|
From: Nuno L. <nun...@is...> - 2009-01-23 00:13:45
Attachments:
vg_dwarf_base_type.txt
|
Hi, Attached is a patch to disable verification of DW_AT_encoding of DW_TAG_base_type. Per my reading of the standard (version 2) it's not stricly necessary to include the encoding and in fact the LLVM compiler sometimes doesn't include it. Regards, Nuno |
|
From: Nicholas N. <n.n...@gm...> - 2009-06-29 05:40:57
|
On Fri, Jan 23, 2009 at 10:13 AM, Nuno Lopes<nun...@is...> wrote: > > Attached is a patch to disable verification of DW_AT_encoding of > DW_TAG_base_type. Per my reading of the standard (version 2) it's not > stricly necessary to include the encoding and in fact the LLVM compiler > sometimes doesn't include it. Nuno, Can you open a bug for this? Can you include a description of what currently happens with LLVM in this case -- does Valgrind abort, emit a warning, fail to read the debug info, etc? Thanks. Nick |
|
From: Tom H. <to...@co...> - 2009-06-29 07:28:00
|
On 29/06/09 06:40, Nicholas Nethercote wrote: > On Fri, Jan 23, 2009 at 10:13 AM, Nuno Lopes<nun...@is...> wrote: > >> Attached is a patch to disable verification of DW_AT_encoding of >> DW_TAG_base_type. Per my reading of the standard (version 2) it's not >> stricly necessary to include the encoding and in fact the LLVM compiler >> sometimes doesn't include it. > > Can you open a bug for this? Can you include a description of what > currently happens with LLVM in this case -- does Valgrind abort, emit > a warning, fail to read the debug info, etc? Thanks. Well page 63 of the current DWARF3 draft says: "A base type entry also has a DW_AT_encoding attribute describing how the base type is encoded and is to be interpreted. The value of this attribute is an integer constant. Which doesn't sound very optional. The DWARF2 standard has similar wording. Tom -- Tom Hughes (to...@co...) http://www.compton.nu/ |
|
From: Nuno L. <nun...@is...> - 2009-06-29 08:01:42
|
>> Attached is a patch to disable verification of DW_AT_encoding of >> DW_TAG_base_type. Per my reading of the standard (version 2) it's not >> stricly necessary to include the encoding and in fact the LLVM compiler >> sometimes doesn't include it. > > Nuno, > > Can you open a bug for this? Can you include a description of what > currently happens with LLVM in this case -- does Valgrind abort, emit > a warning, fail to read the debug info, etc? Thanks. As far as I remember, Julian already commited the patch :) FYI, valgrind was aborting the execution when running code from clang+LLVM. Nuno |