Page MenuHomePhabricator

dougk (Douglas Katzman)
User

Projects

User does not belong to any projects.

User Details

User Since
Mar 31 2015, 10:45 AM (245 w, 4 d)

Recent Activity

Dec 19 2016

dougk added a comment to D26454: Implement no_sanitize_address for global vars.

I think it probably works to have the attribute appertain to any sanitizer. I did not know that it did, so I conservatively assumed that it didn't. I'll go ahead and make things consistent, and confirm that it dtrt.

Dec 19 2016, 9:32 AM

Nov 30 2016

dougk added a comment to D26454: Implement no_sanitize_address for global vars.

Suppression of sanitizing is necessary if the variable is magically a memory-mapped device I/O address.
The linker can arrange for this to be the case using fancy scripts, or even just as simple as a section attribute that requires that you take up exactly a certain number of bytes in the section.
There was some thought that any non-default section should preclude sanitization, but Kostya said that, no, it would make sense to require explicit no-sanitize. I (mistakenly) took that to mean "just do it", for which I apologize.

Nov 30 2016, 12:23 PM

Nov 14 2016

dougk updated the diff for D26454: Implement no_sanitize_address for global vars.

add a sentence about the change in AddressSanitizer.rst

Nov 14 2016, 8:12 AM

Nov 11 2016

dougk updated the diff for D26454: Implement no_sanitize_address for global vars.

changes per Aaron Ballman

Nov 11 2016, 12:43 PM
dougk committed rL286599: Fix mismatched enum value name and diagnostic text..
Fix mismatched enum value name and diagnostic text.
Nov 11 2016, 9:01 AM
dougk closed D26459: Fix mismatched enum value name and diagnostic text. by committing rL286599: Fix mismatched enum value name and diagnostic text..
Nov 11 2016, 9:01 AM

Nov 9 2016

dougk added a comment to D26454: Implement no_sanitize_address for global vars.

Also note: In this change, the diagnostic messages are correct; it's the enum name that's bad. The diagnostic message is wrong for the 'section' attribute, which is fixed by https://reviews.llvm.org/D26459

Nov 9 2016, 12:07 PM
dougk updated the diff for D26459: Fix mismatched enum value name and diagnostic text..

inadvertent omission of attr-section.c test

Nov 9 2016, 9:54 AM
dougk retitled D26459: Fix mismatched enum value name and diagnostic text. from to Fix mismatched enum value name and diagnostic text..
Nov 9 2016, 9:49 AM
dougk committed rL286380: [Sparc] Unbreak test.
[Sparc] Unbreak test
Nov 9 2016, 9:11 AM
dougk retitled D26454: Implement no_sanitize_address for global vars from to Implement no_sanitize_address for global vars.
Nov 9 2016, 8:44 AM
dougk committed rL286376: [Sparc] LLONG is not lock-free atomic on v8.
[Sparc] LLONG is not lock-free atomic on v8
Nov 9 2016, 7:53 AM
dougk closed D26286: [Sparc]: correct the ATOMIC_LLONG_LOCK_FREE macro by committing rL286376: [Sparc] LLONG is not lock-free atomic on v8.
Nov 9 2016, 7:53 AM

Nov 3 2016

dougk updated the diff for D26286: [Sparc]: correct the ATOMIC_LLONG_LOCK_FREE macro.

fix atomics-inlining test

Nov 3 2016, 12:24 PM
dougk retitled D26286: [Sparc]: correct the ATOMIC_LLONG_LOCK_FREE macro from to [Sparc]: correct the ATOMIC_LLONG_LOCK_FREE macro.
Nov 3 2016, 11:11 AM

Oct 25 2016

dougk committed rL285144: [Myriad] add empty file to simulated bin dir.
[Myriad] add empty file to simulated bin dir
Oct 25 2016, 5:08 PM
dougk committed rL285140: [Myriad] Find libc++ adjacent to libstdc++.
[Myriad] Find libc++ adjacent to libstdc++
Oct 25 2016, 4:12 PM

Oct 19 2016

dougk added a comment to D25690: [X86] Correct the number of bits in a fixup.

this changes harms the .byte,.short, etc assembler directives. niravd will take a look

