maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

DebugInfo: Hash DW_OP_convert in loclists when using Split DWARF 

Originally committed in: 1ced28cbe7
            Reverted in: f75301d16d

(reverted due to tests failing on non-linux/x86 targets, tests have since been
generalized and specialized... since Split DWARF isn't supported on non-elf
targets anyway and we have no way to run on "whatever elf target is available"
so they fail on MacOS without an explicit target triple)

This code was incorrectly emitting extra bytes into arbitrary parts of
the object file when it was meant to be hashing them to compute the DWO
ID.

Follow-up patch(es) will refactor this API somewhat to make such bugs
harder to introduce, hopefully.
This commit is contained in:
David Blaikie 2020-02-04 19:25:47 -08:00
parent def55a8efd
commit ec50e10db4
3 changed files with 20 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
llvm/lib/CodeGen/AsmPrinter/ByteStreamer.h
View File

@ -50,7 +50,7 @@ public:
}
void EmitULEB128(uint64_t DWord, const Twine &Comment, unsigned PadTo) override {
AP.OutStreamer->AddComment(Comment);
AP.EmitULEB128(DWord);
AP.EmitULEB128(DWord, nullptr, PadTo);
}
};

2
llvm/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
View File

@ -2241,7 +2241,7 @@ void DwarfDebug::emitDebugLocEntry(ByteStreamer &Streamer,
uint64_t Offset =
CU->ExprRefedBaseTypes[Op.getRawOperand(I)].Die->getOffset();
assert(Offset < (1ULL << (ULEB128PadSize * 7)) && "Offset wont fit");
Asm->EmitULEB128(Offset, nullptr, ULEB128PadSize);
Streamer.EmitULEB128(Offset, "", ULEB128PadSize);
// Make sure comments stay aligned.
for (unsigned J = 0; J < ULEB128PadSize; ++J)
if (Comment != End)

19
llvm/test/DebugInfo/X86/convert-loclist.ll
View File

@ -1,7 +1,24 @@
; RUN: %llc_dwarf -filetype=obj < %s \
; RUN: | llvm-dwarfdump -debug-info -debug-loclists - | FileCheck %s
; RUN: llc -mtriple x86_64-pc-linux -split-dwarf-file=foo.dwo -filetype=obj < %s \
; RUN: | llvm-dwarfdump -debug-info -debug-loclists - | FileCheck %s
; RUN: | llvm-dwarfdump -debug-info -debug-loclists - | FileCheck --check-prefix=SPLIT --check-prefix=CHECK %s
; RUN: llc -mtriple x86_64-pc-linux -split-dwarf-file=foo.dwo -filetype=asm < %s \
; RUN: | FileCheck --check-prefix=ASM %s
; A bit of a brittle test - this is testing the specific DWO_id. The
; alternative would be to test two files with different DW_OP_convert values &
; ensuring the DWO IDs differ when the DW_OP_convert parameter differs.
; So if this test ends up being a brittle pain to maintain, updating the DWO ID
; often - add another IR file with a different DW_OP_convert that's otherwise
; identical and demonstrate that they have different DWO IDs.
; SPLIT: 0x00000000: Compile Unit: {{.*}} DWO_id = 0xafd73565c68bc661
; Regression testing a fairly quirky bug where instead of hashing (see above),
; extra bytes would be emitted into the output assembly in no
; particular/intentional section - so let's check they don't show up at all:
; ASM-NOT: .asciz "\200\200\200"
; CHECK: 0x{{0*}}[[TYPE:.*]]: DW_TAG_base_type
; CHECK-NEXT: DW_AT_name ("DW_ATE_unsigned_32")