[ELF2] Sort PPC64 special sections

[hfinkel]

Summary

PPC64 has several special sections that are intended to be accessed from the TOC base pointer. When a .got is present, the TOC base pointer is .got + 0x8000 (as specified by the ABI). Furthermore, the glibc startup code contains an assumption that a 16-bit relocation can hold the offset from the TOC base value to the beginning of the .toc section. Thus, we need to make sure that .toc appears after .got. This much, at least, is required in practice. The other PPC64 special sections (.toc, .toc1, .opd, etc.) should also be close by to optimize access by smaller TOC-base-pointer offsets.

[ruiu]

ELF/Writer.cpp

361–365

You can simplify this by defining a function, say getPPCRank(StringRef), which returns an integer value according to this table.

0 .tocbss
1 any other name
2 .toc
3 .branch_lt
4 .toc
5 .toc1
6 .opd

Then you can use that function in this function like this

return getPPCRank(A) < getPPCRank(B);

[hfinkel]

Updated to use a ranking function (per review).

[ruiu]

ELF/Writer.cpp

311–317

Sorry, I meant

0 .tocbss
1 any other name
2 .got (* was a typo)
3 .branch_lt
4 .toc
5 .toc1
6 .opd

This is my interpretation of your code. I'm not 100% sure that this ranking is correct, though. So please verify.

Also please remove excessive spaces at .Case("...", <*here*><number>)

[hfinkel]

ELF/Writer.cpp

311–317

I know, but if I do that, then I'll change the order of .got relative to other like sections for all targets. Is that okay?

[ruiu]

ELF/Writer.cpp

311–317

If you don't need that for PPC now, you can leave .got as is.

[hfinkel]

ELF/Writer.cpp

311–317

Okay; I'll leave it like this for now. It is okay for my current test cases, and I'm happy to deal with future enhancements if/when they become necessary.

[ruiu]

OK. LGTM with the StringSwitch style fix.

[hfinkel]

r250100, thanks!

ELF/Writer.cpp

 //===- Writer.cpp ---------------------------------------------------------===//
 //
 //                             The LLVM Linker
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 
 #include "Writer.h"
 #include "Config.h"
 #include "OutputSections.h"
 #include "SymbolTable.h"
 #include "Target.h"
 
+#include "llvm/ADT/StringSwitch.h"
 #include "llvm/Support/FileOutputBuffer.h"
 
 using namespace llvm;
 using namespace llvm::ELF;
 using namespace llvm::object;
 
 using namespace lld;
 using namespace lld::elf2;
for (const Elf_Sym &Sym : File.getLocalSymbols()) {
       error(SymNameOrErr);
       StringRef SymName = *SymNameOrErr;
       if (shouldKeepInSymtab<ELFT>(SymName, Sym))
         Out<ELFT>::SymTab->addSymbol(SymName, true);
     }
   }
 }
 
+// PPC64 has a number of special SHT_PROGBITS+SHF_ALLOC+SHF_WRITE sections that
+// we would like to make sure appear is a specific order to maximize their
+// coverage by a single signed 16-bit offset from the TOC base pointer.
+// Conversely, the special .tocbss section should be first among all SHT_NOBITS
+// sections. This will put it next to the loaded special PPC64 sections (and,
+// thus, within reach of the TOC base pointer).
+static int getPPC64SectionRank(StringRef SectionName) {
+  return StringSwitch<int>(SectionName)
+           .Case(".tocbss",    0)
+           .Case(".branch_lt", 2)
+           .Case(".toc",       3)
+           .Case(".toc1",      4)
+           .Case(".opd",       5)
+           .Default(1);
+}
+
 // Output section ordering is determined by this function.
 template <class ELFT>
 static bool compareOutputSections(OutputSectionBase<ELFT::Is64Bits> *A,
                                   OutputSectionBase<ELFT::Is64Bits> *B) {
   typedef typename ELFFile<ELFT>::uintX_t uintX_t;
 
   uintX_t AFlags = A->getFlags();
   uintX_t BFlags = B->getFlags();
if (AIsExec != BIsExec)
     return BIsExec;
 
   // If we got here we know that both A and B and in the same PT_LOAD.
   // The last requirement we have is to put nobits section last. The
   // reason is that the only thing the dynamic linker will see about
   // them is a p_memsz that is larger than p_filesz. Seeing that it
   // zeros the end of the PT_LOAD, so that has to correspond to the
   // nobits sections.
-  return A->getType() != SHT_NOBITS && B->getType() == SHT_NOBITS;
+  if (A->getType() != SHT_NOBITS && B->getType() == SHT_NOBITS)
+    return true;
+
+  return getPPC64SectionRank(A->getName()) < getPPC64SectionRank(B->getName());
 }
 