Oct 19 2016, 4:38 PM
dougk committed rL284617: [Myriad] Find libc++ headers next to clang binary.
[Myriad] Find libc++ headers next to clang binary
Oct 19 2016, 10:40 AM

Oct 17 2016

dougk added a comment to D25690: [X86] Correct the number of bits in a fixup.

it looks like we need an MCAssembler on which to call getContext() and then Ctx->reportError(Fixup.getLoc(), "mumble").
Some of the other AsmBackends have started to have this plumbing added in.
However most of them actually guard the emission of the message with "if (Ctx)" because we usually don't have the context apparently, so you really can't get a nice source location anyway.
Without this change, we'll silently emit bogus code, so I think we should land it as-is, but do change it from assert to report_fatal_error. Are you ok with that?

Oct 17 2016, 1:01 PM
dougk added a comment to D25690: [X86] Correct the number of bits in a fixup.

We can change it from an assert to report_fatal_error, but we can't actually distinguish at this point whether it was in fact a user error versus whether it was an internal bug.
I guess I agree that dumping core is a bit sad for the user error case. Would you prefer it always to say this instead?

Oct 17 2016, 12:21 PM
dougk retitled D25690: [X86] Correct the number of bits in a fixup from to [X86] Correct the number of bits in a fixup.
Oct 17 2016, 12:09 PM

Oct 14 2016

dougk committed rL284272: Implement no_sanitize_address for global vars.
Implement no_sanitize_address for global vars
Oct 14 2016, 1:04 PM

Oct 7 2016

dougk committed rL283601: Visually align table def with respective enum. NFC.
Visually align table def with respective enum. NFC
Oct 7 2016, 2:11 PM

Oct 5 2016

dougk committed rL283326: [X86] Don't randomly encode %rip where illegal.
[X86] Don't randomly encode %rip where illegal
Oct 5 2016, 8:32 AM
dougk closed D25112: Don't randomly encode %rip where illegal by committing rL283326: [X86] Don't randomly encode %rip where illegal.
Oct 5 2016, 8:32 AM
dougk updated the diff for D25112: Don't randomly encode %rip where illegal.

test Intel syntax

Oct 5 2016, 8:28 AM
dougk committed rL283322: [X86] Fix some tests that didn't assert anything.
[X86] Fix some tests that didn't assert anything
Oct 5 2016, 7:55 AM

Sep 30 2016

dougk updated the diff for D25112: Don't randomly encode %rip where illegal.

handle Intel syntax also

Sep 30 2016, 2:49 PM
dougk retitled D25112: Don't randomly encode %rip where illegal from to Don't randomly encode %rip where illegal.
Sep 30 2016, 11:13 AM

Sep 29 2016

dougk committed rL282757: [Myriad] Silently ignore -fno-split-dwarf-inlining.
[Myriad] Silently ignore -fno-split-dwarf-inlining
Sep 29 2016, 9:11 PM
dougk committed rL282755: Generalize ArgList::AddAllArgs more.
Generalize ArgList::AddAllArgs more
Sep 29 2016, 9:11 PM
dougk committed rL282732: [X86] Avoid "unused" warnings if no asserts.
[X86] Avoid "unused" warnings if no asserts
Sep 29 2016, 9:10 PM

Sep 13 2016

dougk committed rL281359: [Myriad]: set LeonCASA processor feature.
[Myriad]: set LeonCASA processor feature
Sep 13 2016, 11:00 AM

Sep 9 2016

dougk committed rL281071: Myriad: nominally "support" ASAN..
Myriad: nominally "support" ASAN.
Sep 9 2016, 11:29 AM

Aug 29 2016

dougk committed rL279996: [Myriad]: add missing 'mcpu' values.
[Myriad]: add missing 'mcpu' values
Aug 29 2016, 12:51 PM

Jul 25 2016

dougk committed rL276646: [Myriad]: better compatibility with vendor source.
[Myriad]: better compatibility with vendor source
Jul 25 2016, 9:43 AM

Jul 21 2016

dougk committed rL276362: [Sparc]: Fix bug in LowerSTORE due to r275592.
[Sparc]: Fix bug in LowerSTORE due to r275592
Jul 21 2016, 4:36 PM

