[ELF] - Basic support of linkerscript commands: DATA_SEGMENT_ALIGN, DATA_SEGMENT_END, CONSTANT

[grimar]

Summary

It is called basic because:

1. CONSTANT expression can refer to COMMONPAGESIZE and MAXPAGESIZE.

This sizes are usually different and used for possible optimization of
memory consumption.
More details are here: https://sourceware.org/ml/binutils/2002-02/msg00265.html
We currently do not support this optimization, so both CONSTANT(MAXPAGESIZE)
and CONSTANT(COMMONPAGESIZE) just return Target->PageSize value.

2. DATA_SEGMENT_ALIGN and DATA_SEGMENT_END are used as a part of opt.

The latter one is just ignored now.
According to documentation DATA_SEGMENT_ALIGN has 2 possible
calculation, but since we do not support mentioned opt - it
is always calculated now as (ALIGN(MAXPAGESIZE) + (. & (MAXPAGESIZE - 1))).

In general this should work for now until we deside to support this opt.

[ruiu]

This comment is not particularly to this patch, but it seems that you are working on the linker script and starting it from easy part. This patch is, for example, easy to implement. We can add small features such as this later after we complete the main part. There is also a chance that we would end up have to refactor the current code so we would have to rewrite everything -- so adding small features first is a bit risky. Therefore, can you focus on design instead of coding? We need a plan with which we are sure we'll be able to implement all the (well, probably not all, but all sane) linker script features.

[davide]

It would be particularly nice to have designs on how to handle features like 'SECTION', which not only seem to be larger but also used by more complicated linker scripts like, e.g., the ones used by kernels.

[grimar]

In D19663#415938, @ruiu wrote:

This comment is not particularly to this patch, but it seems that you are working on the linker script and starting it from easy part. This patch is, for example, easy to implement. We can add small features such as this later after we complete the main part. There is also a chance that we would end up have to refactor the current code so we would have to rewrite everything -- so adding small features first is a bit risky. Therefore, can you focus on design instead of coding? We need a plan with which we are sure we'll be able to implement all the (well, probably not all, but all sane) linker script features.

Ok.
About this little patch: if was fast to do and I assumed that expression evaluation and linker script itself are more or less independent components. And even if we will rewrite everything, such patches introduces new testcases, that are potentially usefull.

[grimar]

• Rebased.

So I guess after all that LS commits we now are ready to support this.
It opens a way for linker script constants as well.

[evgeny777]

ELF/LinkerScript.cpp

158 ↗

(On Diff #64912)

Looks like you can write

expect("(.)")

[grimar]

ELF/LinkerScript.cpp

158 ↗

(On Diff #64912)

This is 3 different tokens, this will not work.

[evgeny777]

ELF/LinkerScript.cpp

158 ↗

(On Diff #64912)

Yeah, you're right, sorry.

[ruiu]

LGTM

ELF/LinkerScript.cpp

104 ↗

(On Diff #64912)

static?

105–109 ↗

(On Diff #64912)

Are you describing the same thing in two places in the same commit? Maybe you want to leave one of them and remove the other?

137 ↗

(On Diff #64912)

Can you consistently use V as a return value in this function just like other if clauses do?

[grimar]

ELF/LinkerScript.cpp

104 ↗

(On Diff #64912)

Done.

105–109 ↗

(On Diff #64912)

Removed.

137 ↗

(On Diff #64912)

Sure, sorry.

lld/trunk/ELF/LinkerScript.cpp

 
 uint64_t ExprParser::run() {
   uint64_t V = parseExpr();
   if (!atEOF() && !Error)
     setError("stray token: " + peek());
   return V;
 }
 
+uint64_t static getConstantValue(StringRef C) {
+  if (C == "COMMONPAGESIZE" || C == "MAXPAGESIZE")
+    return Target->PageSize;
+  error("unknown constant: " + C);
+  return 0;
+}
+
 // This is a part of the operator-precedence parser to evaluate
 // arithmetic expressions in SECTIONS command. This function evaluates an
 // integer literal, a parenthesized expression, the ALIGN function,
 // or the special variable ".".
 uint64_t ExprParser::parsePrimary() {
   StringRef Tok = next();
   if (Tok == ".")
     return Dot;
   if (Tok == "(") {
     uint64_t V = parseExpr();
     expect(")");
     return V;
   }
   if (Tok == "ALIGN") {
     expect("(");
     uint64_t V = parseExpr();
     expect(")");
     return alignTo(Dot, V);
   }
+  if (Tok == "CONSTANT") {
+    expect("(");
+    uint64_t V = getConstantValue(next());
+    expect(")");
+    return V;
+  }
+  // Documentations says there are two ways to compute
+  // the value of DATA_SEGMENT_ALIGN command, depending on whether the second
+  // uses fewer COMMONPAGESIZE sized pages for the data segment(area between the
+  // result of this expression and `DATA_SEGMENT_END') than the first or not.
+  // That is possible optimization, that we do not support, so we compute that
+  // function always as (ALIGN(MAXPAGESIZE) + (. & (MAXPAGESIZE - 1))) now.
+  if (Tok == "DATA_SEGMENT_ALIGN") {
+    expect("(");
+    uint64_t L = parseExpr();
+    expect(",");
+    parseExpr();
+    expect(")");
+    return alignTo(Dot, L) + (Dot & (L - 1));
+  }
+  // Since we do not support the optimization from comment above,
+  // we can just ignore that command.
+  if (Tok == "DATA_SEGMENT_END") {
+    expect("(");
+    expect(".");
+    expect(")");
+    return Dot;
+  }
   uint64_t V = 0;
   if (Tok.getAsInteger(0, V))
     setError("malformed number: " + Tok);
   return V;
 }
 