 // Until this function is called, common symbols do not belong to any section.
 // This function adds them to end of BSS section.
 template <class ELFT>
 static void addCommonSymbols(std::vector<DefinedCommon<ELFT> *> &Syms) {
   typedef typename ELFFile<ELFT>::uintX_t uintX_t;
   typedef typename ELFFile<ELFT>::Elf_Sym Elf_Sym;
 

test/elf2/basic64be.s

 # RUN: llvm-mc -filetype=obj -triple=powerpc64-unknown-linux %s -o %t
 # RUN: ld.lld2 -discard-all %t -o %t2
 # RUN: llvm-readobj -file-headers -sections -section-data -program-headers %t2 | FileCheck %s
 # REQUIRES: ppc
 
 # exits with return code 42 on linux
 .section        ".opd","aw"
 .global _start
 _start:
 .quad   .Lfoo,.TOC.@tocbase,0
 
+# generate .toc and .toc1 sections to make sure that the ordering is as
+# intended (.toc before .toc1, and both before .opd).
+.section        ".toc1","aw"
+.quad          22, 37, 89, 47
+
+.section        ".toc","aw"
+.quad          45, 86, 72, 24
+
 .text
 .Lfoo:
   li      0,1
   li      3,42
   sc
 
 # CHECK:      ElfHeader {
 # CHECK-NEXT:   Ident {
 # CHECK-NEXT:     Magic: (7F 45 4C 46)
 # CHECK-NEXT:     Class: 64-bit (0x2)
 # CHECK-NEXT:     DataEncoding: BigEndian (0x2)
 # CHECK-NEXT:     FileVersion: 1
 # CHECK-NEXT:     OS/ABI: SystemV (0x0)
 # CHECK-NEXT:     ABIVersion: 0
 # CHECK-NEXT:     Unused: (00 00 00 00 00 00 00)
 # CHECK-NEXT:   }
 # CHECK-NEXT:   Type: Executable (0x2)
 # CHECK-NEXT:   Machine: EM_PPC64 (0x15)
 # CHECK-NEXT:   Version: 1
-# CHECK-NEXT:   Entry: 0x10020000
+# CHECK-NEXT:   Entry: 0x10020040
 # CHECK-NEXT:   ProgramHeaderOffset: 0x40
-# CHECK-NEXT:   SectionHeaderOffset: 0x20078
+# CHECK-NEXT:   SectionHeaderOffset: 0x200C8
 # CHECK-NEXT:   Flags [ (0x0)
 # CHECK-NEXT:   ]
 # CHECK-NEXT:   HeaderSize: 64
 # CHECK-NEXT:   ProgramHeaderEntrySize: 56
 # CHECK-NEXT:   ProgramHeaderCount: 3
 # CHECK-NEXT:   SectionHeaderEntrySize: 64
-# CHECK-NEXT:   SectionHeaderCount: 7
-# CHECK-NEXT:    StringTableSectionIndex: 6
+# CHECK-NEXT:   SectionHeaderCount: 9
+# CHECK-NEXT:   StringTableSectionIndex: 8
 # CHECK-NEXT: }
 # CHECK-NEXT: Sections [
 # CHECK-NEXT:   Section {
 # CHECK-NEXT:     Index: 0
 # CHECK-NEXT:     Name:  (0)
 # CHECK-NEXT:     Type: SHT_NULL (0x0)
 # CHECK-NEXT:     Flags [ (0x0)
 # CHECK-NEXT:     ]
 # CHECK-NEXT:     Address: 0x0
 # CHECK-NEXT:     Offset: 0x0
 # CHECK-NEXT:     Size: 0
 # CHECK-NEXT:     Link: 0
 # CHECK-NEXT:     Info: 0
 # CHECK-NEXT:     AddressAlignment: 0
 # CHECK-NEXT:     EntrySize: 0
 # CHECK-NEXT:     SectionData (
-# CHECK:          )
+# CHECK-NEXT:     )
 # CHECK-NEXT:   }
 # CHECK-NEXT:   Section {
 # CHECK-NEXT:     Index: 1
-# CHECK-NEXT:     Name: .text