May 9 2016

dougk updated the diff for D20079: Avoid gcc warning about casting ptr-to-object to ptr-to-fun.

fix declaration

May 9 2016, 7:19 PM
dougk retitled D20079: Avoid gcc warning about casting ptr-to-object to ptr-to-fun from to Avoid gcc warning about casting ptr-to-object to ptr-to-fun.
May 9 2016, 1:29 PM
dougk committed rL268956: [Myriad] Use Generic_ELF::addClangTargetOptions().
[Myriad] Use Generic_ELF::addClangTargetOptions()
May 9 2016, 12:16 PM

Apr 19 2016

dougk committed rL266797: Fix PR26999 better- RenderDebugEnablingArgs() once only.
Fix PR26999 better- RenderDebugEnablingArgs() once only
Apr 19 2016, 12:01 PM
dougk committed rL266775: Pass dwarf-version to cc1as..
Pass dwarf-version to cc1as.
Apr 19 2016, 10:49 AM

Mar 28 2016

dougk committed rL264623: Use 'alignTo' instead of re-inventing it. NFC.
Use 'alignTo' instead of re-inventing it. NFC
Mar 28 2016, 12:28 PM
dougk committed rL264579: Sparc: silently ignore .proc assembler directive.
Sparc: silently ignore .proc assembler directive
Mar 28 2016, 7:05 AM
dougk closed D18463: Sparc: silently ignore .proc assembler directive by committing rL264579: Sparc: silently ignore .proc assembler directive.
Mar 28 2016, 7:05 AM

Mar 24 2016

dougk retitled D18463: Sparc: silently ignore .proc assembler directive from to Sparc: silently ignore .proc assembler directive.
Mar 24 2016, 3:09 PM

Mar 15 2016

dougk committed rL263596: Myriad: define __myriad2 macro automatically.
Myriad: define __myriad2 macro automatically
Mar 15 2016, 3:39 PM
dougk committed rL263557: Myriad: Add new sparc CPU kinds..
Myriad: Add new sparc CPU kinds.
Mar 15 2016, 9:46 AM
dougk committed rL263556: Myriad: Pass -mcpu to movi{Compile,Asm}.
Myriad: Pass -mcpu to movi{Compile,Asm}
Mar 15 2016, 9:46 AM
dougk closed D18186: Myriad: pass -mcpu flag to moviCompile, no default by committing rL263556: Myriad: Pass -mcpu to movi{Compile,Asm}.
Mar 15 2016, 9:46 AM
dougk updated the diff for D18186: Myriad: pass -mcpu flag to moviCompile, no default.

also moviAsm

Mar 15 2016, 8:37 AM
dougk retitled D18186: Myriad: pass -mcpu flag to moviCompile, no default from to Myriad: pass -mcpu flag to moviCompile, no default.
Mar 15 2016, 8:11 AM

Jan 5 2016

dougk committed rL256897: Avoid assert failure on some invalid cc1 options..
Avoid assert failure on some invalid cc1 options.
Jan 5 2016, 5:41 PM
dougk closed D15882: Avoid assert failure on some invalid cc1 options. by committing rL256897: Avoid assert failure on some invalid cc1 options..
Jan 5 2016, 5:41 PM

Jan 4 2016

dougk retitled D15882: Avoid assert failure on some invalid cc1 options. from to Avoid assert failure on some invalid cc1 options..
Jan 4 2016, 7:51 PM

Dec 16 2015

dougk accepted D15579: Fix the failing windows clang unit tests. NFC.
Dec 16 2015, 10:29 AM

Nov 19 2015

dougk committed rL253647: [Myriad]: handle Preprocess job action (-E).
[Myriad]: handle Preprocess job action (-E)
Nov 19 2015, 9:00 PM

Nov 18 2015

dougk committed rL253476: [Myriad]: fix test for Windows.
[Myriad]: fix test for Windows
Nov 18 2015, 9:22 AM
dougk committed rL253467: [Myriad]: insert -L paths into linker cmd only when they exist..
[Myriad]: insert -L paths into linker cmd only when they exist.
Nov 18 2015, 8:27 AM
dougk closed D14754: [Myriad]: insert -L paths into linker cmd only when they exist. by committing rL253467: [Myriad]: insert -L paths into linker cmd only when they exist..
Nov 18 2015, 8:27 AM

