COFF: Merge .idata, .didat and .edata into .rdata by default.

[pcc]

Summary

This saves a little space and matches what link.exe does.

Tested using the chromium Windows trybots:
https://chromium-review.googlesource.com/c/chromium/src/+/1014784

[pcc]

lld/test/COFF/pdb-publics-import.test

39 ↗

(On Diff #142828)

This is the only weird thing in the test updates.

@zturner I'm not sure if this is a pre-existing problem with our pdb emission?

[pcc]

lld/test/COFF/pdb-publics-import.test

39 ↗

(On Diff #142828)

Actually looks like it was an off-by-one error in llvm-pdbutil -- fix incoming

[zturner]

lld/test/COFF/pdb-publics-import.test

39 ↗

(On Diff #142828)

I don't know if this is a pre-existing problem, it may be due to the patch.

Short version: can you run llvm-pdbutil dump -section-headers on the new PDB? My guess is that you will see less than 2 sections here. At the very least, comparing the output of the tool with this option before and after your patch should provide a clue.

Long version: The printing code here is using the section index (the 2 from 0002:0000) as an index into a list of section names.

It generates this list of section names by looking in the "optional debug stream array" in the DBI stream of the PDB. Essentially this is an array of stream indices, where each array index has a special hardcoded meaning. One of these (at index DbgHeaderType::SectionHdr, whose value is I think 5) represents the index of a stream which contains a list of section headers.

So the printing code looks at index 5 in this array, loads the specified stream, gets the list of section header, and then reads the second one.

If there are less than 2 though, it will print ??? for the section name.

So my suspicion is that something about this patch is causing only 1 entry to be written here instead of 2.

In LLD, this is set on this line:

// Add COFF section header stream.
ExitOnErr(
    DbiBuilder.addDbgStream(pdb::DbgHeaderType::SectionHdr, SectionTable));

So check there and see that SectionTable contains the proper list of sections.

39 ↗

(On Diff #142828)

Oh cool. Could be that too :)

[pcc]

• Revert change to pdb-publics-import.test

[ruiu]

LGTM

COFF/Driver.cpp

void LinkerDriver::link(ArrayRef<const char *> ArgsArr) {
   // Handle /failifmismatch
   for (auto *Arg : Args.filtered(OPT_failifmismatch))
     checkFailIfMismatch(Arg->getValue());
 
   // Handle /merge
   for (auto *Arg : Args.filtered(OPT_merge))
     parseMerge(Arg->getValue());
 
+  // Add default section merging rules after user rules. User rules take
+  // precedence, but we will emit a warning if there is a conflict.
+  parseMerge(".idata=.rdata");
+  parseMerge(".didat=.rdata");
+  parseMerge(".edata=.rdata");
+
   // Handle /section
   for (auto *Arg : Args.filtered(OPT_section))
     parseSection(Arg->getValue());
 
   // Handle /aligncomm
   for (auto *Arg : Args.filtered(OPT_aligncomm))
     parseAligncomm(Arg->getValue());
 

test/COFF/arm64-relocs-imports.test

 # AFTER:  14000106c:      c0 03 5f d6     ret
 # AFTER:  140001070:      10 20 00 40     <unknown>
 # AFTER:  140001074:      01 00 00 00     <unknown>
 # AFTER:  140001078:      09 20 00 00     <unknown>
 # AFTER:  14000107c:      09 00 00 00     <unknown>
 # AFTER:  140001080:      00 20 0e 91     add     x0, x0, #904
 # AFTER:  140001084:      00 04 40 91     add     x0, x0, #1, lsl #12
 # AFTER:  140001088:      00 c4 41 f9     ldr     x0, [x0, #904]
-# AFTER:  14000108c:      10 00 00 d0     adrp    x16, #8192
-# AFTER:  140001090:      10 1e 40 f9     ldr     x16, [x16, #56]
+# AFTER:  14000108c:      10 00 00 b0     adrp    x16, #4096
+# AFTER:  140001090:      10 2a 40 f9     ldr     x16, [x16, #80]
 # AFTER:  140001094:      00 02 1f d6     br      x16
 
