i8* getelementptr inbounds ([1 x i8], [1 x i8]* @empty, i32 0, i32 0))
   ret i32 %ret
-; CHECK-NEXT: ret i32 %ret
 }

llvm/trunk/test/Transforms/InstCombine/snprintf.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 @.str = private unnamed_addr constant [4 x i8] c"str\00", align 1
 @.str.1 = private unnamed_addr constant [3 x i8] c"%%\00", align 1
 @.str.2 = private unnamed_addr constant [3 x i8] c"%c\00", align 1
 @.str.3 = private unnamed_addr constant [3 x i8] c"%s\00", align 1
 
 declare i32 @snprintf(i8*, i64, i8*, ...) #1
 
 define void @test_not_const_fmt(i8* %buf, i8* %fmt) #0 {
 ; CHECK-LABEL: @test_not_const_fmt(
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 (i8*, i64, i8*, ...) @snprintf(i8* [[BUF:%.*]], i64 32, i8* [[FMT:%.*]])
+; CHECK-NEXT:    [[CALL:%.*]] = call i32 (i8*, i64, i8*, ...) @snprintf(i8* nonnull dereferenceable(1) [[BUF:%.*]], i64 32, i8* [[FMT:%.*]])
 ; CHECK-NEXT:    ret void
 ;
   %call = call i32 (i8*, i64, i8*, ...) @snprintf(i8* %buf, i64 32, i8* %fmt) #2
   ret void
 }
 
+; size is '0', do not add nonnull attribute
 define void @test_not_const_fmt_zero_size_return_value(i8* %buf, i8* %fmt) #0 {
 ; CHECK-LABEL: @test_not_const_fmt_zero_size_return_value(
 ; CHECK-NEXT:    [[CALL:%.*]] = call i32 (i8*, i64, i8*, ...) @snprintf(i8* [[BUF:%.*]], i64 0, i8* [[FMT:%.*]])
 ; CHECK-NEXT:    ret void
 ;
   %call = call i32 (i8*, i64, i8*, ...) @snprintf(i8* %buf, i64 0, i8* %fmt) #2
   ret void
 }
 
-
 define void @test_not_const_size(i8* %buf, i64 %size) #0 {
 ; CHECK-LABEL: @test_not_const_size(
 ; CHECK-NEXT:    [[CALL:%.*]] = call i32 (i8*, i64, i8*, ...) @snprintf(i8* [[BUF:%.*]], i64 [[SIZE:%.*]], i8* getelementptr inbounds ([4 x i8], [4 x i8]* @.str, i64 0, i64 0))
 ; CHECK-NEXT:    ret void
 ;
   %call = call i32 (i8*, i64, i8*, ...) @snprintf(i8* %buf, i64 %size, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @.str, i64 0, i64 0)) #2
   ret void
 }
 
 
 define i32 @test_return_value(i8* %buf) #0 {
 ; CHECK-LABEL: @test_return_value(
 ; CHECK-NEXT:    ret i32 3
 ;
   %call = call i32 (i8*, i64, i8*, ...) @snprintf(i8* %buf, i64 0, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @.str, i64 0, i64 0)) #2
   ret i32 %call
 }
 
 define void @test_percentage(i8* %buf) #0 {
 ; CHECK-LABEL: @test_percentage(
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 (i8*, i64, i8*, ...) @snprintf(i8* [[BUF:%.*]], i64 32, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @.str.1, i64 0, i64 0))
+; CHECK-NEXT:    [[CALL:%.*]] = call i32 (i8*, i64, i8*, ...) @snprintf(i8* nonnull dereferenceable(1) [[BUF:%.*]], i64 32, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @.str.1, i64 0, i64 0))
 ; CHECK-NEXT:    ret void
 ;
   %call = call i32 (i8*, i64, i8*, ...) @snprintf(i8* %buf, i64 32, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @.str.1, i64 0, i64 0)) #2
   ret void
 }
 
 define i32 @test_null_buf_return_value() #0 {
 ; CHECK-LABEL: @test_null_buf_return_value(

 ; CHECK-NEXT:    ret i32 1
 ;
   %call = call i32 (i8*, i64, i8*, ...) @snprintf(i8* %buf, i64 0, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @.str.2, i64 0, i64 0), i32 65) #2
   ret i32 %call
 }
 
 define i32 @test_char_wrong_size(i8* %buf) #0 {
 ; CHECK-LABEL: @test_char_wrong_size(
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 (i8*, i64, i8*, ...) @snprintf(i8* [[BUF:%.*]], i64 1, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @.str.2, i64 0, i64 0), i32 65)
+; CHECK-NEXT:    [[CALL:%.*]] = call i32 (i8*, i64, i8*, ...) @snprintf(i8* nonnull dereferenceable(1) [[BUF:%.*]], i64 1, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @.str.2, i64 0, i64 0), i32 65)
 ; CHECK-NEXT:    ret i32 [[CALL]]
 ;
   %call = call i32 (i8*, i64, i8*, ...) @snprintf(i8* %buf, i64 1, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @.str.2, i64 0, i64 0), i32 65) #2
   ret i32 %call
 }
 
 define i32 @test_char_ok_size(i8* %buf) #0 {
 ; CHECK-LABEL: @test_char_ok_size(

 ; CHECK-NEXT:    ret i32 3
 ;
   %call = call i32 (i8*, i64, i8*, ...) @snprintf(i8* %buf, i64 0, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @.str.3, i64 0, i64 0), i8* getelementptr inbounds ([4 x i8], [4 x i8]* @.str, i64 0, i64 0)) #2
   ret i32 %call
 }
 
 define i32 @test_str_wrong_size(i8* %buf) #0 {
 ; CHECK-LABEL: @test_str_wrong_size(
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 (i8*, i64, i8*, ...) @snprintf(i8* [[BUF:%.*]], i64 1, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @.str.3, i64 0, i64 0), i8* getelementptr inbounds ([4 x i8], [4 x i8]* @.str, i64 0, i64 0))
+; CHECK-NEXT:    [[CALL:%.*]] = call i32 (i8*, i64, i8*, ...) @snprintf(i8* nonnull dereferenceable(1) [[BUF:%.*]], i64 1, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @.str.3, i64 0, i64 0), i8* getelementptr inbounds ([4 x i8], [4 x i8]* @.str, i64 0, i64 0))
 ; CHECK-NEXT:    ret i32 [[CALL]]
 ;
   %call = call i32 (i8*, i64, i8*, ...) @snprintf(i8* %buf, i64 1, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @.str.3, i64 0, i64 0), i8* getelementptr inbounds ([4 x i8], [4 x i8]* @.str, i64 0, i64 0)) #2
   ret i32 %call
 }
 
 define i32 @test_str_ok_size(i8* %buf) #0 {
 ; CHECK-LABEL: @test_str_ok_size(
 ; CHECK-NEXT:    [[TMP1:%.*]] = bitcast i8* [[BUF:%.*]] to i32*
 ; CHECK-NEXT:    store i32 7500915, i32* [[TMP1]], align 1
 ; CHECK-NEXT:    ret i32 3
 ;
   %call = call i32 (i8*, i64, i8*, ...) @snprintf(i8* %buf, i64 32, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @.str.3, i64 0, i64 0), i8* getelementptr inbounds ([4 x i8], [4 x i8]* @.str, i64 0, i64 0)) #2
   ret i32 %call
 }

llvm/trunk/test/Transforms/InstCombine/sprintf-1.ll

 @percent_s = constant [3 x i8] c"%s\00"
 
 declare i32 @sprintf(i8*, i8*, ...)
 
 ; Check sprintf(dst, fmt) -> llvm.memcpy(str, fmt, strlen(fmt) + 1, 1).
 
 define void @test_simplify1(i8* %dst) {
 ; CHECK-LABEL: @test_simplify1(
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(13) [[DST:%.*]], i8* align 1 dereferenceable(13) getelementptr inbounds ([13 x i8], [13 x i8]* @hello_world, i32 0, i32 0), i32 13, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(13) [[DST:%.*]], i8* nonnull align 1 dereferenceable(13) getelementptr inbounds ([13 x i8], [13 x i8]* @hello_world, i32 0, i32 0), i32 13, i1 false)
 ; CHECK-NEXT:    ret void
 ;
 ; CHECK-IPRINTF-LABEL: @test_simplify1(
-; CHECK-IPRINTF-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(13) [[DST:%.*]], i8* align 1 dereferenceable(13) getelementptr inbounds ([13 x i8], [13 x i8]* @hello_world, i32 0, i32 0), i32 13, i1 false)
+; CHECK-IPRINTF-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(13) [[DST:%.*]], i8* nonnull align 1 dereferenceable(13) getelementptr inbounds ([13 x i8], [13 x i8]* @hello_world, i32 0, i32 0), i32 13, i1 false)
 ; CHECK-IPRINTF-NEXT:    ret void
 ;
   %fmt = getelementptr [13 x i8], [13 x i8]* @hello_world, i32 0, i32 0
   call i32 (i8*, i8*, ...) @sprintf(i8* %dst, i8* %fmt)
   ret void
 }
 
 define void @test_simplify2(i8* %dst) {
;
   call i32 (i8*, i8*, ...) @sprintf(i8* %dst, i8* %fmt, i8 104)
   ret void
 }
 
 ; Check sprintf(dst, "%s", str) -> llvm.memcpy(dest, str, strlen(str) + 1, 1).
 
 define void @test_simplify5(i8* %dst, i8* %str) {
 ; CHECK-LABEL: @test_simplify5(
-; CHECK-NEXT:    [[STRLEN:%.*]] = call i32 @strlen(i8* [[STR:%.*]])
+; CHECK-NEXT:    [[STRLEN:%.*]] = call i32 @strlen(i8* nonnull dereferenceable(1) [[STR:%.*]])
 ; CHECK-NEXT:    [[LENINC:%.*]] = add i32 [[STRLEN]], 1
 ; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 [[DST:%.*]], i8* align 1 [[STR]], i32 [[LENINC]], i1 false)
 ; CHECK-NEXT:    ret void
 ;
 ; CHECK-IPRINTF-LABEL: @test_simplify5(
-; CHECK-IPRINTF-NEXT:    [[STRLEN:%.*]] = call i32 @strlen(i8* [[STR:%.*]])
+; CHECK-IPRINTF-NEXT:    [[STRLEN:%.*]] = call i32 @strlen(i8* nonnull dereferenceable(1) [[STR:%.*]])
 ; CHECK-IPRINTF-NEXT:    [[LENINC:%.*]] = add i32 [[STRLEN]], 1
 ; CHECK-IPRINTF-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 [[DST:%.*]], i8* align 1 [[STR]], i32 [[LENINC]], i1 false)
 ; CHECK-IPRINTF-NEXT:    ret void
 ;
   %fmt = getelementptr [3 x i8], [3 x i8]* @percent_s, i32 0, i32 0
   call i32 (i8*, i8*, ...) @sprintf(i8* %dst, i8* %fmt, i8* %str)
   ret void
 }
 
 ; Check sprintf(dst, format, ...) -> siprintf(str, format, ...) if no floating.
 
