[llvm] r174479 - Attempt to recover gdb bot after r174445.

David Blaikie dblaikie at gmail.com
Fri Feb 8 15:39:58 PST 2013 

On Fri, Feb 8, 2013 at 3:35 PM, Manman Ren <mren at apple.com> wrote:
>
> On Feb 8, 2013, at 3:31 PM, David Blaikie <dblaikie at gmail.com> wrote:
>
>> On Fri, Feb 8, 2013 at 3:29 PM, Manman Ren <mren at apple.com> wrote:
>>>
>>> On Feb 8, 2013, at 3:21 PM, David Blaikie <dblaikie at gmail.com> wrote:
>>>
>>>> On Fri, Feb 8, 2013 at 3:14 PM, Manman Ren <mren at apple.com> wrote:
>>>>>
>>>>> On Feb 8, 2013, at 11:41 AM, David Blaikie <dblaikie at gmail.com> wrote:
>>>>>
>>>>>> On Tue, Feb 5, 2013 at 4:59 PM, Manman Ren <mren at apple.com> wrote:
>>>>>>> Author: mren
>>>>>>> Date: Tue Feb  5 18:59:41 2013
>>>>>>> New Revision: 174479
>>>>>>>
>>>>>>> URL: http://llvm.org/viewvc/llvm-project?rev=174479&view=rev
>>>>>>> Log:
>>>>>>> Attempt to recover gdb bot after r174445.
>>>>>>>
>>>>>>> Failure: undefined symbol 'Lline_table_start0'.
>>>>>>> Root-cause: we use a symbol subtraction to calculate at_stmt_list, but
>>>>>>> the line table entries are not dumped in the assembly.
>>>>>>> Fix: use zero instead of a symbol subtraction for Compile Unit 0.
>>>>>>
>>>>>> This seems to not have completely fixed problems of this kind. I
>>>>>> discovered more breakage in DragonEgg. Any attempt to build with debug
>>>>>> info in DragonEgg is currently broken due to the original change, even
>>>>>> after this fix.
>>>>>>
>>>>>> You can reproduce this without dragonegg:
>>>>>>
>>>>>> $ echo "int main() { }" | clang -x c++ -g -S - -o hello.s && clang hello.s
>>>>>
>>>>> This simple example worked for me with source up to r174431.
>>>>
>>>> It works for me when I revert r174479 + r174445, fails for me with ToT.
>>> Works for me without any local change:
>>> Maybe it is because our ld version.
>>> I got another email from Duncan, there the version of ld is: GNU ld (GNU Binutils for Debian) 2.20.1-system.20100303
>>
>> I'm not sure how it'd be an ld version issue - when I look at the
>> assembly produced by the above command I see a reference to the
>> .Lline_table_start0 symbol in the debug info & no other mention of it.
>> Seems pretty clearly something that would not be able to be linked...
>>
>> does your hello.s from the above command not contain any mention of
>> line_table_start0?
> No, my hello.s does not have line_table_start0.

That bit I find a little surprising - but I'm probably still not
understanding something.

Eric pointed out on IRC/offline, though, that this probably doesn't
reproduce on Darwin as we unconditionally emit the line table on
Darwin for some reason (perhaps the linker doesn't support the .loc
directives?).

if you try:

clang hello.cpp -g -S -target i386-pc-linux-gnu

you'll see the hello.s file contains a reference to
.Lline_table_start0 but no definition for it - this is a/the problem