+# CHECK-NEXT:     Name: .text (1)
 # CHECK-NEXT:     Type: SHT_PROGBITS (0x1)
 # CHECK-NEXT:     Flags [ (0x6)
 # CHECK-NEXT:       SHF_ALLOC (0x2)
 # CHECK-NEXT:       SHF_EXECINSTR (0x4)
 # CHECK-NEXT:     ]
 # CHECK-NEXT:     Address: 0x10010000
-# CHECK-NEXT:     Offset: 0x1000
+# CHECK-NEXT:     Offset: 0x10000
 # CHECK-NEXT:     Size: 12
 # CHECK-NEXT:     Link: 0
 # CHECK-NEXT:     Info: 0
 # CHECK-NEXT:     AddressAlignment: 4
 # CHECK-NEXT:     EntrySize: 0
 # CHECK-NEXT:     SectionData (
 # CHECK:          )
 # CHECK-NEXT:   }
 # CHECK-NEXT:   Section {
 # CHECK-NEXT:     Index: 2
-# CHECK-NEXT:     Name: .data
+# CHECK-NEXT:     Name: .data (45)
 # CHECK-NEXT:     Type: SHT_PROGBITS (0x1)
 # CHECK-NEXT:     Flags [ (0x3)
 # CHECK-NEXT:       SHF_ALLOC (0x2)
 # CHECK-NEXT:       SHF_WRITE (0x1)
 # CHECK-NEXT:     ]
 # CHECK-NEXT:     Address: 0x1001000C
 # CHECK-NEXT:     Offset: 0x1000C
 # CHECK-NEXT:     Size: 0
 # CHECK-NEXT:     Link: 0
 # CHECK-NEXT:     Info: 0
 # CHECK-NEXT:     AddressAlignment: 4
 # CHECK-NEXT:     EntrySize: 0
 # CHECK-NEXT:     SectionData (
 # CHECK:          )
 # CHECK-NEXT:   }
 # CHECK-NEXT:   Section {
 # CHECK-NEXT:     Index: 3
-# CHECK-NEXT:     Name: .opd
+# CHECK-NEXT:     Name: .toc (24)
 # CHECK-NEXT:     Type: SHT_PROGBITS (0x1)
 # CHECK-NEXT:     Flags [ (0x3)
 # CHECK-NEXT:       SHF_ALLOC (0x2)
 # CHECK-NEXT:       SHF_WRITE (0x1)
 # CHECK-NEXT:     ]
 # CHECK-NEXT:     Address: 0x10020000
-# CHECK-NEXT:     Offset: 0x2000
+# CHECK-NEXT:     Offset: 0x20000
+# CHECK-NEXT:     Size: 32
+# CHECK-NEXT:     Link: 0
+# CHECK-NEXT:     Info: 0
+# CHECK-NEXT:     AddressAlignment: 1
+# CHECK-NEXT:     EntrySize: 0
+# CHECK-NEXT:     SectionData (
+# CHECK-NEXT:      0000: 00000000 0000002D 00000000 00000056 |.......-.......V|
+# CHECK-NEXT:      0010: 00000000 00000048 00000000 00000018 |.......H........|
+# CHECK-NEXT:     )
+# CHECK-NEXT:   }
+# CHECK-NEXT:   Section {
+# CHECK-NEXT:     Index: 4
+# CHECK-NEXT:     Name: .toc1 (51)
+# CHECK-NEXT:     Type: SHT_PROGBITS (0x1)
+# CHECK-NEXT:     Flags [ (0x3)
+# CHECK-NEXT:       SHF_ALLOC (0x2)
+# CHECK-NEXT:       SHF_WRITE (0x1)
+# CHECK-NEXT:     ]
+# CHECK-NEXT:     Address: 0x10020020
+# CHECK-NEXT:     Offset: 0x20020
+# CHECK-NEXT:     Size: 32
+# CHECK-NEXT:     Link: 0
+# CHECK-NEXT:     Info: 0
+# CHECK-NEXT:     AddressAlignment: 1
+# CHECK-NEXT:     EntrySize: 0
+# CHECK-NEXT:     SectionData (
+# CHECK-NEXT:       0000: 00000000 00000016 00000000 00000025  |...............%|
+# CHECK-NEXT:       0010: 00000000 00000059 00000000 0000002F  |.......Y......./|
+# CHECK-NEXT:     )
+# CHECK-NEXT:   }
+# CHECK-NEXT:   Section {