 define void @test_simplify6(i8* %dst) {
 ; CHECK-LABEL: @test_simplify6(
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 (i8*, i8*, ...) @sprintf(i8* [[DST:%.*]], i8* getelementptr inbounds ([3 x i8], [3 x i8]* @percent_d, i32 0, i32 0), i32 187)
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 (i8*, i8*, ...) @sprintf(i8* nonnull dereferenceable(1) [[DST:%.*]], i8* nonnull dereferenceable(1) getelementptr inbounds ([3 x i8], [3 x i8]* @percent_d, i32 0, i32 0), i32 187)
 ; CHECK-NEXT:    ret void
 ;
 ; CHECK-IPRINTF-LABEL: @test_simplify6(
 ; CHECK-IPRINTF-NEXT:    [[TMP1:%.*]] = call i32 (i8*, i8*, ...) @siprintf(i8* [[DST:%.*]], i8* getelementptr inbounds ([3 x i8], [3 x i8]* @percent_d, i32 0, i32 0), i32 187)
 ; CHECK-IPRINTF-NEXT:    ret void
 ;
   %fmt = getelementptr [3 x i8], [3 x i8]* @percent_d, i32 0, i32 0
   call i32 (i8*, i8*, ...) @sprintf(i8* %dst, i8* %fmt, i32 187)
   ret void
 }
 
 define void @test_no_simplify1(i8* %dst) {
 ; CHECK-LABEL: @test_no_simplify1(
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 (i8*, i8*, ...) @sprintf(i8* [[DST:%.*]], i8* getelementptr inbounds ([3 x i8], [3 x i8]* @percent_f, i32 0, i32 0), double 1.870000e+00)
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 (i8*, i8*, ...) @sprintf(i8* nonnull dereferenceable(1) [[DST:%.*]], i8* nonnull dereferenceable(1) getelementptr inbounds ([3 x i8], [3 x i8]* @percent_f, i32 0, i32 0), double 1.870000e+00)
 ; CHECK-NEXT:    ret void
 ;
 ; CHECK-IPRINTF-LABEL: @test_no_simplify1(
-; CHECK-IPRINTF-NEXT:    [[TMP1:%.*]] = call i32 (i8*, i8*, ...) @sprintf(i8* [[DST:%.*]], i8* getelementptr inbounds ([3 x i8], [3 x i8]* @percent_f, i32 0, i32 0), double 1.870000e+00)
+; CHECK-IPRINTF-NEXT:    [[TMP1:%.*]] = call i32 (i8*, i8*, ...) @sprintf(i8* nonnull dereferenceable(1) [[DST:%.*]], i8* nonnull dereferenceable(1) getelementptr inbounds ([3 x i8], [3 x i8]* @percent_f, i32 0, i32 0), double 1.870000e+00)
 ; CHECK-IPRINTF-NEXT:    ret void
 ;
   %fmt = getelementptr [3 x i8], [3 x i8]* @percent_f, i32 0, i32 0
   call i32 (i8*, i8*, ...) @sprintf(i8* %dst, i8* %fmt, double 1.87)
   ret void
 }
 
 define void @test_no_simplify2(i8* %dst, i8* %fmt, double %d) {
 ; CHECK-LABEL: @test_no_simplify2(
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 (i8*, i8*, ...) @sprintf(i8* [[DST:%.*]], i8* [[FMT:%.*]], double [[D:%.*]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 (i8*, i8*, ...) @sprintf(i8* nonnull dereferenceable(1) [[DST:%.*]], i8* nonnull dereferenceable(1) [[FMT:%.*]], double [[D:%.*]])
 ; CHECK-NEXT:    ret void
 ;
 ; CHECK-IPRINTF-LABEL: @test_no_simplify2(
-; CHECK-IPRINTF-NEXT:    [[TMP1:%.*]] = call i32 (i8*, i8*, ...) @sprintf(i8* [[DST:%.*]], i8* [[FMT:%.*]], double [[D:%.*]])
+; CHECK-IPRINTF-NEXT:    [[TMP1:%.*]] = call i32 (i8*, i8*, ...) @sprintf(i8* nonnull dereferenceable(1) [[DST:%.*]], i8* nonnull dereferenceable(1) [[FMT:%.*]], double [[D:%.*]])
 ; CHECK-IPRINTF-NEXT:    ret void
 ;
   call i32 (i8*, i8*, ...) @sprintf(i8* %dst, i8* %fmt, double %d)
   ret void
 }

llvm/trunk/test/Transforms/InstCombine/stpcpy-1.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the stpcpy library call simplifier works correctly.
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 ;
 ; This transformation requires the pointer size, as it assumes that size_t is
 ; the size of a pointer.
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128-n8:16:32"
 
 @hello = constant [6 x i8] c"hello\00"
 @a = common global [32 x i8] zeroinitializer, align 1
 @b = common global [32 x i8] zeroinitializer, align 1
 
 declare i8* @stpcpy(i8*, i8*)
 
 define i8* @test_simplify1() {
 ; CHECK-LABEL: @test_simplify1(
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(6) getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0), i8* align 1 dereferenceable(6) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 6, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(6) getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0), i8* nonnull align 1 dereferenceable(6) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 6, i1 false)
 ; CHECK-NEXT:    ret i8* getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 5)
 ;
 
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
 
   %ret = call i8* @stpcpy(i8* %dst, i8* %src)
   ret i8* %ret
 }
 
 define i8* @test_simplify2() {
 ; CHECK-LABEL: @test_simplify2(
-; CHECK-NEXT:    [[STRLEN:%.*]] = call i32 @strlen(i8* getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0))
+; CHECK-NEXT:    [[STRLEN:%.*]] = call i32 @strlen(i8* nonnull dereferenceable(1) getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0))
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [32 x i8], [32 x i8]* @a, i32 0, i32 [[STRLEN]]
 ; CHECK-NEXT:    ret i8* [[TMP1]]
 ;
 
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
 
   %ret = call i8* @stpcpy(i8* %dst, i8* %dst)
   ret i8* %ret
 }
 
+define void @test_simplify3(i8* %dst) {
+; CHECK-LABEL: @test_simplify3(
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(80) [[DST:%.*]], i8* nonnull align 1 dereferenceable(6) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 6, i1 false)
+; CHECK-NEXT:    ret void
+;
+
+  %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
+
+  call i8* @stpcpy(i8* dereferenceable(80) %dst, i8* %src)
+  ret void
+}
+
 define i8* @test_no_simplify1() {
 ; CHECK-LABEL: @test_no_simplify1(
 ; CHECK-NEXT:    [[RET:%.*]] = call i8* @stpcpy(i8* getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0), i8* getelementptr inbounds ([32 x i8], [32 x i8]* @b, i32 0, i32 0))
 ; CHECK-NEXT:    ret i8* [[RET]]
 ;
 
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %src = getelementptr [32 x i8], [32 x i8]* @b, i32 0, i32 0
 
   %ret = call i8* @stpcpy(i8* %dst, i8* %src)
   ret i8* %ret
 }

llvm/trunk/test/Transforms/InstCombine/stpcpy_chk-1.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test lib call simplification of __stpcpy_chk calls with various values
 ; for src, dst, and slen.
 ;
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 
 @a = common global [60 x i8] zeroinitializer, align 1
 @b = common global [60 x i8] zeroinitializer, align 1
 @.str = private constant [12 x i8] c"abcdefghijk\00"
 
 ; Check cases where slen >= strlen (src).
 
 define i8* @test_simplify1() {
 ; CHECK-LABEL: @test_simplify1(
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
 ; CHECK-NEXT:    ret i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 11)
 ;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [12 x i8], [12 x i8]* @.str, i32 0, i32 0
 
   %ret = call i8* @__stpcpy_chk(i8* %dst, i8* %src, i32 60)
   ret i8* %ret
 }
 
 define i8* @test_simplify2() {
 ; CHECK-LABEL: @test_simplify2(
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
 ; CHECK-NEXT:    ret i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 11)
 ;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [12 x i8], [12 x i8]* @.str, i32 0, i32 0
 
   %ret = call i8* @__stpcpy_chk(i8* %dst, i8* %src, i32 12)
   ret i8* %ret
 }
 
 define i8* @test_simplify3() {
 ; CHECK-LABEL: @test_simplify3(
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
 ; CHECK-NEXT:    ret i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 11)
 ;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [12 x i8], [12 x i8]* @.str, i32 0, i32 0
 
   %ret = call i8* @__stpcpy_chk(i8* %dst, i8* %src, i32 -1)
   ret i8* %ret
 }
;
   %ret = call i8* @__stpcpy_chk(i8* %dst, i8* %src, i32 %len)
   ret i8* %ret
 }
 
 ; Check case where the source and destination are the same.
 
 define i8* @test_simplify6() {
 ; CHECK-LABEL: @test_simplify6(
-; CHECK-NEXT:    [[STRLEN:%.*]] = call i32 @strlen(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0))
+; CHECK-NEXT:    [[STRLEN:%.*]] = call i32 @strlen(i8* nonnull dereferenceable(1) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0))
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 [[STRLEN]]
 ; CHECK-NEXT:    ret i8* [[TMP1]]
 ;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
 
   %len = call i32 @llvm.objectsize.i32.p0i8(i8* %dst, i1 false, i1 false, i1 false)
   %ret = call i8* @__stpcpy_chk(i8* %dst, i8* %dst, i32 %len)
   ret i8* %ret

llvm/trunk/test/Transforms/InstCombine/strchr-1.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the strchr library call simplifier works correctly.
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128-n8:16:32"
 
 @hello = constant [14 x i8] c"hello world\5Cn\00"
 @null = constant [1 x i8] zeroinitializer
 @newlines = constant [3 x i8] c"\0D\0A\00"
 @chp = global i8* zeroinitializer
 
 declare i8* @strchr(i8*, i32)
 
 define void @test_simplify1() {
-; CHECK: store i8* getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 6)
-; CHECK-NOT: call i8* @strchr
-; CHECK: ret void
+; CHECK-LABEL: @test_simplify1(
+; CHECK-NEXT:    store i8* getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 6), i8** @chp, align 4
+; CHECK-NEXT:    ret void
+;
 
   %str = getelementptr [14 x i8], [14 x i8]* @hello, i32 0, i32 0
   %dst = call i8* @strchr(i8* %str, i32 119)
   store i8* %dst, i8** @chp
   ret void
 }
 
 define void @test_simplify2() {
-; CHECK: store i8* null, i8** @chp, align 4
-; CHECK-NOT: call i8* @strchr
-; CHECK: ret void
+; CHECK-LABEL: @test_simplify2(
+; CHECK-NEXT:    store i8* null, i8** @chp, align 4
+; CHECK-NEXT:    ret void
+;
 
   %str = getelementptr [1 x i8], [1 x i8]* @null, i32 0, i32 0
   %dst = call i8* @strchr(i8* %str, i32 119)
   store i8* %dst, i8** @chp
   ret void
 }
 
 define void @test_simplify3() {
-; CHECK: store i8* getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 13)
-; CHECK-NOT: call i8* @strchr
-; CHECK: ret void
+; CHECK-LABEL: @test_simplify3(
+; CHECK-NEXT:    store i8* getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 13), i8** @chp, align 4
+; CHECK-NEXT:    ret void
+;
 
   %src = getelementptr [14 x i8], [14 x i8]* @hello, i32 0, i32 0
   %dst = call i8* @strchr(i8* %src, i32 0)
   store i8* %dst, i8** @chp
   ret void
 }
 
 define void @test_simplify4(i32 %chr) {
-; CHECK: call i8* @memchr
-; CHECK-NOT: call i8* @strchr
-; CHECK: ret void
+; CHECK-LABEL: @test_simplify4(
+; CHECK-NEXT:    [[MEMCHR:%.*]] = call i8* @memchr(i8* nonnull dereferenceable(14) getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 0), i32 [[CHR:%.*]], i32 14)
+; CHECK-NEXT:    store i8* [[MEMCHR]], i8** @chp, align 4
+; CHECK-NEXT:    ret void
+;
 
   %src = getelementptr [14 x i8], [14 x i8]* @hello, i32 0, i32 0
   %dst = call i8* @strchr(i8* %src, i32 %chr)
   store i8* %dst, i8** @chp
   ret void
 }
 
 define void @test_simplify5() {
-; CHECK: store i8* getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 13)
-; CHECK-NOT: call i8* @strchr
-; CHECK: ret void
+; CHECK-LABEL: @test_simplify5(
+; CHECK-NEXT:    store i8* getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 13), i8** @chp, align 4
+; CHECK-NEXT:    ret void
+;
 
   %src = getelementptr [14 x i8], [14 x i8]* @hello, i32 0, i32 0
   %dst = call i8* @strchr(i8* %src, i32 65280)
   store i8* %dst, i8** @chp
   ret void
 }
 
 ; Check transformation strchr(p, 0) -> p + strlen(p)
 define void @test_simplify6(i8* %str) {
-; CHECK: %strlen = call i32 @strlen(i8* %str)
-; CHECK-NOT: call i8* @strchr
-; CHECK: %strchr = getelementptr i8, i8* %str, i32 %strlen
-; CHECK: store i8* %strchr, i8** @chp, align 4
-; CHECK: ret void
+; CHECK-LABEL: @test_simplify6(
+; CHECK-NEXT:    [[STRLEN:%.*]] = call i32 @strlen(i8* nonnull dereferenceable(1) [[STR:%.*]])
+; CHECK-NEXT:    [[STRCHR:%.*]] = getelementptr i8, i8* [[STR]], i32 [[STRLEN]]
+; CHECK-NEXT:    store i8* [[STRCHR]], i8** @chp, align 4
+; CHECK-NEXT:    ret void
+;
 