Nov 17 2015

dougk updated the diff for D14754: [Myriad]: insert -L paths into linker cmd only when they exist..

changes per jyknight

Nov 17 2015, 3:07 PM
dougk committed rL253390: [Myriad]: -nostdlib implies -nostartfiles.
[Myriad]: -nostdlib implies -nostartfiles
Nov 17 2015, 2:36 PM
dougk retitled D14754: [Myriad]: insert -L paths into linker cmd only when they exist. from to [Myriad]: insert -L paths into linker cmd only when they exist..
Nov 17 2015, 1:12 PM
dougk committed rL253372: Use !hasArg with two options instead of !hasArg && !hasArg..
Use !hasArg with two options instead of !hasArg && !hasArg.
Nov 17 2015, 12:30 PM
dougk committed rL253350: Add trivial utility to append -L arguments to linker step. NFC.
Add trivial utility to append -L arguments to linker step. NFC
Nov 17 2015, 9:44 AM

Nov 16 2015

dougk committed rL253213: [Myriad]: pass the 'std=' option to moviCompile.
[Myriad]: pass the 'std=' option to moviCompile
Nov 16 2015, 7:42 AM

Nov 11 2015

dougk closed D13316: Visibly fail if attempting to encode register AH,BH,CH,DH in a REX-prefixed instruction..
Nov 11 2015, 8:34 AM
dougk committed rL252743: Visibly fail if attempting to encode register AH,BH,CH,DH in a REX-prefixed….
Visibly fail if attempting to encode register AH,BH,CH,DH in a REX-prefixed…
Nov 11 2015, 7:53 AM

Nov 10 2015

dougk accepted D14548: Make test/Driver/biarch.c use FileCheck instead of grep.
Nov 10 2015, 1:35 PM
dougk updated the diff for D13316: Visibly fail if attempting to encode register AH,BH,CH,DH in a REX-prefixed instruction..

Simpler fix per Craig Topper

Nov 10 2015, 12:32 PM

Oct 21 2015

dougk committed rL250932: [Myriad]: Always add -L paths even if -nostdlib is set..
[Myriad]: Always add -L paths even if -nostdlib is set.
Oct 21 2015, 12:36 PM

Oct 16 2015

dougk accepted D13813: Teach MyriadToolchain how to find its C++ header paths..

I'd have titled the CL such that emphasis on what people might find generally useful:
"Move addLibStdCXXIncludePaths method from Linux to Generic_GCC .. and incidentally make MyriadToolchain use that"

Oct 16 2015, 9:41 AM

Oct 14 2015

dougk committed rL250363: Remove unnecessary braces in single-line 'if'..
Remove unnecessary braces in single-line 'if'.
Oct 14 2015, 9:12 PM

Oct 13 2015

dougk added a comment to rL249655: Stop messing with the 'g' group of options in CompilerInvocation..

thanks for the test case. Fixed by http://reviews.llvm.org/rL250173

Oct 13 2015, 9:27 AM
dougk committed rL250173: Always pass a -dwarf-version argument to integrated as..
Always pass a -dwarf-version argument to integrated as.
Oct 13 2015, 9:26 AM

Oct 9 2015

dougk committed rL249893: [Myriad]: put libstdc++ and libc in the right order.
[Myriad]: put libstdc++ and libc in the right order
Oct 9 2015, 1:28 PM

Oct 8 2015

dougk committed rL249692: [Myriad]: default the Dwarf version to 2.
[Myriad]: default the Dwarf version to 2
Oct 8 2015, 7:19 AM