> I asked Chad to try, and it worked for him too.
> The email from Duncan is about a more complicated case, so there it is maybe a linker issue.
>
> Thanks,
> Manman
>
>>
>>>
>>> Thanks,
>>> Manman
>>>
>>>>
>>>>> With r174479, we are not using line_table_start0 in at_stmt_list, so it should be okay if we don't emit the line table entries in MCAsmStreamer.
>>>>
>>>> It's not clear to me - have you tried? So far as I can see, with ToT,
>>>> that repro fails to link with the aforementioned linker error. Are you
>>>> not seeing that behavior?
>>>>
>>>>> I noticed that using -disable-dot-loc will dump the line table entries.
>>>>
>>>> Right - because we emit the line table if we don't emit .loc
>>>> directives. If we do emit .loc directives then we skip the line table
>>>> because we figure the linker can figure it out based on the .loc
>>>> directives.
>>>>
>>>> - David
>>>>
>>>>>
>>>>> Thanks,
>>>>> Manman
>>>>>
>>>>>> /tmp/hello-5jpGIX.o:(.debug_info+0x1e): undefined reference to
>>>>>> `.Lline_table_start0'
>>>>>>
>>>>>> Essentially: the assembly code emitted by clang -g is currently
>>>>>> unusable due to this undefined reference
>>>>>>
>>>>>> Eric pointed out that MCAsmStreamer.cpp:1420 - where we don't emit the
>>>>>> line table if we expect the .loc directives to be sufficient for the
>>>>>> assembler to create the line table itself. Since we don't emit the
>>>>>> line table we don't emit the extra special line_table_start symbol
>>>>>> that's being used in the debug info. (this is paraphrased from Eric's
>>>>>> comments based on my rather shallow understanding of the issues)
>>>>>>
>>>>>> - David
>>>>>>
>>>>>>>
>>>>>>> Modified:
>>>>>>>  llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
>>>>>>>  llvm/trunk/test/CodeGen/ARM/2011-01-19-MergedGlobalDbg.ll
>>>>>>>  llvm/trunk/test/CodeGen/ARM/2011-08-02-MergedGlobalDbg.ll
>>>>>>>  llvm/trunk/test/CodeGen/X86/2011-01-24-DbgValue-Before-Use.ll
>>>>>>>  llvm/trunk/test/CodeGen/X86/dbg-value-location.ll
>>>>>>>  llvm/trunk/test/DebugInfo/X86/stringpool.ll
>>>>>>>
>>>>>>> Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
>>>>>>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp?rev=174479&r1=174478&r2=174479&view=diff
>>>>>>> ==============================================================================
>>>>>>> --- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp (original)
>>>>>>> +++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp Tue Feb  5 18:59:41 2013
>>>>>>> @@ -674,6 +674,8 @@ CompileUnit *DwarfDebug::constructCompil
>>>>>>> if (Asm->MAI->doesDwarfUseRelocationsAcrossSections())
>>>>>>>   NewCU->addLabel(Die, dwarf::DW_AT_stmt_list, dwarf::DW_FORM_data4,
>>>>>>>                   LineTableStartSym);
>>>>>>> +  else if (NewCU->getUniqueID() == 0)
>>>>>>> +    NewCU->addUInt(Die, dwarf::DW_AT_stmt_list, dwarf::DW_FORM_data4, 0);
>>>>>>> else
>>>>>>>   NewCU->addDelta(Die, dwarf::DW_AT_stmt_list, dwarf::DW_FORM_data4,
>>>>>>>                   LineTableStartSym, Asm->GetTempSymbol("section_line"));
>>>>>>>
>>>>>>> Modified: llvm/trunk/test/CodeGen/ARM/2011-01-19-MergedGlobalDbg.ll
>>>>>>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/2011-01-19-MergedGlobalDbg.ll?rev=174479&r1=174478&r2=174479&view=diff
>>>>>>> ==============================================================================
>>>>>>> --- llvm/trunk/test/CodeGen/ARM/2011-01-19-MergedGlobalDbg.ll (original)
>>>>>>> +++ llvm/trunk/test/CodeGen/ARM/2011-01-19-MergedGlobalDbg.ll Tue Feb  5 18:59:41 2013
>>>>>>> @@ -17,7 +17,7 @@ target triple = "thumbv7-apple-darwin10"
>>>>>>> ; DW_OP_constu
>>>>>>> ; offset
>>>>>>>
>>>>>>> -;CHECK: .long Lset7
>>>>>>> +;CHECK: .long Lset6
>>>>>>> ;CHECK-NEXT:        @ DW_AT_type