 --- !COFF
 header:
   Machine:         IMAGE_FILE_MACHINE_ARM64
   Characteristics: [  ]
 sections:
   - Name:            .text

test/COFF/delayimports-armnt.yaml

 # REQUIRES: arm
 # RUN: yaml2obj < %s > %t.obj
 # RUN: lld-link %t.obj %p/Inputs/library.lib /subsystem:console \
 # RUN:   /entry:mainCRTStartup /alternatename:__delayLoadHelper2=mainCRTStartup \
 # RUN:   /delayload:library.dll /out:%t.exe
 # RUN: llvm-readobj -coff-imports %t.exe | FileCheck -check-prefix=IMPORT %s
 # RUN: llvm-readobj -coff-basereloc %t.exe | FileCheck -check-prefix=BASEREL %s
 # RUN: llvm-objdump -d %t.exe | FileCheck -check-prefix=DISASM %s
 
 # IMPORT:      Format: COFF-ARM
 # IMPORT-NEXT: Arch: thumb
 # IMPORT-NEXT: AddressSize: 32bit
 # IMPORT-NEXT: DelayImport {
 # IMPORT-NEXT:   Name: library.dll
 # IMPORT-NEXT:   Attributes: 0x1
-# IMPORT-NEXT:   ModuleHandle: 0x2000
-# IMPORT-NEXT:   ImportAddressTable: 0x2008
-# IMPORT-NEXT:   ImportNameTable: 0x3040
+# IMPORT-NEXT:   ModuleHandle: 0x3000
+# IMPORT-NEXT:   ImportAddressTable: 0x3008
+# IMPORT-NEXT:   ImportNameTable: 0x2040
 # IMPORT-NEXT:   BoundDelayImportTable: 0x0
 # IMPORT-NEXT:   UnloadDelayImportTable: 0x0
 # IMPORT-NEXT:   Import {
 # IMPORT-NEXT:     Symbol: function (0)
 # IMPORT-NEXT:     Address: 0x401019
 # IMPORT-NEXT:   }
 # IMPORT-NEXT: }
 #

 # BASEREL-NEXT:     Address: 0x1018
 # BASEREL-NEXT:   }
 # BASEREL-NEXT:   Entry {
 # BASEREL-NEXT:     Type: ARM_MOV32(T)
 # BASEREL-NEXT:     Address: 0x102E
 # BASEREL-NEXT:   }
 # BASEREL-NEXT:   Entry {
 # BASEREL-NEXT:     Type: HIGHLOW
-# BASEREL-NEXT:     Address: 0x2008
+# BASEREL-NEXT:     Address: 0x3008
 # BASEREL-NEXT:   }
 # BASEREL-NEXT:   Entry {
 # BASEREL-NEXT:     Type: ABSOLUTE
-# BASEREL-NEXT:     Address: 0x2000
+# BASEREL-NEXT:     Address: 0x3000
 # BASEREL-NEXT:   }
 # BASEREL-NEXT: ]
 #
-# DISASM:      401018:       42 f2 08 0c     movw r12, #8200
+# DISASM:      401018:       43 f2 08 0c     movw r12, #12296
 # DISASM-NEXT: 40101c:       c0 f2 40 0c     movt    r12, #64
 # DISASM-NEXT: 401020:       2d e9 0f 48     push.w  {r0, r1, r2, r3, r11, lr}
 # DISASM-NEXT: 401024:       0d f2 10 0b     addw    r11, sp, #16
 # DISASM-NEXT: 401028:       2d ed 10 0b     vpush   {d0, d1, d2, d3, d4, d5, d6, d7}
 # DISASM-NEXT: 40102c:       61 46           mov     r1, r12
-# DISASM-NEXT: 40102e:       43 f2 00 00     movw r0, #12288
+# DISASM-NEXT: 40102e:       42 f2 00 00     movw r0, #8192
 # DISASM-NEXT: 401032:       c0 f2 40 00     movt    r0, #64
 # DISASM-NEXT: 401036:       ff f7 e3 ff     bl      #-58
 # DISASM-NEXT: 40103a:       84 46           mov     r12, r0
 # DISASM-NEXT: 40103c:       bd ec 10 0b     vpop    {d0, d1, d2, d3, d4, d5, d6, d7}
 # DISASM-NEXT: 401040:       bd e8 0f 48     pop.w   {r0, r1, r2, r3, r11, lr}
 # DISASM-NEXT: 401044:       60 47           bx      r12
 