   %dst = call i8* @strchr(i8* %str, i32 0)
   store i8* %dst, i8** @chp
   ret void
 }
 
 ; Check transformation strchr("\r\n", C) != nullptr -> (C & 9217) != 0
 define i1 @test_simplify7(i32 %C) {
-; CHECK-LABEL: @test_simplify7
-; CHECK-NEXT: [[TRUNC:%.*]] = trunc i32 %C to i16
-; CHECK-NEXT: [[TRUNC_AND:%.*]] = and i16 [[TRUNC]], 255
-; CHECK-NEXT: %memchr.bounds = icmp ult i16 [[TRUNC_AND]], 16
-; CHECK-NEXT: [[SHL:%.*]] = shl i16 1, [[TRUNC_AND]]
-; CHECK-NEXT: [[AND:%.*]] = and i16 [[SHL]], 9217
-; CHECK-NEXT: %memchr.bits = icmp ne i16 [[AND]], 0
-; CHECK-NEXT: %memchr1 = and i1 %memchr.bounds, %memchr.bits
-; CHECK-NEXT: ret i1 %memchr1
+; CHECK-LABEL: @test_simplify7(
+; CHECK-NEXT:    [[TMP1:%.*]] = trunc i32 [[C:%.*]] to i16
+; CHECK-NEXT:    [[TMP2:%.*]] = and i16 [[TMP1]], 255
+; CHECK-NEXT:    [[MEMCHR_BOUNDS:%.*]] = icmp ult i16 [[TMP2]], 16
+; CHECK-NEXT:    [[TMP3:%.*]] = shl i16 1, [[TMP2]]
+; CHECK-NEXT:    [[TMP4:%.*]] = and i16 [[TMP3]], 9217
+; CHECK-NEXT:    [[MEMCHR_BITS:%.*]] = icmp ne i16 [[TMP4]], 0
+; CHECK-NEXT:    [[MEMCHR1:%.*]] = and i1 [[MEMCHR_BOUNDS]], [[MEMCHR_BITS]]
+; CHECK-NEXT:    ret i1 [[MEMCHR1]]
+;
 
   %dst = call i8* @strchr(i8* getelementptr inbounds ([3 x i8], [3 x i8]* @newlines, i64 0, i64 0), i32 %C)
   %cmp = icmp ne i8* %dst, null
   ret i1 %cmp
 }
+
+define i8* @test1(i8* %str, i32 %c) {
+; CHECK-LABEL: @test1(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @strchr(i8* nonnull dereferenceable(1) [[STR:%.*]], i32 [[C:%.*]])
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+
+  %ret = call i8* @strchr(i8* %str, i32 %c)
+  ret i8* %ret
+}
+
+define i8* @test2(i8* %str, i32 %c) "null-pointer-is-valid"="true" {
+; CHECK-LABEL: @test2(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @strchr(i8* [[STR:%.*]], i32 [[C:%.*]])
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+
+  %ret = call i8* @strchr(i8* %str, i32 %c)
+  ret i8* %ret
+}

llvm/trunk/test/Transforms/InstCombine/strcmp-1.ll

 ; strcmp(x, y)   -> memcmp(x, y, <known length>)
 ; (This transform is rather difficult to trigger in a useful manner)
 define i32 @test5(i1 %b) {
 ; CHECK-LABEL: @test5(
 ; CHECK: %memcmp = call i32 @memcmp(i8* getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i8* %str2, i32 5)
 ; CHECK: ret i32 %memcmp
 ; NOBCMP-LABEL: @test5(
 ; NOBCMP-NEXT:    [[STR2:%.*]] = select i1 [[B:%.*]], i8* getelementptr inbounds ([5 x i8], [5 x i8]* @hell, i32 0, i32 0), i8* getelementptr inbounds ([5 x i8], [5 x i8]* @bell, i32 0, i32 0)
-; NOBCMP-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* dereferenceable(5) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i8* dereferenceable(5) [[STR2]], i32 5)
+; NOBCMP-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(5) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i8* nonnull dereferenceable(5) [[STR2]], i32 5)
 ; NOBCMP-NEXT:    ret i32 [[MEMCMP]]
 ;
 ; BCMP-LABEL: @test5(
 ; BCMP-NEXT:    [[STR2:%.*]] = select i1 [[B:%.*]], i8* getelementptr inbounds ([5 x i8], [5 x i8]* @hell, i32 0, i32 0), i8* getelementptr inbounds ([5 x i8], [5 x i8]* @bell, i32 0, i32 0)
-; BCMP-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* dereferenceable(5) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i8* dereferenceable(5) [[STR2]], i32 5)
+; BCMP-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(5) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i8* nonnull dereferenceable(5) [[STR2]], i32 5)
 ; BCMP-NEXT:    ret i32 [[MEMCMP]]
 ;
   %str1 = getelementptr inbounds [6 x i8], [6 x i8]* @hello, i32 0, i32 0
   %temp1 = getelementptr inbounds [5 x i8], [5 x i8]* @hell, i32 0, i32 0
   %temp2 = getelementptr inbounds [5 x i8], [5 x i8]* @bell, i32 0, i32 0
   %str2 = select i1 %b, i8* %temp1, i8* %temp2
   %temp3 = call i32 @strcmp(i8* %str1, i8* %str2)
   ret i32 %temp3
;
   %temp1 = call i32 @strcmp(i8* %str, i8* %str)
   ret i32 %temp1
 }
 
 ; strcmp(x, y) == 0  -> bcmp(x, y, <known length>)
 define i1 @test7(i1 %b) {
 ; NOBCMP-LABEL: @test7(
 ; NOBCMP-NEXT:    [[STR2:%.*]] = select i1 [[B:%.*]], i8* getelementptr inbounds ([5 x i8], [5 x i8]* @hell, i32 0, i32 0), i8* getelementptr inbounds ([5 x i8], [5 x i8]* @bell, i32 0, i32 0)
-; NOBCMP-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* dereferenceable(5) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i8* dereferenceable(5) [[STR2]], i32 5)
+; NOBCMP-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(5) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i8* nonnull dereferenceable(5) [[STR2]], i32 5)
 ; NOBCMP-NEXT:    [[RES:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; NOBCMP-NEXT:    ret i1 [[RES]]
 ;
 ; BCMP-LABEL: @test7(
 ; BCMP-NEXT:    [[STR2:%.*]] = select i1 [[B:%.*]], i8* getelementptr inbounds ([5 x i8], [5 x i8]* @hell, i32 0, i32 0), i8* getelementptr inbounds ([5 x i8], [5 x i8]* @bell, i32 0, i32 0)
-; BCMP-NEXT:    [[BCMP:%.*]] = call i32 @bcmp(i8* dereferenceable(5) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i8* dereferenceable(5) [[STR2]], i32 5)
+; BCMP-NEXT:    [[BCMP:%.*]] = call i32 @bcmp(i8* nonnull dereferenceable(5) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i8* nonnull dereferenceable(5) [[STR2]], i32 5)
 ; BCMP-NEXT:    [[RES:%.*]] = icmp eq i32 [[BCMP]], 0
 ; BCMP-NEXT:    ret i1 [[RES]]
 ;
 
 
   %str1 = getelementptr inbounds [6 x i8], [6 x i8]* @hello, i32 0, i32 0
   %temp1 = getelementptr inbounds [5 x i8], [5 x i8]* @hell, i32 0, i32 0
   %temp2 = getelementptr inbounds [5 x i8], [5 x i8]* @bell, i32 0, i32 0
   %str2 = select i1 %b, i8* %temp1, i8* %temp2
   %temp3 = call i32 @strcmp(i8* %str1, i8* %str2)
   %res = icmp eq i32 %temp3, 0
   ret i1 %res
 }

llvm/trunk/test/Transforms/InstCombine/strcmp-memcmp.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
 
 @key = constant [4 x i8] c"key\00", align 1
 @abc = constant [8 x i8] c"abc\00de\00\00", align 1
 
 declare void @use(i32)
 
 define i32 @strcmp_memcmp([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strcmp_memcmp(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* nonnull %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 declare i32 @strcmp(i8* nocapture, i8* nocapture)
 
 define i32 @strcmp_memcmp2([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strcmp_memcmp2(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull %string)
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strcmp_memcmp3([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strcmp_memcmp3(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp ne i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* nonnull %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
   %cmp = icmp ne i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strcmp_memcmp4([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strcmp_memcmp4(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp ne i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull %string)
   %cmp = icmp ne i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strcmp_memcmp5([5 x i8]* dereferenceable (5) %buf) {
 ; CHECK-LABEL: @strcmp_memcmp5(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [5 x i8], [5 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [5 x i8], [5 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* nonnull align 1 %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strcmp_memcmp6([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strcmp_memcmp6(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* nonnull %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
   %cmp = icmp sgt i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strcmp_memcmp7([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strcmp_memcmp7(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
 ; CHECK-NEXT:    [[MEMCMP_LOBIT:%.*]] = lshr i32 [[MEMCMP]], 31
 ; CHECK-NEXT:    ret i32 [[MEMCMP_LOBIT]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull %string)
   %cmp = icmp slt i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strcmp_memcmp8([4 x i8]* dereferenceable (4) %buf) {
 ; CHECK-LABEL: @strcmp_memcmp8(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [4 x i8], [4 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [4 x i8], [4 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* nonnull %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strcmp_memcmp9([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strcmp_memcmp9(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* dereferenceable(4) getelementptr inbounds ([8 x i8], [8 x i8]* @abc, i64 0, i64 0), i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([8 x i8], [8 x i8]* @abc, i64 0, i64 0), i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* nonnull %string, i8* getelementptr inbounds ([8 x i8], [8 x i8]* @abc, i64 0, i64 0))
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 
 define i32 @strncmp_memcmp([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(2) [[STRING]], i8* dereferenceable(2) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 2)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(2) [[STRING]], i8* nonnull dereferenceable(2) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 2)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* nonnull %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 2)
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 declare i32 @strncmp(i8* nocapture, i8* nocapture, i64)
 
 define i32 @strncmp_memcmp2([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp2(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp ne i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* nonnull %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 11)
   %cmp = icmp ne i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strncmp_memcmp3([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp3(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull %string, i64 11)
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strncmp_memcmp4([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp4(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* nonnull %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 5)
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strncmp_memcmp5([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp5(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull %string, i64 5)
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 
 define i32 @strncmp_memcmp6([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp6(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp ne i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull %string, i64 5)
   %cmp = icmp ne i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strncmp_memcmp7([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp7(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* nonnull %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 4)
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strncmp_memcmp8([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp8(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(3) [[STRING]], i8* dereferenceable(3) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 3)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(3) [[STRING]], i8* nonnull dereferenceable(3) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 3)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* nonnull %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 3)
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strncmp_memcmp9([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp9(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull %string, i64 5)
   %cmp = icmp sgt i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strncmp_memcmp10([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp10(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
 ; CHECK-NEXT:    [[MEMCMP_LOBIT:%.*]] = lshr i32 [[MEMCMP]], 31
 ; CHECK-NEXT:    ret i32 [[MEMCMP_LOBIT]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull %string, i64 5)
   %cmp = icmp slt i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strncmp_memcmp11([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp11(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull %string, i64 12)
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strncmp_memcmp12([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp12(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(4) [[STRING]], i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull %string, i64 12)
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strncmp_memcmp13([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp13(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(2) [[STRING]], i8* dereferenceable(2) getelementptr inbounds ([8 x i8], [8 x i8]* @abc, i64 0, i64 0), i64 2)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(2) [[STRING]], i8* nonnull dereferenceable(2) getelementptr inbounds ([8 x i8], [8 x i8]* @abc, i64 0, i64 0), i64 2)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* nonnull %string, i8* getelementptr inbounds ([8 x i8], [8 x i8]* @abc, i64 0, i64 0), i64 2)
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strncmp_memcmp14([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp14(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* dereferenceable(4) getelementptr inbounds ([8 x i8], [8 x i8]* @abc, i64 0, i64 0), i64 4)
+; CHECK-NEXT:    [[MEMCMP:%.*]] = call i32 @memcmp(i8* nonnull dereferenceable(4) [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([8 x i8], [8 x i8]* @abc, i64 0, i64 0), i64 4)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[MEMCMP]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* nonnull %string, i8* getelementptr inbounds ([8 x i8], [8 x i8]* @abc, i64 0, i64 0), i64 12)
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 ; Negative tests
 define i32 @strcmp_memcmp_bad([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strcmp_memcmp_bad(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strcmp(i8* nonnull [[STRING]], i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
+; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strcmp(i8* nonnull [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[CALL]], 3
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* nonnull %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
   %cmp = icmp sgt i32 %call, 3
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strcmp_memcmp_bad2([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strcmp_memcmp_bad2(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strcmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull [[STRING]])
+; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strcmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull [[STRING]])
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp slt i32 [[CALL]], 3
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull %string)
   %cmp = icmp slt i32 %call, 3
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strcmp_memcmp_bad3([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strcmp_memcmp_bad3(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strcmp(i8* nonnull [[STRING]], i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
+; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strcmp(i8* nonnull [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
 ; CHECK-NEXT:    ret i32 [[CALL]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* nonnull %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
   ret i32 %call
 }
 