>>>>>>> ;CHECK-NEXT:        @ DW_AT_decl_file
>>>>>>> ;CHECK-NEXT:        @ DW_AT_decl_line
>>>>>>>
>>>>>>> Modified: llvm/trunk/test/CodeGen/ARM/2011-08-02-MergedGlobalDbg.ll
>>>>>>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/2011-08-02-MergedGlobalDbg.ll?rev=174479&r1=174478&r2=174479&view=diff
>>>>>>> ==============================================================================
>>>>>>> --- llvm/trunk/test/CodeGen/ARM/2011-08-02-MergedGlobalDbg.ll (original)
>>>>>>> +++ llvm/trunk/test/CodeGen/ARM/2011-08-02-MergedGlobalDbg.ll Tue Feb  5 18:59:41 2013
>>>>>>> @@ -8,7 +8,7 @@
>>>>>>> ; DW_OP_constu
>>>>>>> ; offset
>>>>>>>
>>>>>>> -;CHECK: .long Lset34
>>>>>>> +;CHECK: .long Lset33
>>>>>>> ;CHECK-NEXT:        @ DW_AT_type
>>>>>>> ;CHECK-NEXT:        @ DW_AT_decl_file
>>>>>>> ;CHECK-NEXT:        @ DW_AT_decl_line
>>>>>>>
>>>>>>> Modified: llvm/trunk/test/CodeGen/X86/2011-01-24-DbgValue-Before-Use.ll
>>>>>>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/2011-01-24-DbgValue-Before-Use.ll?rev=174479&r1=174478&r2=174479&view=diff
>>>>>>> ==============================================================================
>>>>>>> --- llvm/trunk/test/CodeGen/X86/2011-01-24-DbgValue-Before-Use.ll (original)
>>>>>>> +++ llvm/trunk/test/CodeGen/X86/2011-01-24-DbgValue-Before-Use.ll Tue Feb  5 18:59:41 2013
>>>>>>> @@ -4,7 +4,7 @@ target datalayout = "e-p:64:64:64-i1:8:8
>>>>>>> target triple = "x86_64-apple-darwin10.0.0"
>>>>>>>
>>>>>>> ; Check debug info for variable z_s
>>>>>>> -;CHECK: .long Lset14
>>>>>>> +;CHECK: .long Lset13
>>>>>>> ;CHECK-NEXT:  ## DW_AT_decl_file
>>>>>>> ;CHECK-NEXT:  ## DW_AT_decl_line
>>>>>>> ;CHECK-NEXT:  ## DW_AT_type
>>>>>>>
>>>>>>> Modified: llvm/trunk/test/CodeGen/X86/dbg-value-location.ll
>>>>>>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/dbg-value-location.ll?rev=174479&r1=174478&r2=174479&view=diff
>>>>>>> ==============================================================================
>>>>>>> --- llvm/trunk/test/CodeGen/X86/dbg-value-location.ll (original)
>>>>>>> +++ llvm/trunk/test/CodeGen/X86/dbg-value-location.ll Tue Feb  5 18:59:41 2013
>>>>>>> @@ -4,7 +4,7 @@ target datalayout = "e-p:64:64:64-i1:8:8
>>>>>>> target triple = "x86_64-apple-darwin10.0.0"
>>>>>>> ;Radar 8950491
>>>>>>>
>>>>>>> -;CHECK: .long Lset6
>>>>>>> +;CHECK: .long Lset5
>>>>>>> ;CHECK-NEXT:        ## DW_AT_decl_file
>>>>>>> ;CHECK-NEXT:        ## DW_AT_decl_line
>>>>>>> ;CHECK-NEXT:        ## DW_AT_type
>>>>>>>
>>>>>>> Modified: llvm/trunk/test/DebugInfo/X86/stringpool.ll
>>>>>>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/DebugInfo/X86/stringpool.ll?rev=174479&r1=174478&r2=174479&view=diff
>>>>>>> ==============================================================================
>>>>>>> --- llvm/trunk/test/DebugInfo/X86/stringpool.ll (original)
>>>>>>> +++ llvm/trunk/test/DebugInfo/X86/stringpool.ll Tue Feb  5 18:59:41 2013
>>>>>>> @@ -23,8 +23,8 @@
>>>>>>> ; LINUX-NEXT: .quad   yyyy
>>>>>>>
>>>>>>> ; Verify that we refer to 'yyyy' without a relocation.
>>>>>>> -; DARWIN: Lset6 = Linfo_string3-Linfo_string          ## DW_AT_name
>>>>>>> -; DARWIN-NEXT:        .long   Lset6
>>>>>>> +; DARWIN: Lset5 = Linfo_string3-Linfo_string          ## DW_AT_name
>>>>>>> +; DARWIN-NEXT:        .long   Lset5
>>>>>>> ; DARWIN-NEXT:        .long   39                      ## DW_AT_type
>>>>>>> ; DARWIN-NEXT:        .byte   1                       ## DW_AT_external
>>>>>>> ; DARWIN-NEXT:        .byte   1                       ## DW_AT_decl_file
>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> llvm-commits mailing list
>>>>>>> llvm-commits at cs.uiuc.edu
>>>>>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>>>>
>>>
>

More information about the llvm-commits mailing list