 --- !COFF

test/COFF/delayimports.test

 # RUN: lld-link /out:%t.exe /entry:main /subsystem:console \
 # RUN:   %p/Inputs/hello64.obj %p/Inputs/std64.lib /delayload:STD64.DLL \
 # RUN:   /alternatename:__delayLoadHelper2=main
 # RUN: llvm-readobj -coff-imports %t.exe | FileCheck -check-prefix=IMPORT %s
 # RUN: llvm-readobj -coff-basereloc %t.exe | FileCheck -check-prefix=BASEREL %s
 
 IMPORT:      DelayImport {
 IMPORT-NEXT:   Name: std64.dll
 IMPORT-NEXT:   Attributes: 0x1
-IMPORT-NEXT:   ModuleHandle: 0x2018
-IMPORT-NEXT:   ImportAddressTable: 0x2020
-IMPORT-NEXT:   ImportNameTable: 0x3040
+IMPORT-NEXT:   ModuleHandle: 0x3018
+IMPORT-NEXT:   ImportAddressTable: 0x3020
+IMPORT-NEXT:   ImportNameTable: 0x2040
 IMPORT-NEXT:   BoundDelayImportTable: 0x0
 IMPORT-NEXT:   UnloadDelayImportTable: 0x0
 IMPORT-NEXT:   Import {
 IMPORT-NEXT:     Symbol: ExitProcess (0)
 IMPORT-NEXT:     Address: 0x140001066
 IMPORT-NEXT:   }
 IMPORT-NEXT:   Import {
 IMPORT-NEXT:     Symbol:  (50)
 IMPORT-NEXT:     Address: 0x1400010BD
 IMPORT-NEXT:   }
 IMPORT-NEXT:   Import {
 IMPORT-NEXT:     Symbol: MessageBoxA (0)
 IMPORT-NEXT:     Address: 0x140001114
 IMPORT-NEXT:   }
 IMPORT-NEXT: }
 
 BASEREL:      BaseReloc [
 BASEREL-NEXT:   Entry {
 BASEREL-NEXT:     Type: DIR64
-BASEREL-NEXT:     Address: 0x2020
+BASEREL-NEXT:     Address: 0x3020
 BASEREL-NEXT:   }
 BASEREL-NEXT:   Entry {
 BASEREL-NEXT:     Type: DIR64
-BASEREL-NEXT:     Address: 0x2028
+BASEREL-NEXT:     Address: 0x3028
 BASEREL-NEXT:   }
 BASEREL-NEXT:   Entry {
 BASEREL-NEXT:     Type: DIR64
-BASEREL-NEXT:     Address: 0x2030
+BASEREL-NEXT:     Address: 0x3030
 BASEREL-NEXT:   }

test/COFF/delayimports32.test

 IMPORT:      Format: COFF-i386
 IMPORT-NEXT: Arch: i386
 IMPORT-NEXT: AddressSize: 32bit
 IMPORT-NEXT: DelayImport {
 IMPORT-NEXT:   Name: std32.dll
 IMPORT-NEXT:   Attributes: 0x1
 IMPORT-NEXT:   ModuleHandle: 0x3018
 IMPORT-NEXT:   ImportAddressTable: 0x3020
-IMPORT-NEXT:   ImportNameTable: 0x4040
+IMPORT-NEXT:   ImportNameTable: 0x20DC
 IMPORT-NEXT:   BoundDelayImportTable: 0x0
 IMPORT-NEXT:   UnloadDelayImportTable: 0x0
 IMPORT-NEXT:   Import {
 IMPORT-NEXT:     Symbol: ExitProcess (0)
 IMPORT-NEXT:     Address: 0x401029
 IMPORT-NEXT:   }
 IMPORT-NEXT:   Import {
 IMPORT-NEXT:     Symbol: MessageBoxA (0)