 
 define i32 @strcmp_memcmp_bad4(i8* nocapture readonly %buf) {
 ; CHECK-LABEL: @strcmp_memcmp_bad4(
-; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @strcmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* [[BUF:%.*]])
+; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @strcmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(1) [[BUF:%.*]])
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[CALL]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %call = tail call i32 @strcmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* %buf)
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 
 define i32 @strcmp_memcmp_bad5([3 x i8]* dereferenceable (3) %buf) {
 ; CHECK-LABEL: @strcmp_memcmp_bad5(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [3 x i8], [3 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strcmp(i8* nonnull [[STRING]], i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
+; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strcmp(i8* nonnull [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[CALL]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [3 x i8], [3 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* nonnull %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strcmp_memcmp_bad6([4 x i8]* dereferenceable (4) %buf, i8* nocapture readonly %k) {
 ; CHECK-LABEL: @strcmp_memcmp_bad6(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [4 x i8], [4 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strcmp(i8* nonnull [[STRING]], i8* [[K:%.*]])
+; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strcmp(i8* nonnull [[STRING]], i8* nonnull dereferenceable(1) [[K:%.*]])
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[CALL]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [4 x i8], [4 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* nonnull %string, i8* %k)
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strcmp_memcmp_bad7(i8* nocapture readonly %k) {
 ; CHECK-LABEL: @strcmp_memcmp_bad7(
-; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @strcmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* [[K:%.*]])
+; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @strcmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(1) [[K:%.*]])
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[CALL]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %call = tail call i32 @strcmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* %k)
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strcmp_memcmp_bad8([4 x i8]* dereferenceable (4) %buf) {
 ; CHECK-LABEL: @strcmp_memcmp_bad8(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [4 x i8], [4 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strcmp(i8* nonnull [[STRING]], i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
+; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strcmp(i8* nonnull [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
 ; CHECK-NEXT:    tail call void @use(i32 [[CALL]])
 ; CHECK-NEXT:    ret i32 0
 ;
   %string = getelementptr inbounds [4 x i8], [4 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* nonnull %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
   tail call void @use(i32 %call)
   ret i32 0
 }
 
 define i32 @strncmp_memcmp_bad([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp_bad(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strncmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull [[STRING]], i64 5)
+; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strncmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull [[STRING]], i64 5)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[CALL]], 3
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull %string, i64 5)
   %cmp = icmp sgt i32 %call, 3
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 
 define i32 @strncmp_memcmp_bad1([12 x i8]* dereferenceable (12) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp_bad1(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strncmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull [[STRING]], i64 5)
+; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strncmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull [[STRING]], i64 5)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp slt i32 [[CALL]], 3
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull %string, i64 5)
   %cmp = icmp slt i32 %call, 3
   %conv = zext i1 %cmp to i32
;
   %call = call i32 @strncmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull %string, i64 %n)
   %cmp = icmp slt i32 %call, 1
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strncmp_memcmp_bad3(i8* nocapture readonly %k) {
 ; CHECK-LABEL: @strncmp_memcmp_bad3(
-; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @strncmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* [[K:%.*]], i64 2)
+; CHECK-NEXT:    [[CALL:%.*]] = tail call i32 @strncmp(i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* nonnull dereferenceable(1) [[K:%.*]], i64 2)
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[CALL]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %call = tail call i32 @strncmp(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i8* %k, i64 2)
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 define i32 @strncmp_memcmp_bad4([4 x i8]* dereferenceable (4) %buf) {
 ; CHECK-LABEL: @strncmp_memcmp_bad4(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [4 x i8], [4 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strncmp(i8* nonnull [[STRING]], i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 2)
+; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strncmp(i8* nonnull [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 2)
 ; CHECK-NEXT:    tail call void @use(i32 [[CALL]])
 ; CHECK-NEXT:    ret i32 0
 ;
   %string = getelementptr inbounds [4 x i8], [4 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strncmp(i8* nonnull %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0), i64 2)
   tail call void @use(i32 %call)
   ret i32 0
 }
 
 define i32 @strcmp_memcmp_msan([12 x i8]* dereferenceable (12) %buf) sanitize_memory {
 ; CHECK-LABEL: @strcmp_memcmp_msan(
 ; CHECK-NEXT:    [[STRING:%.*]] = getelementptr inbounds [12 x i8], [12 x i8]* [[BUF:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strcmp(i8* nonnull [[STRING]], i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
+; CHECK-NEXT:    [[CALL:%.*]] = call i32 @strcmp(i8* nonnull [[STRING]], i8* nonnull dereferenceable(4) getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[CALL]], 0
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i1 [[CMP]] to i32
 ; CHECK-NEXT:    ret i32 [[CONV]]
 ;
   %string = getelementptr inbounds [12 x i8], [12 x i8]* %buf, i64 0, i64 0
   %call = call i32 @strcmp(i8* nonnull %string, i8* getelementptr inbounds ([4 x i8], [4 x i8]* @key, i64 0, i64 0))
   %cmp = icmp eq i32 %call, 0
   %conv = zext i1 %cmp to i32
   ret i32 %conv
 }
 
 declare i32 @memcmp(i8* nocapture, i8* nocapture, i64)

llvm/trunk/test/Transforms/InstCombine/strcpy-1.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the strcpy library call simplifier works correctly.
 ; rdar://6839935
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 ;
 ; This transformation requires the pointer size, as it assumes that size_t is
 ; the size of a pointer.
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128-n8:16:32"
 
 @hello = constant [6 x i8] c"hello\00"
 @a = common global [32 x i8] zeroinitializer, align 1
 @b = common global [32 x i8] zeroinitializer, align 1
 
 declare i8* @strcpy(i8*, i8*)
 
 define void @test_simplify1() {
 ; CHECK-LABEL: @test_simplify1(
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(6) getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0), i8* nonnull align 1 dereferenceable(6) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 6, i1 false)
+; CHECK-NEXT:    ret void
+;
 
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
 
   call i8* @strcpy(i8* %dst, i8* %src)
-; CHECK: @llvm.memcpy.p0i8.p0i8.i32
   ret void
 }
 
 define i8* @test_simplify2() {
 ; CHECK-LABEL: @test_simplify2(
+; CHECK-NEXT:    ret i8* getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0)
+;
 
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
 
   %ret = call i8* @strcpy(i8* %dst, i8* %dst)
-; CHECK: ret i8* getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0)
   ret i8* %ret
 }
 
+
+define void @test_simplify3(i8* %dst) {
+; CHECK-LABEL: @test_simplify3(
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(80) [[DST:%.*]], i8* nonnull align 1 dereferenceable(6) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 6, i1 false)
+; CHECK-NEXT:    ret void
+;
+
+  %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
+
+  call i8* @strcpy(i8* dereferenceable(80) %dst, i8* %src)
+  ret void
+}
+
 define i8* @test_no_simplify1() {
 ; CHECK-LABEL: @test_no_simplify1(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @strcpy(i8* nonnull dereferenceable(1) getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0), i8* nonnull dereferenceable(1) getelementptr inbounds ([32 x i8], [32 x i8]* @b, i32 0, i32 0))
+; CHECK-NEXT:    ret i8* [[RET]]
+;
 
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %src = getelementptr [32 x i8], [32 x i8]* @b, i32 0, i32 0
 
   %ret = call i8* @strcpy(i8* %dst, i8* %src)
-; CHECK: call i8* @strcpy
   ret i8* %ret
 }

llvm/trunk/test/Transforms/InstCombine/strcpy_chk-1.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test lib call simplification of __strcpy_chk calls with various values
 ; for src, dst, and slen.
 ;
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 
 @a = common global [60 x i8] zeroinitializer, align 1
 @b = common global [60 x i8] zeroinitializer, align 1
 @.str = private constant [12 x i8] c"abcdefghijk\00"
 
 ; Check cases where slen >= strlen (src).
 
 define i8* @test_simplify1() {
 ; CHECK-LABEL: @test_simplify1(
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
 ; CHECK-NEXT:    ret i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0)
 ;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [12 x i8], [12 x i8]* @.str, i32 0, i32 0
 
   %ret = call i8* @__strcpy_chk(i8* %dst, i8* %src, i32 60)
   ret i8* %ret
 }
 
 define i8* @test_simplify2() {
 ; CHECK-LABEL: @test_simplify2(
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
 ; CHECK-NEXT:    ret i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0)
 ;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [12 x i8], [12 x i8]* @.str, i32 0, i32 0
 
   %ret = call i8* @__strcpy_chk(i8* %dst, i8* %src, i32 12)
   ret i8* %ret
 }
 
 define i8* @test_simplify3() {
 ; CHECK-LABEL: @test_simplify3(
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
 ; CHECK-NEXT:    ret i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0)
 ;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [12 x i8], [12 x i8]* @.str, i32 0, i32 0
 
   %ret = call i8* @__strcpy_chk(i8* %dst, i8* %src, i32 -1)
   ret i8* %ret
 }
 
 ; Check cases where there are no string constants.
 
 define i8* @test_simplify4() {
 ; CHECK-LABEL: @test_simplify4(
-; CHECK-NEXT:    [[STRCPY:%.*]] = call i8* @strcpy(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i32 0, i32 0))
+; CHECK-NEXT:    [[STRCPY:%.*]] = call i8* @strcpy(i8* nonnull dereferenceable(1) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* nonnull dereferenceable(1) getelementptr inbounds ([60 x i8], [60 x i8]* @b, i32 0, i32 0))
 ; CHECK-NEXT:    ret i8* [[STRCPY]]
 ;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
 
   %ret = call i8* @__strcpy_chk(i8* %dst, i8* %src, i32 -1)
   ret i8* %ret
 }

llvm/trunk/test/Transforms/InstCombine/strcspn-1.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the strcspn library call simplifier works correctly.
 ;
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
 
 @abcba = constant [6 x i8] c"abcba\00"
 @abc = constant [4 x i8] c"abc\00"
 @null = constant [1 x i8] zeroinitializer
 
 declare i64 @strcspn(i8*, i8*)
 
 ; Check strcspn(s, "") -> strlen(s).
 
 define i64 @test_simplify1(i8* %str) {
 ; CHECK-LABEL: @test_simplify1(
+; CHECK-NEXT:    [[STRLEN:%.*]] = call i64 @strlen(i8* nonnull dereferenceable(1) [[STR:%.*]])
+; CHECK-NEXT:    ret i64 [[STRLEN]]
+;
   %pat = getelementptr [1 x i8], [1 x i8]* @null, i32 0, i32 0
 
   %ret = call i64 @strcspn(i8* %str, i8* %pat)
-; CHECK-NEXT: [[VAR:%[a-z]+]] = call i64 @strlen(i8* %str)
   ret i64 %ret
-; CHECK-NEXT: ret i64 [[VAR]]
 }
 
 ; Check strcspn("", s) -> 0.
 