Oct 7 2015

dougk committed rL249658: Use itostr(), not std::to_string() because of Android..
Use itostr(), not std::to_string() because of Android.
Oct 7 2015, 10:27 PM
dougk committed rL249657: Unbreak 'debug-options' test when builder is Darwin.
Unbreak 'debug-options' test when builder is Darwin
Oct 7 2015, 10:04 PM
dougk abandoned D12845: [Sparc]: Myriad toolchain should assume -gdwarf-3 in cc1 invocation .
Oct 7 2015, 9:34 PM
dougk committed rL249655: Stop messing with the 'g' group of options in CompilerInvocation..
Stop messing with the 'g' group of options in CompilerInvocation.
Oct 7 2015, 9:29 PM
dougk closed D13221: Make CompilerInvocation's use of the debug options more understandable. by committing rL249655: Stop messing with the 'g' group of options in CompilerInvocation..
Oct 7 2015, 9:27 PM
dougk added a comment to D13221: Make CompilerInvocation's use of the debug options more understandable..
Oct 7 2015, 10:59 AM
dougk updated the diff for D13221: Make CompilerInvocation's use of the debug options more understandable..

spell 'di-kind=' as 'debug-info-kind=', spell 'full' as 'standalone', spell 'line-tables' as 'line-tables-only', spell internal option name as debug_info_kind_EQ

Oct 7 2015, 8:09 AM
dougk added a comment to D13221: Make CompilerInvocation's use of the debug options more understandable..

James, you're right, 'gdwarf-2' followed by 'line-tables-only' works, but the opposite order doesn't.
So it halfway works, which is better than not working at all.

Oct 7 2015, 8:04 AM

Oct 2 2015

dougk added a comment to D13221: Make CompilerInvocation's use of the debug options more understandable..

A few more remarks:

Oct 2 2015, 4:36 PM
dougk added a comment to D13221: Make CompilerInvocation's use of the debug options more understandable..

A few more remarks:

Oct 2 2015, 4:31 PM
dougk committed rL249166: [Myriad]: Accept '-nostdlib' option.
[Myriad]: Accept '-nostdlib' option
Oct 2 2015, 11:41 AM
dougk committed rL249138: Fix bogus comment..
Fix bogus comment.
Oct 2 2015, 7:43 AM

Sep 30 2015

dougk retitled D13316: Visibly fail if attempting to encode register AH,BH,CH,DH in a REX-prefixed instruction. from to Visibly fail if attempting to encode register AH,BH,CH,DH in a REX-prefixed instruction..
Sep 30 2015, 4:33 PM
dougk updated the diff for D13221: Make CompilerInvocation's use of the debug options more understandable..

Fix CollectArgsForIntegratedAs and make sure that no other tests talk about "-g" after the Driver has produced the command invocation, except for xcore in which apparently xcc does accept -g.

Sep 30 2015, 12:55 PM
dougk committed rL248912: Alias "-ggdbN" to "-gN" options, fixing some incompatibilities..
Alias "-ggdbN" to "-gN" options, fixing some incompatibilities.
Sep 30 2015, 8:57 AM

Sep 29 2015

dougk updated the diff for D13221: Make CompilerInvocation's use of the debug options more understandable..

Replaced the cc1 arguments with two three-valued options. Fixed all the tests.

Sep 29 2015, 9:23 AM

Sep 28 2015

dougk added a comment to D13221: Make CompilerInvocation's use of the debug options more understandable..

I think the best thing to do is introduce new cc1 arguments that control the Opts explicitly, and cease accepting the equivalent driver arguments in cc1.
I would propose '-gdwarf-version={2|3|4}' and '-gdebug-info-kind={line-tables|limited|full}'

Sep 28 2015, 12:38 PM
dougk retitled D13221: Make CompilerInvocation's use of the debug options more understandable. from to Make CompilerInvocation's use of the debug options more understandable..
Sep 28 2015, 12:18 PM

Sep 18 2015

dougk committed rL247995: Use None instead of an explicit constructor..
Use None instead of an explicit constructor.
Sep 18 2015, 8:24 AM
dougk abandoned D10841: [Shave]: allow Clang to run the target linker..
Sep 18 2015, 5:00 AM

Sep 17 2015

dougk committed rL247948: [Myriad]: add "<sysroot>/include" to standard search path.
[Myriad]: add "<sysroot>/include" to standard search path
Sep 17 2015, 2:23 PM
dougk committed rL247932: Try to unbreak windows compiler after r247926..
Try to unbreak windows compiler after r247926.
Sep 17 2015, 1:26 PM