 uint64_t ExprParser::parseTernary(uint64_t Cond) {
   next();

lld/trunk/test/ELF/linkerscript-locationcounter.s

 # RUN:  . = 0x1 > 0x0 ? 0x20000 : 0x999999; \
 # RUN:  .great : { *(.great) } \
 # RUN:  . = 0x1 >= 0x1 ? 0x21000 : 0x999999; \
 # RUN:  .greateq : { *(.greateq) } \
 # RUN:  . = 0x1 == 0x1 ? 0x22000 : 0x999999; \
 # RUN:  .eq : { *(.eq) } \
 # RUN:  . = 0x2 != 0x1 ? 0x23000 : 0x999999; \
 # RUN:  .neq : { *(.neq) } \
+# RUN:  . = CONSTANT (MAXPAGESIZE) * 0x24; \
+# RUN:  .maxpagesize : { *(.maxpagesize) } \
+# RUN:  . = CONSTANT (COMMONPAGESIZE) * 0x25; \
+# RUN:  .commonpagesize : { *(.commonpagesize) } \
+# RUN:  . = DATA_SEGMENT_ALIGN (CONSTANT (MAXPAGESIZE), CONSTANT (COMMONPAGESIZE)); \
+# RUN:  .datasegmentalign : { *(.datasegmentalign) } \
+# RUN:  . = DATA_SEGMENT_END (.); \
 # RUN: }" > %t.script
 # RUN: ld.lld %t --script %t.script -o %t2
 # RUN: llvm-readobj -s %t2 | FileCheck %s
 
 # CHECK: Section {
 # CHECK:   Index: 1
 # CHECK:   Name: .plus
 # CHECK-NEXT:   Type: SHT_PROGBITS

 # CHECK-NEXT:   Address: 0x23000
 # CHECK-NEXT:   Offset:
 # CHECK-NEXT:   Size:
 # CHECK-NEXT:   Link:
 # CHECK-NEXT:   Info:
 # CHECK-NEXT:   AddressAlignment:
 # CHECK-NEXT:   EntrySize:
 # CHECK-NEXT: }
+# CHECK-NEXT: Section {
+# CHECK-NEXT:   Index:
+# CHECK-NEXT:   Name: .maxpagesize
+# CHECK-NEXT:   Type: SHT_PROGBITS
+# CHECK-NEXT:   Flags [
+# CHECK-NEXT:     SHF_ALLOC
+# CHECK-NEXT:   ]
+# CHECK-NEXT:   Address: 0x24000
+# CHECK-NEXT:   Offset:
+# CHECK-NEXT:   Size:
+# CHECK-NEXT:   Link:
+# CHECK-NEXT:   Info:
+# CHECK-NEXT:   AddressAlignment:
+# CHECK-NEXT:   EntrySize: 0
+# CHECK-NEXT: }
+# CHECK-NEXT: Section {
+# CHECK-NEXT:   Index:
+# CHECK-NEXT:   Name: .commonpagesize
+# CHECK-NEXT:   Type: SHT_PROGBITS
+# CHECK-NEXT:   Flags [
+# CHECK-NEXT:     SHF_ALLOC
+# CHECK-NEXT:   ]
+# CHECK-NEXT:   Address: 0x25000
+# CHECK-NEXT:   Offset:
+# CHECK-NEXT:   Size:
+# CHECK-NEXT:   Link:
+# CHECK-NEXT:   Info:
+# CHECK-NEXT:   AddressAlignment:
+# CHECK-NEXT:   EntrySize:
+# CHECK-NEXT: }
+# CHECK-NEXT: Section {
+# CHECK-NEXT:   Index:
+# CHECK-NEXT:   Name: .datasegmentalign
+# CHECK-NEXT:   Type: SHT_PROGBITS
+# CHECK-NEXT:   Flags [
+# CHECK-NEXT:     SHF_ALLOC
+# CHECK-NEXT:   ]
+# CHECK-NEXT:   Address: 0x26008
+# CHECK-NEXT:   Offset:
+# CHECK-NEXT:   Size:
+# CHECK-NEXT:   Link:
+# CHECK-NEXT:   Info:
+# CHECK-NEXT:   AddressAlignment:
+# CHECK-NEXT:   EntrySize:
+# CHECK-NEXT: }
 
 ## Mailformed number error.
 # RUN: echo "SECTIONS { \
 # RUN:  . = 0x12Q41; \
 # RUN: }" > %t.script
 # RUN: not ld.lld %t --script %t.script -o %t2 2>&1 | \
 # RUN:  FileCheck --check-prefix=NUMERR %s
 # NUMERR: malformed number: 0x12Q41

 .section .greateq, "a"
 .quad 0
 
 .section .eq, "a"
 .quad 0
 
 .section .neq, "a"
 .quad 0
+
+.section .maxpagesize, "a"
+.quad 0
+
+.section .commonpagesize, "a"
+.quad 0
+
+.section .datasegmentalign, "a"
+.quad 0