 define i64 @test_simplify2(i8* %pat) {
 ; CHECK-LABEL: @test_simplify2(
+; CHECK-NEXT:    ret i64 0
+;
   %str = getelementptr [1 x i8], [1 x i8]* @null, i32 0, i32 0
 
   %ret = call i64 @strcspn(i8* %str, i8* %pat)
   ret i64 %ret
-; CHECK-NEXT: ret i64 0
 }
 
 ; Check strcspn(s1, s2), where s1 and s2 are constants.
 
 define i64 @test_simplify3() {
 ; CHECK-LABEL: @test_simplify3(
+; CHECK-NEXT:    ret i64 0
+;
   %str = getelementptr [6 x i8], [6 x i8]* @abcba, i32 0, i32 0
   %pat = getelementptr [4 x i8], [4 x i8]* @abc, i32 0, i32 0
 
   %ret = call i64 @strcspn(i8* %str, i8* %pat)
   ret i64 %ret
-; CHECK-NEXT: ret i64 0
 }
 
 ; Check cases that shouldn't be simplified.
 
 define i64 @test_no_simplify1(i8* %str, i8* %pat) {
 ; CHECK-LABEL: @test_no_simplify1(
+; CHECK-NEXT:    [[RET:%.*]] = call i64 @strcspn(i8* [[STR:%.*]], i8* [[PAT:%.*]])
+; CHECK-NEXT:    ret i64 [[RET]]
+;
 
   %ret = call i64 @strcspn(i8* %str, i8* %pat)
-; CHECK-NEXT: %ret = call i64 @strcspn(i8* %str, i8* %pat)
   ret i64 %ret
-; CHECK-NEXT: ret i64 %ret
 }

llvm/trunk/test/Transforms/InstCombine/strlen-1.ll

 ;
   %str_l = call i32 @strlen(i8* %str_p)
   %ne_null = icmp ne i32 %str_l, 0
   ret i1 %ne_null
 }
 
 define i32 @test_simplify9(i1 %x) {
 ; CHECK-LABEL: @test_simplify9(
-; CHECK-NEXT:    [[TMP1:%.*]] = select i1 %x, i32 5, i32 6
+; CHECK-NEXT:    [[TMP1:%.*]] = select i1 [[X:%.*]], i32 5, i32 6
 ; CHECK-NEXT:    ret i32 [[TMP1]]
 ;
   %hello = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
   %longer = getelementptr [7 x i8], [7 x i8]* @longer, i32 0, i32 0
   %s = select i1 %x, i8* %hello, i8* %longer
   %l = call i32 @strlen(i8* %s)
   ret i32 %l
 }
 
 ; Check the case that should be simplified to a sub instruction.
 ; strlen(@hello + x) --> 5 - x
 
 define i32 @test_simplify10(i32 %x) {
 ; CHECK-LABEL: @test_simplify10(
-; CHECK-NEXT:    [[TMP1:%.*]] = sub i32 5, %x
+; CHECK-NEXT:    [[TMP1:%.*]] = sub i32 5, [[X:%.*]]
 ; CHECK-NEXT:    ret i32 [[TMP1]]
 ;
   %hello_p = getelementptr inbounds [6 x i8], [6 x i8]* @hello, i32 0, i32 %x
   %hello_l = call i32 @strlen(i8* %hello_p)
   ret i32 %hello_l
 }
 
 ; strlen(@null_hello_mid + (x & 7)) --> 9 - (x & 7)
 
 define i32 @test_simplify11(i32 %x) {
 ; CHECK-LABEL: @test_simplify11(
-; CHECK-NEXT:    [[AND:%.*]] = and i32 %x, 7
+; CHECK-NEXT:    [[AND:%.*]] = and i32 [[X:%.*]], 7
 ; CHECK-NEXT:    [[TMP1:%.*]] = sub nuw nsw i32 9, [[AND]]
 ; CHECK-NEXT:    ret i32 [[TMP1]]
 ;
   %and = and i32 %x, 7
   %hello_p = getelementptr inbounds [13 x i8], [13 x i8]* @null_hello_mid, i32 0, i32 %and
   %hello_l = call i32 @strlen(i8* %hello_p)
   ret i32 %hello_l
 }
 
 ; Check cases that shouldn't be simplified.
 
 define i32 @test_no_simplify1() {
 ; CHECK-LABEL: @test_no_simplify1(
-; CHECK-NEXT:    [[A_L:%.*]] = call i32 @strlen(i8* getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0))
+; CHECK-NEXT:    [[A_L:%.*]] = call i32 @strlen(i8* nonnull dereferenceable(1) getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0))
 ; CHECK-NEXT:    ret i32 [[A_L]]
 ;
   %a_p = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %a_l = call i32 @strlen(i8* %a_p)
   ret i32 %a_l
 }
 
 ; strlen(@null_hello + x) should not be simplified to a sub instruction.
 
 define i32 @test_no_simplify2(i32 %x) {
 ; CHECK-LABEL: @test_no_simplify2(
-; CHECK-NEXT:    [[HELLO_P:%.*]] = getelementptr inbounds [7 x i8], [7 x i8]* @null_hello, i32 0, i32 %x
+; CHECK-NEXT:    [[HELLO_P:%.*]] = getelementptr inbounds [7 x i8], [7 x i8]* @null_hello, i32 0, i32 [[X:%.*]]
 ; CHECK-NEXT:    [[HELLO_L:%.*]] = call i32 @strlen(i8* nonnull [[HELLO_P]])
 ; CHECK-NEXT:    ret i32 [[HELLO_L]]
 ;
   %hello_p = getelementptr inbounds [7 x i8], [7 x i8]* @null_hello, i32 0, i32 %x
   %hello_l = call i32 @strlen(i8* %hello_p)
   ret i32 %hello_l
 }
 
 define i32 @test_no_simplify2_no_null_opt(i32 %x) #0 {
 ; CHECK-LABEL: @test_no_simplify2_no_null_opt(
-; CHECK-NEXT:    [[HELLO_P:%.*]] = getelementptr inbounds [7 x i8], [7 x i8]* @null_hello, i32 0, i32 %x
+; CHECK-NEXT:    [[HELLO_P:%.*]] = getelementptr inbounds [7 x i8], [7 x i8]* @null_hello, i32 0, i32 [[X:%.*]]
 ; CHECK-NEXT:    [[HELLO_L:%.*]] = call i32 @strlen(i8* [[HELLO_P]])
 ; CHECK-NEXT:    ret i32 [[HELLO_L]]
 ;
   %hello_p = getelementptr inbounds [7 x i8], [7 x i8]* @null_hello, i32 0, i32 %x
   %hello_l = call i32 @strlen(i8* %hello_p)
   ret i32 %hello_l
 }
 
 ; strlen(@null_hello_mid + (x & 15)) should not be simplified to a sub instruction.
 
 define i32 @test_no_simplify3(i32 %x) {
 ; CHECK-LABEL: @test_no_simplify3(
-; CHECK-NEXT:    [[AND:%.*]] = and i32 %x, 15
+; CHECK-NEXT:    [[AND:%.*]] = and i32 [[X:%.*]], 15
 ; CHECK-NEXT:    [[HELLO_P:%.*]] = getelementptr inbounds [13 x i8], [13 x i8]* @null_hello_mid, i32 0, i32 [[AND]]
 ; CHECK-NEXT:    [[HELLO_L:%.*]] = call i32 @strlen(i8* nonnull [[HELLO_P]])
 ; CHECK-NEXT:    ret i32 [[HELLO_L]]
 ;
   %and = and i32 %x, 15
   %hello_p = getelementptr inbounds [13 x i8], [13 x i8]* @null_hello_mid, i32 0, i32 %and
   %hello_l = call i32 @strlen(i8* %hello_p)
   ret i32 %hello_l
 }
 
 define i32 @test_no_simplify3_on_null_opt(i32 %x) #0 {
 ; CHECK-LABEL: @test_no_simplify3_on_null_opt(
-; CHECK-NEXT:    [[AND:%.*]] = and i32 %x, 15
+; CHECK-NEXT:    [[AND:%.*]] = and i32 [[X:%.*]], 15
 ; CHECK-NEXT:    [[HELLO_P:%.*]] = getelementptr inbounds [13 x i8], [13 x i8]* @null_hello_mid, i32 0, i32 [[AND]]
 ; CHECK-NEXT:    [[HELLO_L:%.*]] = call i32 @strlen(i8* [[HELLO_P]])
 ; CHECK-NEXT:    ret i32 [[HELLO_L]]
 ;
   %and = and i32 %x, 15
   %hello_p = getelementptr inbounds [13 x i8], [13 x i8]* @null_hello_mid, i32 0, i32 %and
   %hello_l = call i32 @strlen(i8* %hello_p)
   ret i32 %hello_l
 }
 
+define i32 @test1(i8* %str) {
+; CHECK-LABEL: @test1(
+; CHECK-NEXT:    [[LEN:%.*]] = tail call i32 @strlen(i8* nonnull dereferenceable(1) [[STR:%.*]]) #1
+; CHECK-NEXT:    ret i32 [[LEN]]
+;
+  %len = tail call i32 @strlen(i8* %str) nounwind
+  ret i32 %len
+}
+
+define i32 @test2(i8* %str) #0 {
+; CHECK-LABEL: @test2(
+; CHECK-NEXT:    [[LEN:%.*]] = tail call i32 @strlen(i8* [[STR:%.*]]) #1
+; CHECK-NEXT:    ret i32 [[LEN]]
+;
+  %len = tail call i32 @strlen(i8* %str) nounwind
+  ret i32 %len
+}
+
 attributes #0 = { "null-pointer-is-valid"="true" }

llvm/trunk/test/Transforms/InstCombine/strncat-2.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the strncat libcall simplifier works correctly.
 ;
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 
 @hello = constant [6 x i8] c"hello\00"
 @empty = constant [1 x i8] c"\00"
 @a = common global [32 x i8] zeroinitializer, align 1
-
 declare i8* @strncat(i8*, i8*, i32)
 
 define void @test_simplify1() {
 ; CHECK-LABEL: @test_simplify1(
-; CHECK-NOT: call i8* @strncat
-; CHECK: ret void
+; CHECK-NEXT:    [[STRLEN:%.*]] = call i32 @strlen(i8* nonnull dereferenceable(1) getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0))
+; CHECK-NEXT:    [[ENDPTR:%.*]] = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 [[STRLEN]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(6) [[ENDPTR]], i8* nonnull align 1 dereferenceable(6) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 6, i1 false)
+; CHECK-NEXT:    ret void
+;
 
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
   call i8* @strncat(i8* %dst, i8* %src, i32 13)
   ret void
 }
 
 define void @test_simplify2() {
 ; CHECK-LABEL: @test_simplify2(
-; CHECK-NEXT: ret void
+; CHECK-NEXT:    ret void
+;
 
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %src = getelementptr [1 x i8], [1 x i8]* @empty, i32 0, i32 0
   call i8* @strncat(i8* %dst, i8* %src, i32 13)
   ret void
 }
 
 define void @test_simplify3() {
 ; CHECK-LABEL: @test_simplify3(
-; CHECK-NEXT: ret void
+; CHECK-NEXT:    ret void
+;
 
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
   call i8* @strncat(i8* %dst, i8* %src, i32 0)
   ret void
 }
 
 define void @test_nosimplify1() {
 ; CHECK-LABEL: @test_nosimplify1(
-; CHECK: call i8* @strncat
-; CHECK: ret void
+; CHECK-NEXT:    [[TMP1:%.*]] = call i8* @strncat(i8* nonnull dereferenceable(1) getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0), i8* nonnull dereferenceable(6) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 1)
+; CHECK-NEXT:    ret void
+;
 
