Sourabh Singh Tomar via llvm-dev
2019-Sep-28 16:54 UTC
[llvm-dev] [DebugInfo] emission of loclists.dwo
Hi Folks, While working on dwarf5 features in clang, I noticed some subtleties wrt locationlists-- In split-dwarf + dwarf5 mode - debug_loclists.dwo section is not generated, instead debug_loc.dwo is getting populated in dwo file. + Hexdump shows that pre-standardized encoding used while encoding. This is causing problems for llvm-dwarfdump in dumping location list. I'm guessing, llvm-dwarfdump is assuming it should be accessed according to dwarf5, hence failing{not reporting} ??? In, split-mode +dwarf-4{default} mode -- everything is fine. section generation and llvm-dwarfdump. + While fixing this issue, I've been trying to bring this section debug_loclists.dwo from ground up -- creating LoclistsDWOSection and so forth. Can anybody faced with these issues earlier, validate these findings?, work in progress if it's their any?, approach / solution taken to fix these ??? Thanks ! -- Sourabh Singh Tomar -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20190928/1ceed17f/attachment.html>
David Blaikie via llvm-dev
2019-Sep-29 22:51 UTC
[llvm-dev] [DebugInfo] emission of loclists.dwo
Yep, I've been looking at making some improvements to location list emission (using base address specifiers, etc) & came across this hole as well. Don't have any immediate fix myself - but I'm pretty sure we don't ever want to use the direct address forms (LLE_base_address, LLE_start_length, etc) even in non-Split DWARF (sticking to the address pool forms - base_addressx, startx_length, etc) - so it should be fine to use exactly the same code for debug_loclists and debug_loclists.dwo On Sat, Sep 28, 2019 at 9:54 AM Sourabh Singh Tomar via llvm-dev < llvm-dev at lists.llvm.org> wrote:> Hi Folks, > > While working on dwarf5 features in clang, I noticed some subtleties wrt > locationlists-- > > In split-dwarf + dwarf5 mode - > debug_loclists.dwo section is not generated, instead debug_loc.dwo is > getting populated in dwo file. + Hexdump shows that pre-standardized > encoding used while encoding. > > This is causing problems for llvm-dwarfdump in dumping location list. > I'm guessing, llvm-dwarfdump is assuming it should be accessed according > to dwarf5, hence failing{not reporting} ??? > > In, split-mode +dwarf-4{default} mode -- everything is fine. section > generation and llvm-dwarfdump. > > + While fixing this issue, I've been trying to bring this section > debug_loclists.dwo from ground up -- creating LoclistsDWOSection and so > forth. > > Can anybody faced with these issues earlier, validate these findings?, > work in progress if it's their any?, approach / solution taken to fix these > ??? > > Thanks ! > -- Sourabh Singh Tomar > > > _______________________________________________ > LLVM Developers mailing list > llvm-dev at lists.llvm.org > https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20190929/8fdd38ea/attachment.html>