 BASEREL-NEXT:   }
 BASEREL-NEXT:   Entry {
 BASEREL-NEXT:     Type: HIGHLOW
 BASEREL-NEXT:     Address: 0x3024
 BASEREL-NEXT:   }
 BASEREL-NEXT: ]
 
 DISASM:      102b:      68 20 30 40 00  pushl   $4206624
-DISASM-NEXT: 1030:      68 00 40 40 00  pushl   $4210688
+DISASM-NEXT: 1030:      68 9c 20 40 00  pushl   $4202652
 DISASM-NEXT: 1035:      e8 c6 ff ff ff  calll   -58 <.text>
 DISASM-NEXT: 103a:      5a      popl    %edx
 DISASM-NEXT: 103b:      59      popl    %ecx
 DISASM-NEXT: 103c:      ff e0   jmpl    *%eax
 DISASM-NEXT: 103e:      51      pushl   %ecx
 DISASM-NEXT: 103f:      52      pushl   %edx
 DISASM-NEXT: 1040:      68 24 30 40 00  pushl   $4206628
-DISASM-NEXT: 1045:      68 00 40 40 00  pushl   $4210688
+DISASM-NEXT: 1045:      68 9c 20 40 00  pushl   $4202652
 DISASM-NEXT: 104a:      e8 b1 ff ff ff  calll   -79 <.text>
 DISASM-NEXT: 104f:      5a      popl    %edx
 DISASM-NEXT: 1050:      59      popl    %ecx
 DISASM-NEXT: 1051:      ff e0   jmpl    *%eax

test/COFF/export-armnt.yaml

 # REQUIRES: arm
 
 # RUN: yaml2obj < %s > %t.obj
 #
 # RUN: lld-link /out:%t.dll /dll %t.obj /export:exportfn1 /export:exportfn2 /export:exportdata,data
 # RUN: llvm-objdump -p %t.dll | FileCheck %s
 
 # CHECK:      Export Table:
 # CHECK:      DLL name: export-armnt.yaml.tmp.dll
 # CHECK:      Ordinal      RVA  Name
 # CHECK-NEXT:       0        0
-# CHECK-NEXT:       1   0x2000  exportdata
+# CHECK-NEXT:       1   0x3000  exportdata
 # CHECK-NEXT:       2   0x1005  exportfn1
 # CHECK-NEXT:       3   0x1009  exportfn2
 # CHECK-NEXT:       4   0x1009  exportfn3
 
 --- !COFF
 header:
   Machine:         IMAGE_FILE_MACHINE_ARMNT
   Characteristics: []