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
   call i8* @strncat(i8* %dst, i8* %src, i32 1)
   ret void
 }
+
+; strncat(nonnull x, nonnull y, n)  -> strncat(nonnull x, y, n)
+define i8* @test1(i8* %str1, i8* %str2, i32 %n) {
+; CHECK-LABEL: @test1(
+; CHECK-NEXT:    [[TEMP1:%.*]] = call i8* @strncat(i8* nonnull [[STR1:%.*]], i8* nonnull [[STR2:%.*]], i32 [[N:%.*]])
+; CHECK-NEXT:    ret i8* [[TEMP1]]
+;
+
+  %temp1 = call i8* @strncat(i8* nonnull %str1, i8* nonnull %str2, i32 %n)
+  ret i8* %temp1
+}
+
+; strncat(x, y, 0)  -> x
+define i8* @test2(i8* %str1, i8* %str2, i32 %n) {
+; CHECK-LABEL: @test2(
+; CHECK-NEXT:    ret i8* [[STR1:%.*]]
+;
+
+  %temp1 = call i8* @strncat(i8* %str1, i8* %str2, i32 0)
+  ret i8* %temp1
+}
+
+; strncat(x, y, 5)  -> strncat(nonnull x, nonnull y, 5)
+define i8* @test3(i8* %str1, i8* %str2, i32 %n) {
+; CHECK-LABEL: @test3(
+; CHECK-NEXT:    [[TEMP1:%.*]] = call i8* @strncat(i8* nonnull dereferenceable(1) [[STR1:%.*]], i8* nonnull dereferenceable(1) [[STR2:%.*]], i32 5)
+; CHECK-NEXT:    ret i8* [[TEMP1]]
+;
+
+  %temp1 = call i8* @strncat(i8* %str1, i8* %str2, i32 5)
+  ret i8* %temp1
+}
+
+define i8* @test4(i8* %str1, i8* %str2, i32 %n) "null-pointer-is-valid"="true" {
+; CHECK-LABEL: @test4(
+; CHECK-NEXT:    [[TEMP1:%.*]] = call i8* @strncat(i8* [[STR1:%.*]], i8* [[STR2:%.*]], i32 [[N:%.*]])
+; CHECK-NEXT:    ret i8* [[TEMP1]]
+;
+
+  %temp1 = call i8* @strncat(i8* %str1, i8* %str2, i32 %n)
+  ret i8* %temp1
+}
+
+define i8* @test5(i8* %str, i32 %n) {
+; CHECK-LABEL: @test5(
+; CHECK-NEXT:    [[STRLEN:%.*]] = call i32 @strlen(i8* nonnull dereferenceable(1) [[STR:%.*]])
+; CHECK-NEXT:    [[ENDPTR:%.*]] = getelementptr i8, i8* [[STR]], i32 [[STRLEN]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(6) [[ENDPTR]], i8* nonnull align 1 dereferenceable(6) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 6, i1 false)
+; CHECK-NEXT:    ret i8* [[STR]]
+;
+  %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
+  %temp1 = call i8* @strncat(i8* %str, i8* %src, i32 10)
+  ret i8* %temp1
+}

llvm/trunk/test/Transforms/InstCombine/strncat-3.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the strncat libcall simplifier works correctly.
 ;
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 
 @hello = constant [6 x i8] c"hello\00"
 @empty = constant [1 x i8] c"\00"
 @a = common global [32 x i8] zeroinitializer, align 1
 
 declare i16* @strncat(i8*, i8*, i32)
 
 define void @test_nosimplify1() {
 ; CHECK-LABEL: @test_nosimplify1(
-; CHECK: call i16* @strncat
-; CHECK: ret void
+; CHECK-NEXT:    [[TMP1:%.*]] = call i16* @strncat(i8* getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0), i8* getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 13)
+; CHECK-NEXT:    ret void
+;
 
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
   call i16* @strncat(i8* %dst, i8* %src, i32 13)
   ret void
 }

llvm/trunk/test/Transforms/InstCombine/strncmp-1.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the strncmp library call simplifier works correctly.
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 
 @hello = constant [6 x i8] c"hello\00"
 @hell = constant [5 x i8] c"hell\00"
 @bell = constant [5 x i8] c"bell\00"
 @null = constant [1 x i8] zeroinitializer
 
 declare i32 @strncmp(i8*, i8*, i32)
 
 ; strncmp("", x, n) -> -*x
 define i32 @test1(i8* %str2) {
 ; CHECK-LABEL: @test1(
-; CHECK: %strcmpload = load i8, i8* %str
-; CHECK: %1 = zext i8 %strcmpload to i32
-; CHECK: %2 = sub nsw i32 0, %1
-; CHECK: ret i32 %2
+; CHECK-NEXT:    [[STRCMPLOAD:%.*]] = load i8, i8* [[STR2:%.*]], align 1
+; CHECK-NEXT:    [[TMP1:%.*]] = zext i8 [[STRCMPLOAD]] to i32
+; CHECK-NEXT:    [[TMP2:%.*]] = sub nsw i32 0, [[TMP1]]
+; CHECK-NEXT:    ret i32 [[TMP2]]
+;
 
   %str1 = getelementptr inbounds [1 x i8], [1 x i8]* @null, i32 0, i32 0
   %temp1 = call i32 @strncmp(i8* %str1, i8* %str2, i32 10)
   ret i32 %temp1
 }
 
 ; strncmp(x, "", n) -> *x
 define i32 @test2(i8* %str1) {
 ; CHECK-LABEL: @test2(
-; CHECK: %strcmpload = load i8, i8* %str1
-; CHECK: %1 = zext i8 %strcmpload to i32
-; CHECK: ret i32 %1
+; CHECK-NEXT:    [[STRCMPLOAD:%.*]] = load i8, i8* [[STR1:%.*]], align 1
+; CHECK-NEXT:    [[TMP1:%.*]] = zext i8 [[STRCMPLOAD]] to i32
+; CHECK-NEXT:    ret i32 [[TMP1]]
+;
 
   %str2 = getelementptr inbounds [1 x i8], [1 x i8]* @null, i32 0, i32 0
   %temp1 = call i32 @strncmp(i8* %str1, i8* %str2, i32 10)
   ret i32 %temp1
 }
 
 ; strncmp(x, y, n)  -> cnst
 define i32 @test3() {
 ; CHECK-LABEL: @test3(
-; CHECK: ret i32 -1
+; CHECK-NEXT:    ret i32 -1
+;
 
   %str1 = getelementptr inbounds [5 x i8], [5 x i8]* @hell, i32 0, i32 0
   %str2 = getelementptr inbounds [6 x i8], [6 x i8]* @hello, i32 0, i32 0
   %temp1 = call i32 @strncmp(i8* %str1, i8* %str2, i32 10)
   ret i32 %temp1
 }
 
 define i32 @test4() {
 ; CHECK-LABEL: @test4(
-; CHECK: ret i32 1
+; CHECK-NEXT:    ret i32 1
+;
 
   %str1 = getelementptr inbounds [5 x i8], [5 x i8]* @hell, i32 0, i32 0
   %str2 = getelementptr inbounds [1 x i8], [1 x i8]* @null, i32 0, i32 0
   %temp1 = call i32 @strncmp(i8* %str1, i8* %str2, i32 10)
   ret i32 %temp1
 }
 
 define i32 @test5() {
 ; CHECK-LABEL: @test5(
-; CHECK: ret i32 0
+; CHECK-NEXT:    ret i32 0
+;
 
   %str1 = getelementptr inbounds [5 x i8], [5 x i8]* @hell, i32 0, i32 0
   %str2 = getelementptr inbounds [6 x i8], [6 x i8]* @hello, i32 0, i32 0
   %temp1 = call i32 @strncmp(i8* %str1, i8* %str2, i32 4)
   ret i32 %temp1
 }
 
 ; strncmp(x,y,1) -> memcmp(x,y,1)
 define i32 @test6(i8* %str1, i8* %str2) {
 ; CHECK-LABEL: @test6(
-; CHECK: [[LOAD1:%[a-z]+]] = load i8, i8* %str1, align 1
-; CHECK: [[ZEXT1:%[a-z]+]] = zext i8 [[LOAD1]] to i32
-; CHECK: [[LOAD2:%[a-z]+]] = load i8, i8* %str2, align 1
-; CHECK: [[ZEXT2:%[a-z]+]] = zext i8 [[LOAD2]] to i32
-; CHECK: [[RET:%[a-z]+]] = sub nsw i32 [[ZEXT1]], [[ZEXT2]]
-; CHECK: ret i32 [[RET]]
+; CHECK-NEXT:    [[LHSC:%.*]] = load i8, i8* [[STR1:%.*]], align 1
+; CHECK-NEXT:    [[LHSV:%.*]] = zext i8 [[LHSC]] to i32
+; CHECK-NEXT:    [[RHSC:%.*]] = load i8, i8* [[STR2:%.*]], align 1
+; CHECK-NEXT:    [[RHSV:%.*]] = zext i8 [[RHSC]] to i32
+; CHECK-NEXT:    [[CHARDIFF:%.*]] = sub nsw i32 [[LHSV]], [[RHSV]]
+; CHECK-NEXT:    ret i32 [[CHARDIFF]]
+;
 
   %temp1 = call i32 @strncmp(i8* %str1, i8* %str2, i32 1)
   ret i32 %temp1
 }
 
 ; strncmp(x,y,0)   -> 0
 define i32 @test7(i8* %str1, i8* %str2) {
 ; CHECK-LABEL: @test7(
-; CHECK: ret i32 0
+; CHECK-NEXT:    ret i32 0
+;
 
   %temp1 = call i32 @strncmp(i8* %str1, i8* %str2, i32 0)
   ret i32 %temp1
 }
 
 ; strncmp(x,x,n)  -> 0
 define i32 @test8(i8* %str, i32 %n) {
 ; CHECK-LABEL: @test8(
-; CHECK: ret i32 0
+; CHECK-NEXT:    ret i32 0
+;
 
   %temp1 = call i32 @strncmp(i8* %str, i8* %str, i32 %n)
   ret i32 %temp1
 }
+
+; strncmp(nonnull x, nonnull y, n)  -> strncmp(x, y, n)
+define i32 @test9(i8* %str1, i8* %str2, i32 %n) {
+; CHECK-LABEL: @test9(
+; CHECK-NEXT:    [[TEMP1:%.*]] = call i32 @strncmp(i8* nonnull [[STR1:%.*]], i8* nonnull [[STR2:%.*]], i32 [[N:%.*]])
+; CHECK-NEXT:    ret i32 [[TEMP1]]
+;
+
+  %temp1 = call i32 @strncmp(i8* nonnull %str1, i8* nonnull %str2, i32 %n)
+  ret i32 %temp1
+}
+
+; strncmp(nonnull x, nonnull y, 0)  -> 0
+define i32 @test10(i8* %str1, i8* %str2, i32 %n) {
+; CHECK-LABEL: @test10(
+; CHECK-NEXT:    ret i32 0
+;
+
+  %temp1 = call i32 @strncmp(i8* nonnull %str1, i8* nonnull %str2, i32 0)
+  ret i32 %temp1
+}
+
+; strncmp(x, y, 5)  -> strncmp(nonnull x, nonnull y, 5)
+define i32 @test11(i8* %str1, i8* %str2, i32 %n) {
+; CHECK-LABEL: @test11(
+; CHECK-NEXT:    [[TEMP1:%.*]] = call i32 @strncmp(i8* nonnull dereferenceable(1) [[STR1:%.*]], i8* nonnull dereferenceable(1) [[STR2:%.*]], i32 5)
+; CHECK-NEXT:    ret i32 [[TEMP1]]
+;
+
+  %temp1 = call i32 @strncmp(i8* %str1, i8* %str2, i32 5)
+  ret i32 %temp1
+}
+
+define i32 @test12(i8* %str1, i8* %str2, i32 %n) "null-pointer-is-valid"="true" {
+; CHECK-LABEL: @test12(
+; CHECK-NEXT:    [[TEMP1:%.*]] = call i32 @strncmp(i8* [[STR1:%.*]], i8* [[STR2:%.*]], i32 [[N:%.*]])
+; CHECK-NEXT:    ret i32 [[TEMP1]]
+;
+
+  %temp1 = call i32 @strncmp(i8* %str1, i8* %str2, i32 %n)
+  ret i32 %temp1
+}

llvm/trunk/test/Transforms/InstCombine/strncmp-2.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the strncmp library call simplifier works correctly.
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 
 @hello = constant [6 x i8] c"hello\00"
 @hell = constant [5 x i8] c"hell\00"
 
 declare i16 @strncmp(i8*, i8*, i32)
 
 define i16 @test_nosimplify() {
 ; CHECK-LABEL: @test_nosimplify(
-; CHECK: call i16 @strncmp
-; CHECK: ret i16 %temp1
+; CHECK-NEXT:    [[TEMP1:%.*]] = call i16 @strncmp(i8* getelementptr inbounds ([5 x i8], [5 x i8]* @hell, i32 0, i32 0), i8* getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 10)
+; CHECK-NEXT:    ret i16 [[TEMP1]]
+;
 