+# CHECK-NEXT:     Index: 5
+# CHECK-NEXT:     Name: .opd (19)
+# CHECK-NEXT:     Type: SHT_PROGBITS (0x1)
+# CHECK-NEXT:     Flags [ (0x3)
+# CHECK-NEXT:       SHF_ALLOC (0x2)
+# CHECK-NEXT:       SHF_WRITE (0x1)
+# CHECK-NEXT:     ]
+# CHECK-NEXT:     Address: 0x10020040
+# CHECK-NEXT:     Offset: 0x20040
 # CHECK-NEXT:     Size: 24
 # CHECK-NEXT:     Link: 0
 # CHECK-NEXT:     Info: 0
 # CHECK-NEXT:     AddressAlignment: 1
 # CHECK-NEXT:     EntrySize: 0
 # CHECK-NEXT:     SectionData (
-# CHECK-NEXT:      0000: 00000000 10010000 00000000 00000000  |................|
-# CHECK-NEXT:      0010: 00000000 00000000                    |........|
+# CHECK-NEXT:       0000: 00000000 10010000 00000000 00000000 |................|
+# CHECK-NEXT:       0010: 00000000 00000000                    |........|
 # CHECK-NEXT:     )
 # CHECK-NEXT:   }
 # CHECK-NEXT:   Section {
-# CHECK-NEXT:     Index: 4
-# CHECK-NEXT:     Name: .bss
+# CHECK-NEXT:     Index: 6
+# CHECK-NEXT:     Name: .bss (14)
 # CHECK-NEXT:     Type: SHT_NOBITS (0x8)
 # CHECK-NEXT:     Flags [ (0x3)
 # CHECK-NEXT:       SHF_ALLOC (0x2)
 # CHECK-NEXT:       SHF_WRITE (0x1)
 # CHECK-NEXT:     ]
-# CHECK-NEXT:     Address: 0x10020018
-# CHECK-NEXT:     Offset: 0x20018
+# CHECK-NEXT:     Address: 0x10020058
+# CHECK-NEXT:     Offset: 0x20058
 # CHECK-NEXT:     Size: 0
 # CHECK-NEXT:     Link: 0
 # CHECK-NEXT:     Info: 0
 # CHECK-NEXT:     AddressAlignment: 4
 # CHECK-NEXT:     EntrySize: 0
 # CHECK-NEXT:   }
 # CHECK-NEXT:   Section {
-# CHECK-NEXT:     Index: 5
-# CHECK-NEXT:     Name: .symtab
-# CHECK-NEXT:     Type: SHT_SYMTAB
-# CHECK-NEXT:     Flags [
+# CHECK-NEXT:     Index: 7
+# CHECK-NEXT:     Name: .symtab (37)
+# CHECK-NEXT:     Type: SHT_SYMTAB (0x2)
+# CHECK-NEXT:     Flags [ (0x0)
 # CHECK-NEXT:     ]
 # CHECK-NEXT:     Address: 0x0
-# CHECK-NEXT:     Offset: 0x20018
+# CHECK-NEXT:     Offset: 0x20058
 # CHECK-NEXT:     Size: 48
-# CHECK-NEXT:     Link: 6
+# CHECK-NEXT:     Link: 8
 # CHECK-NEXT:     Info: 1
 # CHECK-NEXT:     AddressAlignment: 8
 # CHECK-NEXT:     EntrySize: 24
 # CHECK-NEXT:     SectionData (
 # CHECK:          )
 # CHECK-NEXT:   }
 # CHECK-NEXT:   Section {
-# CHECK-NEXT:     Index: 6
-# CHECK-NEXT:     Name: .strtab
+# CHECK-NEXT:     Index: 8
+# CHECK-NEXT:     Name: .strtab (29)
 # CHECK-NEXT:     Type: SHT_STRTAB (0x3)
 # CHECK-NEXT:     Flags [ (0x0)
 # CHECK-NEXT:     ]
 # CHECK-NEXT:     Address: 0x0
-# CHECK-NEXT:     Offset: 0x20048
-# CHECK-NEXT:     Size: 46
+# CHECK-NEXT:     Offset: 0x20088
+# CHECK-NEXT:     Size: 57
