[PATCH] D27765: Renumber testcase metadata nodes after D26769. [NFC]

Adrian Prantl via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Wed Dec 14 11:18:56 PST 2016

aprantl added inline comments.

Comment at: test/CodeGen/AArch64/arm64-2011-03-17-AsmPrinterCrash.ll:23-44
-!0 = !DIGlobalVariableExpression(var: !DIGlobalVariable(name: "vsplive", line: 617, isLocal: true, isDefinition: true, scope: !1, file: !2, type: !6))
-!1 = distinct !DISubprogram(name: "drt_vsprintf", line: 616, isLocal: false, isDefinition: true, virtualIndex: 6, flags: DIFlagPrototyped, isOptimized: false, unit: !3, file: !20, scope: !2, type: !4)
-!2 = !DIFile(filename: "print.i", directory: "/Volumes/Ebi/echeng/radars/r9146594")
-!3 = distinct !DICompileUnit(language: DW_LANG_C99, producer: "clang version 3.0 (http://llvm.org/git/clang.git git:/git/puzzlebox/clang.git/ c4d1aea01c4444eb81bdbf391f1be309127c3cf1)", isOptimized: true, emissionKind: FullDebug, file: !20, enums: !21, retainedTypes: !21, globals: !{!0})
-!4 = !DISubroutineType(types: !5)
-!5 = !{!6}
-!6 = !DIBasicType(tag: DW_TAG_base_type, name: "int", size: 32, align: 32, encoding: DW_ATE_signed)
dblaikie wrote:
> Is dropping all this debug info intentional?
> & theer are other 'interesting' changes in other files, but rather than discussing all of them:
> Perhaps it'd be better to just run this script over the files that are changed by your previous commit? Ah, I guess that's what you're already doing.
> *ponders*
Lots of testcases have unreachable MDNodes in them that are leftover from previous lossy conversions. The script has to drop them.

You're correct, I only ran it on the files containing a DIGlobalVariableExpression and tthat survived the llvm-as | llvm-dis loop. There are a bunch of testcases that don't even do that because of incomplete metadata (missing version etc.). I'll fix up those manually. In a separate commit.

Comment at: test/DebugInfo/ARM/big-endian-bitfield.ll:39
 ; CHECK:      DW_AT_data_bit_offset      [DW_FORM_data1]	(0x05)
-!9 = !DIDerivedType(tag: DW_TAG_member, name: "k", scope: !5, file: !1, line: 3, baseType: !8, size: 6, offset: 5)
+!9 = !DIBasicType(name: "int", size: 32, encoding: DW_ATE_signed)
 ; CHECK: DW_TAG_member
dblaikie wrote:
> This one got thrown off by the DIBasicType here - worth considering a fix to the script to improve this case?
I can have the sync heuristic also ignore ![0-9]. Maybe that works.

Comment at: test/DebugInfo/COFF/globals-discarded.ll:31
+!6 = distinct !DIGlobalVariableExpression(var: !7)
+!7 = !DIGlobalVariable(name: "_OptionsStorage", scope: !2, file: !3, line: 3, type: !8, isLocal: true, isDefinition: true)
dblaikie wrote:
> Does the script insert blank lines? Might not be worth doing so in a block of contiguous IR without any comments, etc?
I can modify it to only insert a blank line if the previous line contained a `;`.


More information about the llvm-commits mailing list