   %str1 = getelementptr inbounds [5 x i8], [5 x i8]* @hell, i32 0, i32 0
   %str2 = getelementptr inbounds [6 x i8], [6 x i8]* @hello, i32 0, i32 0
   %temp1 = call i16 @strncmp(i8* %str1, i8* %str2, i32 10)
   ret i16 %temp1
 }

llvm/trunk/test/Transforms/InstCombine/strncpy-1.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the strncpy library call simplifier works correctly.
 ;
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 
 @hello = constant [6 x i8] c"hello\00"
 @null = constant [1 x i8] zeroinitializer
 @null_hello = constant [7 x i8] c"\00hello\00"
 @a = common global [32 x i8] zeroinitializer, align 1
 @b = common global [32 x i8] zeroinitializer, align 1
 
 declare i8* @strncpy(i8*, i8*, i32)
 declare i32 @puts(i8*)
 
 ; Check a bunch of strncpy invocations together.
 
 define i32 @test_simplify1() {
 ; CHECK-LABEL: @test_simplify1(
-; CHECK-NOT: call i8* @strncpy
-; CHECK: call i32 @puts
+; CHECK-NEXT:    [[TARGET:%.*]] = alloca [1024 x i8], align 1
+; CHECK-NEXT:    [[ARG1:%.*]] = getelementptr inbounds [1024 x i8], [1024 x i8]* [[TARGET]], i32 0, i32 0
+; CHECK-NEXT:    store i8 0, i8* [[ARG1]], align 1
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(6) [[ARG1]], i8* nonnull align 1 dereferenceable(6) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 6, i1 false)
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* nonnull align 1 dereferenceable(42) [[ARG1]], i8 0, i32 42, i1 false)
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* nonnull align 1 dereferenceable(42) [[ARG1]], i8 0, i32 42, i1 false)
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @puts(i8* nonnull [[ARG1]])
+; CHECK-NEXT:    ret i32 0
+;
   %target = alloca [1024 x i8]
   %arg1 = getelementptr [1024 x i8], [1024 x i8]* %target, i32 0, i32 0
   store i8 0, i8* %arg1
 
   %arg2 = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
   %rslt1 = call i8* @strncpy(i8* %arg1, i8* %arg2, i32 6)
 
   %arg3 = getelementptr [1 x i8], [1 x i8]* @null, i32 0, i32 0
   %rslt2 = call i8* @strncpy(i8* %rslt1, i8* %arg3, i32 42)
 
   %arg4 = getelementptr [7 x i8], [7 x i8]* @null_hello, i32 0, i32 0
   %rslt3 = call i8* @strncpy(i8* %rslt2, i8* %arg4, i32 42)
 
   call i32 @puts( i8* %rslt3 )
   ret i32 0
 }
 
 ; Check strncpy(x, "", y) -> memset(x, '\0', y, 1).
 
 define void @test_simplify2() {
 ; CHECK-LABEL: @test_simplify2(
+; CHECK-NEXT:    call void @llvm.memset.p0i8.i32(i8* nonnull align 1 dereferenceable(32) getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0), i8 0, i32 32, i1 false)
+; CHECK-NEXT:    ret void
+;
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %src = getelementptr [1 x i8], [1 x i8]* @null, i32 0, i32 0
 
   call i8* @strncpy(i8* %dst, i8* %src, i32 32)
-; CHECK: call void @llvm.memset.p0i8.i32
   ret void
 }
 
 ; Check strncpy(x, y, 0) -> x.
 
 define i8* @test_simplify3() {
 ; CHECK-LABEL: @test_simplify3(
+; CHECK-NEXT:    ret i8* getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0)
+;
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
 
   %ret = call i8* @strncpy(i8* %dst, i8* %src, i32 0)
   ret i8* %ret
-; CHECK: ret i8* getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0)
 }
 
 ; Check  strncpy(x, s, c) -> memcpy(x, s, c, 1) [s and c are constant].
 
 define void @test_simplify4() {
 ; CHECK-LABEL: @test_simplify4(
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(6) getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0), i8* nonnull align 1 dereferenceable(6) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 6, i1 false)
+; CHECK-NEXT:    ret void
+;
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
 
   call i8* @strncpy(i8* %dst, i8* %src, i32 6)
-; CHECK: call void @llvm.memcpy.p0i8.p0i8.i32
   ret void
 }
 
+define void @test_simplify5(i8* %dst) {
+; CHECK-LABEL: @test_simplify5(
+; CHECK-NEXT:    [[TMP1:%.*]] = call i8* @strncpy(i8* nonnull dereferenceable(8) [[DST:%.*]], i8* nonnull dereferenceable(6) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 32)
+; CHECK-NEXT:    ret void
+;
+  %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
+  call i8* @strncpy(i8* dereferenceable(8) %dst, i8* %src, i32 32)
+  ret void
+}
+
+define void @test_simplify6(i8* %dst) {
+; CHECK-LABEL: @test_simplify6(
+; CHECK-NEXT:    [[TMP1:%.*]] = call i8* @strncpy(i8* nonnull dereferenceable(80) [[DST:%.*]], i8* nonnull dereferenceable(6) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 32)
+; CHECK-NEXT:    ret void
+;
+  %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
+  call i8* @strncpy(i8* dereferenceable(80) %dst, i8* %src, i32 32)
+  ret void
+}
+
+define void @test_simplify7(i8* %dst, i32 %n) {
+; CHECK-LABEL: @test_simplify7(
+; CHECK-NEXT:    [[TMP1:%.*]] = call i8* @strncpy(i8* nonnull dereferenceable(80) [[DST:%.*]], i8* getelementptr inbounds ([1 x i8], [1 x i8]* @null, i32 0, i32 0), i32 [[N:%.*]])
+; CHECK-NEXT:    ret void
+;
+  %src = getelementptr [1 x i8], [1 x i8]* @null, i32 0, i32 0
+  call i8* @strncpy(i8* dereferenceable(80) %dst, i8* %src, i32 %n)
+  ret void
+}
+
+define i8* @test1(i8* %dst, i8* %src, i32 %n) {
+; CHECK-LABEL: @test1(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @strncpy(i8* nonnull [[DST:%.*]], i8* nonnull [[SRC:%.*]], i32 [[N:%.*]])
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+  %ret = call i8* @strncpy(i8* nonnull %dst, i8* nonnull %src, i32 %n)
+  ret i8* %ret
+}
+
+define i8* @test2(i8* %dst) {
+; CHECK-LABEL: @test2(
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(5) [[DST:%.*]], i8* nonnull align 1 dereferenceable(6) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 5, i1 false)
+; CHECK-NEXT:    ret i8* [[DST]]
+;
+  %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
+  %ret = call i8* @strncpy(i8* nonnull %dst, i8* nonnull %src, i32 5)
+  ret i8* %ret
+}
+
 ; Check cases that shouldn't be simplified.
 
 define void @test_no_simplify1() {
 ; CHECK-LABEL: @test_no_simplify1(
+; CHECK-NEXT:    [[TMP1:%.*]] = call i8* @strncpy(i8* nonnull dereferenceable(1) getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0), i8* nonnull dereferenceable(1) getelementptr inbounds ([32 x i8], [32 x i8]* @b, i32 0, i32 0), i32 32)
+; CHECK-NEXT:    ret void
+;
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %src = getelementptr [32 x i8], [32 x i8]* @b, i32 0, i32 0
 
   call i8* @strncpy(i8* %dst, i8* %src, i32 32)
-; CHECK: call i8* @strncpy
   ret void
 }
 
 define void @test_no_simplify2() {
 ; CHECK-LABEL: @test_no_simplify2(
+; CHECK-NEXT:    [[TMP1:%.*]] = call i8* @strncpy(i8* nonnull dereferenceable(1) getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0), i8* nonnull dereferenceable(6) getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 8)
+; CHECK-NEXT:    ret void
+;
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
 
   call i8* @strncpy(i8* %dst, i8* %src, i32 8)
-; CHECK: call i8* @strncpy
   ret void
 }

llvm/trunk/test/Transforms/InstCombine/strncpy-2.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the strncpy library call simplifier works correctly.
 ;
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 
 @hello = constant [6 x i8] c"hello\00"
 @a = common global [32 x i8] zeroinitializer, align 1
 
 declare i16* @strncpy(i8*, i8*, i32)
 
 ; Check that 'strncpy' functions with the wrong prototype aren't simplified.
 
 define void @test_no_simplify1() {
 ; CHECK-LABEL: @test_no_simplify1(
+; CHECK-NEXT:    [[TMP1:%.*]] = call i16* @strncpy(i8* getelementptr inbounds ([32 x i8], [32 x i8]* @a, i32 0, i32 0), i8* getelementptr inbounds ([6 x i8], [6 x i8]* @hello, i32 0, i32 0), i32 6)
+; CHECK-NEXT:    ret void
+;
   %dst = getelementptr [32 x i8], [32 x i8]* @a, i32 0, i32 0
   %src = getelementptr [6 x i8], [6 x i8]* @hello, i32 0, i32 0
 
   call i16* @strncpy(i8* %dst, i8* %src, i32 6)
-; CHECK: call i16* @strncpy
   ret void
 }

llvm/trunk/test/Transforms/InstCombine/strncpy_chk-1.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test lib call simplification of __strncpy_chk calls with various values
 ; for len and dstlen.
 ;
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 
 @a = common global [60 x i8] zeroinitializer, align 1
 @b = common global [60 x i8] zeroinitializer, align 1
 @.str = private constant [12 x i8] c"abcdefghijk\00"
 
 ; Check cases where dstlen >= len
 
 define i8* @test_simplify1() {
 ; CHECK-LABEL: @test_simplify1(
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
 ; CHECK-NEXT:    ret i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0)
 ;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [12 x i8], [12 x i8]* @.str, i32 0, i32 0
 
   %ret = call i8* @__strncpy_chk(i8* %dst, i8* %src, i32 12, i32 60)
   ret i8* %ret
 }
 
 define i8* @test_simplify2() {
 ; CHECK-LABEL: @test_simplify2(
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* nonnull align 1 dereferenceable(12) getelementptr inbounds ([12 x i8], [12 x i8]* @.str, i32 0, i32 0), i32 12, i1 false)
 ; CHECK-NEXT:    ret i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0)
 ;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [12 x i8], [12 x i8]* @.str, i32 0, i32 0
 
   %ret = call i8* @__strncpy_chk(i8* %dst, i8* %src, i32 12, i32 12)
   ret i8* %ret
 }
 
 define i8* @test_simplify3() {
 ; CHECK-LABEL: @test_simplify3(
-; CHECK-NEXT:    [[STRNCPY:%.*]] = call i8* @strncpy(i8* getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* getelementptr inbounds ([60 x i8], [60 x i8]* @b, i32 0, i32 0), i32 12)
+; CHECK-NEXT:    [[STRNCPY:%.*]] = call i8* @strncpy(i8* nonnull dereferenceable(1) getelementptr inbounds ([60 x i8], [60 x i8]* @a, i32 0, i32 0), i8* nonnull dereferenceable(1) getelementptr inbounds ([60 x i8], [60 x i8]* @b, i32 0, i32 0), i32 12)
 ; CHECK-NEXT:    ret i8* [[STRNCPY]]
 ;
   %dst = getelementptr inbounds [60 x i8], [60 x i8]* @a, i32 0, i32 0
   %src = getelementptr inbounds [60 x i8], [60 x i8]* @b, i32 0, i32 0
 
   %ret = call i8* @__strncpy_chk(i8* %dst, i8* %src, i32 12, i32 60)
   ret i8* %ret
 }

llvm/trunk/test/Transforms/InstCombine/strpbrk-1.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the strpbrk library call simplifier works correctly.
 ;
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 
 @hello = constant [12 x i8] c"hello world\00"
 @w = constant [2 x i8] c"w\00"
 @null = constant [1 x i8] zeroinitializer
 
 declare i8* @strpbrk(i8*, i8*)
 