 # CHECK-NEXT:     Link: 0
 # CHECK-NEXT:     Info: 0
 # CHECK-NEXT:     AddressAlignment: 1
 # CHECK-NEXT:     EntrySize: 0
 # CHECK-NEXT:     SectionData (
 # CHECK:          )
 # CHECK-NEXT:   }
 # CHECK-NEXT: ]
 # CHECK-NEXT: ProgramHeaders [
-# CHECK-NEXT:  ProgramHeader {
-# CHECK-NEXT:    Type: PT_LOAD (0x1)
-# CHECK-NEXT:    Offset: 0x0
-# CHECK-NEXT:    VirtualAddress: 0x10000000
-# CHECK-NEXT:    PhysicalAddress: 0x10000000
-# CHECK-NEXT:    FileSize: 232
-# CHECK-NEXT:    MemSize: 232
-# CHECK-NEXT:    Flags [
-# CHECK-NEXT:      PF_R
-# CHECK-NEXT:    ]
-# CHECK-NEXT:    Alignment: 65536
-# CHECK-NEXT:  }
-# CHECK-NEXT:  ProgramHeader {
-# CHECK-NEXT:    Type: PT_LOAD (0x1)
-# CHECK-NEXT:    Offset: 0x1000
-# CHECK-NEXT:    VirtualAddress: 0x10010000
-# CHECK-NEXT:    PhysicalAddress: 0x10010000
-# CHECK-NEXT:    FileSize: 12
-# CHECK-NEXT:    MemSize: 12
-# CHECK-NEXT:    Flags [ (0x5)
-# CHECK-NEXT:      PF_R (0x4)
-# CHECK-NEXT:      PF_X (0x1)
-# CHECK-NEXT:    ]
-# CHECK-NEXT:    Alignment: 65536
-# CHECK-NEXT:  }
-# CHECK-NEXT:  ProgramHeader {
-# CHECK-NEXT:    Type: PT_LOAD (0x1)
-# CHECK-NEXT:    Offset: 0x2000
-# CHECK-NEXT:    VirtualAddress: 0x10020000
-# CHECK-NEXT:    PhysicalAddress: 0x10020000
-# CHECK-NEXT:    FileSize: 24
-# CHECK-NEXT:    MemSize: 24
-# CHECK-NEXT:    Flags [ (0x6)
-# CHECK-NEXT:      PF_R (0x4)
-# CHECK-NEXT:      PF_W (0x2)
-# CHECK-NEXT:    ]
-# CHECK-NEXT:    Alignment: 65536
-# CHECK-NEXT:  }
-# CHECK-NEXT:]
+# CHECK-NEXT:   ProgramHeader {
+# CHECK-NEXT:     Type: PT_LOAD (0x1)
+# CHECK-NEXT:     Offset: 0x0
+# CHECK-NEXT:     VirtualAddress: 0x10000000
+# CHECK-NEXT:     PhysicalAddress: 0x10000000
+# CHECK-NEXT:     FileSize: 232
+# CHECK-NEXT:     MemSize: 232
+# CHECK-NEXT:     Flags [ (0x4)
+# CHECK-NEXT:       PF_R (0x4)
+# CHECK-NEXT:     ]
+# CHECK-NEXT:     Alignment: 65536
+# CHECK-NEXT:   }
+# CHECK-NEXT:   ProgramHeader {
+# CHECK-NEXT:     Type: PT_LOAD (0x1)
+# CHECK-NEXT:     Offset: 0x10000
+# CHECK-NEXT:     VirtualAddress: 0x10010000
+# CHECK-NEXT:     PhysicalAddress: 0x10010000
+# CHECK-NEXT:     FileSize: 12
+# CHECK-NEXT:     MemSize: 12
+# CHECK-NEXT:     Flags [ (0x5)
+# CHECK-NEXT:       PF_R (0x4)
+# CHECK-NEXT:       PF_X (0x1)
+# CHECK-NEXT:     ]
+# CHECK-NEXT:     Alignment: 65536
+# CHECK-NEXT:   }
+# CHECK-NEXT:   ProgramHeader {
+# CHECK-NEXT:     Type: PT_LOAD (0x1)
+# CHECK-NEXT:     Offset: 0x20000
+# CHECK-NEXT:     VirtualAddress: 0x10020000
+# CHECK-NEXT:     PhysicalAddress: 0x10020000
+# CHECK-NEXT:     FileSize: 88
+# CHECK-NEXT:     MemSize: 88
+# CHECK-NEXT:     Flags [ (0x6)
+# CHECK-NEXT:       PF_R (0x4)
+# CHECK-NEXT:       PF_W (0x2)
+# CHECK-NEXT:     ]
+# CHECK-NEXT:     Alignment: 65536
+# CHECK-NEXT:   }
+# CHECK-NEXT: ]
+