[lld] Add more complete support for the INCLUDE command.

[itessier]

Summary

This change adds INCLUDE support to the MEMORY and SECTION commands, and to output sections, as per:

https://sourceware.org/binutils/docs/ld/File-Commands.html#File-Commands

[grimar]

Looks good!
A suggestion about the test is below.

I think you can write the tests as *.test (see lld/test/ELF/linkerscript/implicit-program-header.test for example).
Idea is to make tests containing large linker scripts input easier to read by avoiding using echo for them.

You can write the following line instead of asm code I think:
# RUN: echo 'nop; .section .data,"aw"; .quad 0' | llvm-mc -filetype=obj -triple=x86_64-unknown-linux - -o %t.o

[ruiu]

test/ELF/linkerscript/memory-include.s

3 ↗

(On Diff #168544)

Please construct a linker script file like this

echo foo > %t.inc
echo bar >> %t.inc
echo baz >> %t.inc

the reason being that %t.inc will have newline characters this way. If you use line concatenation, all lines are concatenated into a single line, and if something goes wrong, its error message is hard to read because a line is too long.

[ruiu]

Or, as George pointed out, you can make linker script as independent files.

[itessier]

I converted all the tests into .test files. I removed the empty file test in memory-include.test since it requires modifying the linker script. I could add a separate linker script in the input data directory for this case, or maybe just simplify the remaining tests by removing the empty file checks?

[ruiu]

LGTM

[grimar]

LGTM with a minor nit.

test/ELF/linkerscript/section-include.test

11

Please align this:

# CHECK1:      .data  00000008 0000000000002000 DATA
# CHECK1-NEXT: .data3 00000008 0000000000002008 DATA

18

And this.

[itessier]

Done, aligned the check lines.

[itessier]

Can somebody submit this for me? I don't have SVN access.

ELF/ScriptParser.cpp

void ScriptParser::readSections() {
   expect("{");
   std::vector<BaseCommand *> V;
   while (!errorCount() && !consume("}")) {
     StringRef Tok = next();
     if (Tok == "OVERLAY") {
       for (BaseCommand *Cmd : readOverlay())
         V.push_back(Cmd);
       continue;
+    } else if (Tok == "INCLUDE") {
+      readInclude();
+      continue;
     }
 
     if (BaseCommand *Cmd = readAssignment(Tok))
       V.push_back(Cmd);
     else
       V.push_back(readOutputSectionDescription(Tok));
   }
 
while (!errorCount() && !consume("}")) {
     } else if (Tok == "CONSTRUCTORS") {
       // CONSTRUCTORS is a keyword to make the linker recognize C++ ctors/dtors
       // by name. This is for very old file formats such as ECOFF/XCOFF.
       // For ELF, we should ignore.
     } else if (Tok == "FILL") {
       Cmd->Filler = readFill();
     } else if (Tok == "SORT") {
       readSort();
+    } else if (Tok == "INCLUDE") {
+      readInclude();
     } else if (peek() == "(") {
       Cmd->SectionCommands.push_back(readInputSectionDescription(Tok));
     } else {
       setError("unknown command " + Tok);
     }
   }
 
   if (consume(">"))

 
 // Parse the MEMORY command as specified in:
 // https://sourceware.org/binutils/docs/ld/MEMORY.html
 //
 // MEMORY { name [(attr)] : ORIGIN = origin, LENGTH = len ... }
 void ScriptParser::readMemory() {
   expect("{");
   while (!errorCount() && !consume("}")) {
-    StringRef Name = next();
+    StringRef Tok = next();
+    if (Tok == "INCLUDE") {
+      readInclude();
+      continue;
+    }
 
     uint32_t Flags = 0;
     uint32_t NegFlags = 0;
     if (consume("(")) {
       std::tie(Flags, NegFlags) = readMemoryAttributes();
       expect(")");
     }
     expect(":");
 
     uint64_t Origin = readMemoryAssignment("ORIGIN", "org", "o");
     expect(",");
     uint64_t Length = readMemoryAssignment("LENGTH", "len", "l");
 
     // Add the memory region to the region map.
     MemoryRegion *MR =
-        make<MemoryRegion>(Name, Origin, Length, Flags, NegFlags);
-    if (!Script->MemoryRegions.insert({Name, MR}).second)
-      setError("region '" + Name + "' already defined");
+        make<MemoryRegion>(Tok, Origin, Length, Flags, NegFlags);
+    if (!Script->MemoryRegions.insert({Tok, MR}).second)
+      setError("region '" + Tok + "' already defined");
   }
 }
 