 ; Check strpbrk(s, "") -> NULL.
 
 define i8* @test_simplify1(i8* %str) {
 ; CHECK-LABEL: @test_simplify1(
+; CHECK-NEXT:    ret i8* null
+;
   %pat = getelementptr [1 x i8], [1 x i8]* @null, i32 0, i32 0
 
   %ret = call i8* @strpbrk(i8* %str, i8* %pat)
   ret i8* %ret
-; CHECK-NEXT: ret i8* null
 }
 
 ; Check strpbrk("", s) -> NULL.
 
 define i8* @test_simplify2(i8* %pat) {
 ; CHECK-LABEL: @test_simplify2(
+; CHECK-NEXT:    ret i8* null
+;
   %str = getelementptr [1 x i8], [1 x i8]* @null, i32 0, i32 0
 
   %ret = call i8* @strpbrk(i8* %str, i8* %pat)
   ret i8* %ret
-; CHECK-NEXT: ret i8* null
 }
 
 ; Check strpbrk(s1, s2), where s1 and s2 are constants.
 
 define i8* @test_simplify3() {
 ; CHECK-LABEL: @test_simplify3(
+; CHECK-NEXT:    ret i8* getelementptr inbounds ([12 x i8], [12 x i8]* @hello, i32 0, i32 6)
+;
   %str = getelementptr [12 x i8], [12 x i8]* @hello, i32 0, i32 0
   %pat = getelementptr [2 x i8], [2 x i8]* @w, i32 0, i32 0
 
   %ret = call i8* @strpbrk(i8* %str, i8* %pat)
   ret i8* %ret
-; CHECK-NEXT: ret i8* getelementptr inbounds ([12 x i8], [12 x i8]* @hello, i32 0, i32 6)
 }
 
 ; Check strpbrk(s, "a") -> strchr(s, 'a').
 
 define i8* @test_simplify4(i8* %str) {
 ; CHECK-LABEL: @test_simplify4(
+; CHECK-NEXT:    [[STRCHR:%.*]] = call i8* @strchr(i8* nonnull dereferenceable(1) [[STR:%.*]], i32 119)
+; CHECK-NEXT:    ret i8* [[STRCHR]]
+;
   %pat = getelementptr [2 x i8], [2 x i8]* @w, i32 0, i32 0
 
   %ret = call i8* @strpbrk(i8* %str, i8* %pat)
-; CHECK-NEXT: [[VAR:%[a-z]+]] = call i8* @strchr(i8* %str, i32 119)
   ret i8* %ret
-; CHECK-NEXT: ret i8* [[VAR]]
 }
 
 ; Check cases that shouldn't be simplified.
 
 define i8* @test_no_simplify1(i8* %str, i8* %pat) {
 ; CHECK-LABEL: @test_no_simplify1(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @strpbrk(i8* [[STR:%.*]], i8* [[PAT:%.*]])
+; CHECK-NEXT:    ret i8* [[RET]]
+;
 
   %ret = call i8* @strpbrk(i8* %str, i8* %pat)
-; CHECK-NEXT: %ret = call i8* @strpbrk(i8* %str, i8* %pat)
   ret i8* %ret
-; CHECK-NEXT: ret i8* %ret
 }

llvm/trunk/test/Transforms/InstCombine/strrchr-1.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the strrchr library call simplifier works correctly.
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 
 @hello = constant [14 x i8] c"hello world\5Cn\00"
 @null = constant [1 x i8] zeroinitializer
 @chp = global i8* zeroinitializer
 
 declare i8* @strrchr(i8*, i32)
 
 define void @test_simplify1() {
-; CHECK: store i8* getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 6)
-; CHECK-NOT: call i8* @strrchr
-; CHECK: ret void
+; CHECK-LABEL: @test_simplify1(
+; CHECK-NEXT:    store i8* getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 6), i8** @chp, align 4
+; CHECK-NEXT:    ret void
+;
 
   %str = getelementptr [14 x i8], [14 x i8]* @hello, i32 0, i32 0
   %dst = call i8* @strrchr(i8* %str, i32 119)
   store i8* %dst, i8** @chp
   ret void
 }
 
 define void @test_simplify2() {
-; CHECK: store i8* null, i8** @chp, align 4
-; CHECK-NOT: call i8* @strrchr
-; CHECK: ret void
+; CHECK-LABEL: @test_simplify2(
+; CHECK-NEXT:    store i8* null, i8** @chp, align 4
+; CHECK-NEXT:    ret void
+;
 
   %str = getelementptr [1 x i8], [1 x i8]* @null, i32 0, i32 0
   %dst = call i8* @strrchr(i8* %str, i32 119)
   store i8* %dst, i8** @chp
   ret void
 }
 
 define void @test_simplify3() {
-; CHECK: store i8* getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 13)
-; CHECK-NOT: call i8* @strrchr
-; CHECK: ret void
+; CHECK-LABEL: @test_simplify3(
+; CHECK-NEXT:    store i8* getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 13), i8** @chp, align 4
+; CHECK-NEXT:    ret void
+;
 
   %src = getelementptr [14 x i8], [14 x i8]* @hello, i32 0, i32 0
   %dst = call i8* @strrchr(i8* %src, i32 0)
   store i8* %dst, i8** @chp
   ret void
 }
 
 define void @test_simplify4() {
-; CHECK: store i8* getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 13)
-; CHECK-NOT: call i8* @strrchr
-; CHECK: ret void
+; CHECK-LABEL: @test_simplify4(
+; CHECK-NEXT:    store i8* getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 13), i8** @chp, align 4
+; CHECK-NEXT:    ret void
+;
 
   %src = getelementptr [14 x i8], [14 x i8]* @hello, i32 0, i32 0
   %dst = call i8* @strrchr(i8* %src, i32 65280)
   store i8* %dst, i8** @chp
   ret void
 }
 
 define void @test_nosimplify1(i32 %chr) {
 ; CHECK-LABEL: @test_nosimplify1(
-; CHECK: call i8* @strrchr
-; CHECK: ret void
+; CHECK-NEXT:    [[DST:%.*]] = call i8* @strrchr(i8* nonnull dereferenceable(1) getelementptr inbounds ([14 x i8], [14 x i8]* @hello, i32 0, i32 0), i32 [[CHR:%.*]])
+; CHECK-NEXT:    store i8* [[DST]], i8** @chp, align 4
+; CHECK-NEXT:    ret void
+;
 
   %src = getelementptr [14 x i8], [14 x i8]* @hello, i32 0, i32 0
   %dst = call i8* @strrchr(i8* %src, i32 %chr)
   store i8* %dst, i8** @chp
   ret void
 }
+
+define i8* @test1(i8* %str, i32 %c) {
+; CHECK-LABEL: @test1(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @strrchr(i8* nonnull dereferenceable(1) [[STR:%.*]], i32 [[C:%.*]])
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+
+  %ret = call i8* @strrchr(i8* %str, i32 %c)
+  ret i8* %ret
+}
+
+define i8* @test2(i8* %str, i32 %c) "null-pointer-is-valid"="true" {
+; CHECK-LABEL: @test2(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @strrchr(i8* [[STR:%.*]], i32 [[C:%.*]])
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+
+  %ret = call i8* @strrchr(i8* %str, i32 %c)
+  ret i8* %ret
+}

llvm/trunk/test/Transforms/InstCombine/strstr-1.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; Test that the strstr library call simplifier works correctly.
 ;
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
 
 @.str = private constant [1 x i8] zeroinitializer
 @.str1 = private constant [2 x i8] c"a\00"
 @.str2 = private constant [6 x i8] c"abcde\00"
 @.str3 = private constant [4 x i8] c"bcd\00"
 
 declare i8* @strstr(i8*, i8*)
 
 ; Check strstr(str, "") -> str.
 
 define i8* @test_simplify1(i8* %str) {
 ; CHECK-LABEL: @test_simplify1(
+; CHECK-NEXT:    ret i8* [[STR:%.*]]
+;
   %pat = getelementptr inbounds [1 x i8], [1 x i8]* @.str, i32 0, i32 0
   %ret = call i8* @strstr(i8* %str, i8* %pat)
   ret i8* %ret
-; CHECK-NEXT: ret i8* %str
 }
 
 ; Check strstr(str, "a") -> strchr(str, 'a').
 
 define i8* @test_simplify2(i8* %str) {
 ; CHECK-LABEL: @test_simplify2(
+; CHECK-NEXT:    [[STRCHR:%.*]] = call i8* @strchr(i8* nonnull dereferenceable(1) [[STR:%.*]], i32 97)
+; CHECK-NEXT:    ret i8* [[STRCHR]]
+;
   %pat = getelementptr inbounds [2 x i8], [2 x i8]* @.str1, i32 0, i32 0
   %ret = call i8* @strstr(i8* %str, i8* %pat)
   ret i8* %ret
-; CHECK-NEXT: @strchr(i8* %str, i32 97)
 }
 
 ; Check strstr("abcde", "bcd") -> "abcde" + 1.
 
 define i8* @test_simplify3() {
 ; CHECK-LABEL: @test_simplify3(
+; CHECK-NEXT:    ret i8* getelementptr inbounds ([6 x i8], [6 x i8]* @.str2, i64 0, i64 1)
+;
   %str = getelementptr inbounds [6 x i8], [6 x i8]* @.str2, i32 0, i32 0
   %pat = getelementptr inbounds [4 x i8], [4 x i8]* @.str3, i32 0, i32 0
   %ret = call i8* @strstr(i8* %str, i8* %pat)
   ret i8* %ret
-; CHECK-NEXT: getelementptr inbounds ([6 x i8], [6 x i8]* @.str2, i64 0, i64 1)
 }
 
 ; Check strstr(str, str) -> str.
 
 define i8* @test_simplify4(i8* %str) {
 ; CHECK-LABEL: @test_simplify4(
+; CHECK-NEXT:    ret i8* [[STR:%.*]]
+;
   %ret = call i8* @strstr(i8* %str, i8* %str)
   ret i8* %ret
-; CHECK-NEXT: ret i8* %str
 }
 
 ; Check strstr(str, pat) == str -> strncmp(str, pat, strlen(str)) == 0.
 
 define i1 @test_simplify5(i8* %str, i8* %pat) {
 ; CHECK-LABEL: @test_simplify5(
+; CHECK-NEXT:    [[STRLEN:%.*]] = call i64 @strlen(i8* nonnull dereferenceable(1) [[PAT:%.*]])
+; CHECK-NEXT:    [[STRNCMP:%.*]] = call i32 @strncmp(i8* [[STR:%.*]], i8* [[PAT]], i64 [[STRLEN]])
+; CHECK-NEXT:    [[CMP1:%.*]] = icmp eq i32 [[STRNCMP]], 0
+; CHECK-NEXT:    ret i1 [[CMP1]]
+;
   %ret = call i8* @strstr(i8* %str, i8* %pat)
   %cmp = icmp eq i8* %ret, %str
   ret i1 %cmp
-; CHECK: [[LEN:%[a-z]+]] = call {{i[0-9]+}} @strlen(i8* %pat)
-; CHECK: [[NCMP:%[a-z]+]] = call {{i[0-9]+}} @strncmp(i8* %str, i8* %pat, {{i[0-9]+}} [[LEN]])
-; CHECK: icmp eq {{i[0-9]+}} [[NCMP]], 0
-; CHECK: ret i1
+}
+
+define i8* @test1(i8* %str1, i8* %str2) {
+; CHECK-LABEL: @test1(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @strstr(i8* nonnull dereferenceable(1) [[STR1:%.*]], i8* nonnull dereferenceable(1) [[STR2:%.*]])
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+  %ret = call i8* @strstr(i8* %str1, i8* %str2)
+  ret i8* %ret
+}
+
+define i8* @test2(i8* %str1, i8* %str2) "null-pointer-is-valid"="true" {
+; CHECK-LABEL: @test2(
+; CHECK-NEXT:    [[RET:%.*]] = call i8* @strstr(i8* [[STR1:%.*]], i8* [[STR2:%.*]])
+; CHECK-NEXT:    ret i8* [[RET]]
+;
+  %ret = call i8* @strstr(i8* %str1, i8* %str2)
+  ret i8* %ret
 }