test/COFF/hello32.test

 HEADER-NEXT:   SizeOfStackReserve: 1048576
 HEADER-NEXT:   SizeOfStackCommit: 4096
 HEADER-NEXT:   SizeOfHeapReserve: 1048576
 HEADER-NEXT:   SizeOfHeapCommit: 4096
 HEADER-NEXT:   NumberOfRvaAndSize: 16
 HEADER-NEXT:   DataDirectory {
 HEADER-NEXT:     ExportTableRVA: 0x0
 HEADER-NEXT:     ExportTableSize: 0x0
-HEADER-NEXT:     ImportTableRVA: 0x3000
+HEADER-NEXT:     ImportTableRVA: 0x2000
 HEADER-NEXT:     ImportTableSize: 0x28
 HEADER-NEXT:     ResourceTableRVA: 0x0
 HEADER-NEXT:     ResourceTableSize: 0x0
 HEADER-NEXT:     ExceptionTableRVA: 0x0
 HEADER-NEXT:     ExceptionTableSize: 0x0
 HEADER-NEXT:     CertificateTableRVA: 0x0
 HEADER-NEXT:     CertificateTableSize: 0x0
 HEADER-NEXT:     BaseRelocationTableRVA: 0x4000
 HEADER-NEXT:     BaseRelocationTableSize: 0x10
 HEADER-NEXT:     DebugRVA: 0x0
 HEADER-NEXT:     DebugSize: 0x0
 HEADER-NEXT:     ArchitectureRVA: 0x0
 HEADER-NEXT:     ArchitectureSize: 0x0
 HEADER-NEXT:     GlobalPtrRVA: 0x0
 HEADER-NEXT:     GlobalPtrSize: 0x0
 HEADER-NEXT:     TLSTableRVA: 0x0
 HEADER-NEXT:     TLSTableSize: 0x0
 HEADER-NEXT:     LoadConfigTableRVA: 0x0
 HEADER-NEXT:     LoadConfigTableSize: 0x0
 HEADER-NEXT:     BoundImportRVA: 0x0
 HEADER-NEXT:     BoundImportSize: 0x0
-HEADER-NEXT:     IATRVA: 0x3034
+HEADER-NEXT:     IATRVA: 0x2034
 HEADER-NEXT:     IATSize: 0xC
 HEADER-NEXT:     DelayImportDescriptorRVA: 0x0
 HEADER-NEXT:     DelayImportDescriptorSize: 0x0
 HEADER-NEXT:     CLRRuntimeHeaderRVA: 0x0
 HEADER-NEXT:     CLRRuntimeHeaderSize: 0x0
 HEADER-NEXT:     ReservedRVA: 0x0
 HEADER-NEXT:     ReservedSize: 0x0
 HEADER-NEXT:   }

 HEADER-NEXT:   AddressOfNewExeHeader: 120
 HEADER-NEXT: }
 
 IMPORTS: Format: COFF-i386
 IMPORTS: Arch: i386
 IMPORTS: AddressSize: 32bit
 IMPORTS: Import {
 IMPORTS:   Name: std32.dll
-IMPORTS:   ImportLookupTableRVA: 0x3028
-IMPORTS:   ImportAddressTableRVA: 0x3034
+IMPORTS:   ImportLookupTableRVA: 0x2028
+IMPORTS:   ImportAddressTableRVA: 0x2034
 IMPORTS:   Symbol: ExitProcess (0)
 IMPORTS:   Symbol: MessageBoxA (1)
 IMPORTS: }
 
 BASEREL: BaseReloc [
 BASEREL:   Entry {
 BASEREL:     Type: HIGHLOW
 BASEREL:     Address: 0x1005
 BASEREL:   }
 BASEREL:   Entry {
 BASEREL:     Type: HIGHLOW
 BASEREL:     Address: 0x100C
 BASEREL:   }
 BASEREL: ]

test/COFF/imports.test

 # RUN:   %p/Inputs/hello64.obj %p/Inputs/std64.lib /include:ExitProcess
 # RUN: llvm-objdump -d %t.exe | FileCheck -check-prefix=TEXT %s
 # RUN: llvm-readobj -coff-imports %t.exe | FileCheck -check-prefix=IMPORT %s
 
 TEXT: Disassembly of section .text:
 TEXT-NEXT: .text:
 TEXT-NEXT: subq    $40, %rsp
 TEXT-NEXT: movq    $0, %rcx
-TEXT-NEXT: leaq    4084(%rip), %rdx
-TEXT-NEXT: leaq    4071(%rip), %r8
+TEXT-NEXT: leaq    8180(%rip), %rdx
+TEXT-NEXT: leaq    8167(%rip), %r8
 TEXT-NEXT: movl    $0, %r9d
 TEXT-NEXT: callq   60
 TEXT-NEXT: movl    $0, %ecx
 TEXT-NEXT: callq   18
 TEXT-NEXT: callq   29
-TEXT:      jmpq    *8194(%rip)
-TEXT:      jmpq    *8186(%rip)
-TEXT:      jmpq    *8178(%rip)
+TEXT:      jmpq    *4098(%rip)
+TEXT:      jmpq    *4090(%rip)
+TEXT:      jmpq    *4082(%rip)
 