 // This function parses the attributes used to match against section
 // flags when placing output sections in a memory region. These flags
 // are only used when an explicit memory region name is not used.
 std::pair<uint32_t, uint32_t> ScriptParser::readMemoryAttributes() {
   uint32_t Flags = 0;

test/ELF/linkerscript/memory-include.test

+# REQUIRES: x86
+
+# RUN: echo '.section .text,"ax"; .global _start; nop' > %t.s
+# RUN: echo '.section .data,"aw"; .quad 0' >> %t.s
+# RUN: llvm-mc -filetype=obj -triple=x86_64-unknown-linux %t.s -o %t.o
+
+# RUN: echo "RAM2 (rwx): ORIGIN = 0x3000, LENGTH = 0x100" > %t.inc
+# RUN: ld.lld -o %t.elf --script %s %t.o -L %T
+# RUN: llvm-objdump -section-headers %t.elf | FileCheck %s
+# CHECK: .data         00000008 0000000000002000 DATA
+# CHECK: .data2        00000008 0000000000003000 DATA
+
+MEMORY {
+  ROM (rwx): ORIGIN = 0x1000, LENGTH = 0x100
+  RAM (rwx): ORIGIN = 0x2000, LENGTH = 0x100
+  INCLUDE "memory-include.test.tmp.inc"
+}
+
+SECTIONS {
+  .text : { *(.text*) } > ROM
+  .data : { *(.data*) } > RAM
+  .data2 : { QUAD(0) } > RAM2
+}

test/ELF/linkerscript/output-section-include.test

+# REQUIRES: x86
+
+# RUN: echo '.section .text,"ax"; .global _start; nop' > %t.s
+# RUN: echo '.section .data,"aw"; .quad 0' >> %t.s
+# RUN: llvm-mc -filetype=obj -triple=x86_64-unknown-linux %t.s -o %t.o
+
+## Empty include file.
+# RUN: echo "" > %t.inc
+# RUN: ld.lld -o %t.elf --script %s %t.o -L %T
+# RUN: llvm-objdump -section-headers %t.elf | FileCheck %s --check-prefix=CHECK1
+# CHECK1: .data         00000008 0000000000002000 DATA
+
+## Non-empty include file.
+# RUN: echo "QUAD(0)" > %t.inc
+# RUN: ld.lld -o %t.elf --script %s %t.o -L %T
+# RUN: llvm-objdump -section-headers %t.elf | FileCheck %s --check-prefix=CHECK2
+# CHECK2: .data         00000010 0000000000002000 DATA
+
+MEMORY {
+  ROM (rwx): ORIGIN = 0x1000, LENGTH = 0x100
+  RAM (rwx): ORIGIN = 0x2000, LENGTH = 0x100
+}
+
+SECTIONS {
+  .text : { *(.text*) } > ROM
+  .data : {
+    *(.data*)
+    INCLUDE "output-section-include.test.tmp.inc"
+  } > RAM
+}

test/ELF/linkerscript/section-include.test

+# REQUIRES: x86
+
+# RUN: echo '.section .text,"ax"; .global _start; nop' > %t.s
+# RUN: echo '.section .data,"aw"; .quad 0' >> %t.s
+# RUN: llvm-mc -filetype=obj -triple=x86_64-unknown-linux %t.s -o %t.o
+
+## Empty include file.
+# RUN: echo "" > %t.inc
+# RUN: ld.lld -o %t.elf --script %s %t.o -L %T
+# RUN: llvm-objdump -section-headers %t.elf | FileCheck %s --check-prefix=CHECK1
+# CHECK1: .data         00000008 0000000000002000 DATA
+# CHECK1-NEXT: .data3        00000008 0000000000002008 DATA
+
+## Non-empty include file.
+# RUN: echo ".data2 : { QUAD(0) } > RAM" > %t.inc
+# RUN: ld.lld -o %t.elf --script %s %t.o -L %T
+# RUN: llvm-objdump -section-headers %t.elf | FileCheck %s --check-prefix=CHECK2
+# CHECK2: .data         00000008 0000000000002000 DATA
+# CHECK2-NEXT: .data2        00000008 0000000000002008 DATA
+# CHECK2-NEXT: .data3        00000008 0000000000002010 DATA
+
+MEMORY {
+  ROM (rwx): ORIGIN = 0x1000, LENGTH = 0x100
+  RAM (rwx): ORIGIN = 0x2000, LENGTH = 0x100
+}
+
+SECTIONS {
+  .text : { *(.text*) } > ROM
+  .data : { *(.data*) } > RAM
+  INCLUDE "section-include.test.tmp.inc"
+  .data3 : { QUAD(0) } > RAM
+}