 IMPORT:      Import {
 IMPORT-NEXT:   Name: std64.dll
-IMPORT-NEXT:   ImportLookupTableRVA: 0x3028
-IMPORT-NEXT:   ImportAddressTableRVA: 0x3048
+IMPORT-NEXT:   ImportLookupTableRVA: 0x2028
+IMPORT-NEXT:   ImportAddressTableRVA: 0x2048
 IMPORT-NEXT:   Symbol: ExitProcess (0)
 IMPORT-NEXT:   Symbol:  (50)
 IMPORT-NEXT:   Symbol: MessageBoxA (1)
 IMPORT-NEXT: }

test/COFF/pdb-publics-import.test

 CHECK-NEXT:    Records
 CHECK-NEXT:      112 | S_PUB32 [size = 20] `main`
 CHECK-NEXT:            flags = function, addr = 0001:0000
 CHECK-NEXT:       64 | S_PUB32 [size = 24] `exportfn1`
 CHECK-NEXT:            flags = function, addr = 0001:0016
 CHECK-NEXT:       88 | S_PUB32 [size = 24] `exportfn2`
 CHECK-NEXT:            flags = function, addr = 0001:0032
 CHECK-NEXT:       32 | S_PUB32 [size = 32] `__imp_exportfn2`
-CHECK-NEXT:            flags = none, addr = 0003:0072
+CHECK-NEXT:            flags = none, addr = 0002:0240
 CHECK-NEXT:        0 | S_PUB32 [size = 32] `__imp_exportfn1`
-CHECK-NEXT:            flags = none, addr = 0003:0064
+CHECK-NEXT:            flags = none, addr = 0002:0232
 
 CHECK:                         Section Contributions
 CHECK-NEXT: ============================================================
     main
 CHECK-NEXT:   SC[.text]   | mod = 0, 0001:0000, size = 8, data crc = 0, reloc crc = 0
 CHECK-NEXT:                 IMAGE_SCN_CNT_CODE | IMAGE_SCN_ALIGN_4BYTES | IMAGE_SCN_MEM_EXECUTE |
 CHECK-NEXT:                 IMAGE_SCN_MEM_READ
     exportfn1 thunk

test/COFF/symtab.test

 # CHECK-NEXT:     BaseType: Null (0x0)
 # CHECK-NEXT:     ComplexType: Null (0x0)
 # CHECK-NEXT:     StorageClass: External (0x2)
 # CHECK-NEXT:     AuxSymbolCount: 0
 # CHECK-NEXT:   }
 # CHECK-NEXT:   Symbol {
 # CHECK-NEXT:     Name: message
 # CHECK-NEXT:     Value: 6
-# CHECK-NEXT:     Section: .text2 (5)
+# CHECK-NEXT:     Section: .text2
 # CHECK-NEXT:     BaseType: Null (0x0)
 # CHECK-NEXT:     ComplexType: Null (0x0)
 # CHECK-NEXT:     StorageClass: Static (0x3)
 # CHECK-NEXT:     AuxSymbolCount: 0
 # CHECK-NEXT:   }
 # CHECK-NEXT:   Symbol {
 # CHECK-NEXT:     Name: main
 # CHECK-NEXT:     Value: 0
 # CHECK-NEXT:     Section: .text (1)
 # CHECK-NEXT:     BaseType: Null (0x0)
 # CHECK-NEXT:     ComplexType: Null (0x0)
 # CHECK-NEXT:     StorageClass: External (0x2)
 # CHECK-NEXT:     AuxSymbolCount: 0
 # CHECK-NEXT:   }
 # CHECK-NEXT:   Symbol {
 # CHECK-NEXT:     Name: caption
 # CHECK-NEXT:     Value: 0
-# CHECK-NEXT:     Section: .text2 (5)
+# CHECK-NEXT:     Section: .text2
 # CHECK-NEXT:     BaseType: Null (0x0)
 # CHECK-NEXT:     ComplexType: Null (0x0)
 # CHECK-NEXT:     StorageClass: Static (0x3)
 # CHECK-NEXT:     AuxSymbolCount: 0
 # CHECK-NEXT:   }
 # CHECK-NEXT:   Symbol {
 # CHECK-NEXT:     Name: abs_symbol
 # CHECK-NEXT:     Value: 2662186735