[Lldb-commits] [PATCH] D46606: General cleanup to minimize the .debug_types patch

clayborg created this revision. clayborg added reviewers: labath, aprantl. Herald added a subscriber: JDevlieghere. This cleanup is designed to make the https://reviews.llvm.org/D32167 patch smaller and easier to read. Cleanup in this patch: Allow DWARFUnit subclasses to hand out the data that

Re: [Lldb-commits] [RFC] Type lookup for template types is broken...

I made a small experiment where I manually edited the DW_AT_name in the assembler output: Baseline: $ lldb with_params (lldb) target create "with_params" (lldb) b 5 (lldb) r Process 28369 stopped * thread #1, name = 'with_params', stop reason = breakpoint 1.1 frame #0: 0x004004fc wi

[Lldb-commits] [PATCH] D46588: [WIP][LLDB-MI] Add possibility to set breakpoints without selecting a target.

clayborg added a comment. Add tests and this will be good to go. Comment at: tools/lldb-mi/MICmdCmdBreak.cpp:166 + + if (sbTarget == rSessionInfo.GetDebugger().GetDummyTarget()) +m_bBrkPtIsPending = true; polyakov.alex wrote: > I want to discuss is it a go

[Lldb-commits] [PATCH] D46588: [WIP][LLDB-MI] Add possibility to set breakpoints without selecting a target.

polyakov.alex added inline comments. Comment at: tools/lldb-mi/MICmdCmdBreak.cpp:166 + + if (sbTarget == rSessionInfo.GetDebugger().GetDummyTarget()) +m_bBrkPtIsPending = true; I want to discuss is it a good way to use 'pending' breakpoint's here? Reposito

[Lldb-commits] [PATCH] D46588: [WIP][LLDB-MI] Add possibility to set breakpoints without selecting a target.

polyakov.alex created this revision. polyakov.alex added reviewers: aprantl, clayborg. polyakov.alex added a project: LLDB. Herald added subscribers: llvm-commits, ki.stfu. It's not a final version of the patch, testcase will be added. Repository: rL LLVM https://reviews.llvm.org/D46588 File

Re: [Lldb-commits] [RFC] Type lookup for template types is broken...

Clang can control the emission of or not in the name, using the "debugger tuning" feature. It sounds like neither LLDB nor PS4 will mind losing the in the name, and it makes the indexing simpler. If the tuning is set for GDB then we can still emit the in the names. It would make me uncomfo

[Lldb-commits] [PATCH] D46529: Add support to object files for accessing the .debug_types section

This revision was automatically updated to reflect the committed changes. Closed by commit rL331777: Add support to object files for accessing the .debug_types section (authored by gclayton, committed by ). Herald added a subscriber: llvm-commits. Changed prior to commit: https://reviews.llvm.o

[Lldb-commits] [lldb] r331777 - Add support to object files for accessing the .debug_types section

Author: gclayton Date: Tue May 8 10:19:24 2018 New Revision: 331777 URL: http://llvm.org/viewvc/llvm-project?rev=331777&view=rev Log: Add support to object files for accessing the .debug_types section In an effort to make the .debug_types patch smaller, breaking out the part that reads the .deb

[Lldb-commits] [PATCH] D46529: Add support to object files for accessing the .debug_types section

clayborg added a comment. Pavel, let me know if you are good with this. https://reviews.llvm.org/D46529 ___ lldb-commits mailing list lldb-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits

[Lldb-commits] [PATCH] D46576: [DWARF] Align non-accelerated function fullname searching with the apple-tables path

clayborg added a comment. the class would be something like: class DWARFIndex { virtual bool Initialize(SymbolFileDWARF &dwarf); // Do manual indexing, or load indexes. Return true if successful (found apple tables, DWARF 5 tables, or had debug info to manually index) virtual ... Lookup(Co

[Lldb-commits] [PATCH] D46576: [DWARF] Align non-accelerated function fullname searching with the apple-tables path

clayborg added a comment. Before adding the DWARF 5 index support, we should virtualize this searching and indexing into a base class. Then we make 1 class for manual DWARF indexing, one for Apple indexes, and one for DWARF5. Then we can unit test each one easily. Thoughts? https://reviews.ll

[Lldb-commits] [PATCH] D46529: Add support to object files for accessing the .debug_types section

davide accepted this revision. davide added a comment. Thanks for taking the time to split this! https://reviews.llvm.org/D46529 ___ lldb-commits mailing list lldb-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits

[Lldb-commits] [PATCH] D46576: [DWARF] Align non-accelerated function fullname searching with the apple-tables path

clayborg accepted this revision. clayborg added a comment. Much better. https://reviews.llvm.org/D46576 ___ lldb-commits mailing list lldb-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits

[Lldb-commits] [PATCH] D46529: Add support to object files for accessing the .debug_types section

clayborg added a comment. BTW: I added the .debug_info verification to the lit test as there was a section made, but it wasn't checked. https://reviews.llvm.org/D46529 ___ lldb-commits mailing list lldb-commits@lists.llvm.org http://lists.llvm.org/

[Lldb-commits] [PATCH] D46576: [DWARF] Align non-accelerated function fullname searching with the apple-tables path

labath updated this revision to Diff 145717. labath added a comment. Stopped inserting demangled names into the fullname index and made the FindFunctions index search in all three indices (fullname, basename, method). The last two shouldn't really contain full names, but they seem to be required f

[Lldb-commits] [PATCH] D46529: Add support to object files for accessing the .debug_types section

clayborg updated this revision to Diff 145715. clayborg added a comment. Added lit test. https://reviews.llvm.org/D46529 Files: include/lldb/lldb-enumerations.h lit/Modules/elf-section-types.yaml source/Core/Section.cpp source/Plugins/ObjectFile/ELF/ObjectFileELF.cpp source/Plugins/Ob

[Lldb-commits] [PATCH] D46576: [DWARF] Align non-accelerated function fullname searching with the apple-tables path

clayborg added inline comments. Comment at: source/Plugins/SymbolFile/DWARF/DWARFUnit.cpp:928-929 Mangled mangled(ConstString(mangled_cstr), true); +func_basenames.Insert(mangled.GetMangledName(), + DIERef(cu_offset, die.G

[Lldb-commits] [PATCH] D46576: [DWARF] Align non-accelerated function fullname searching with the apple-tables path

labath added inline comments. Comment at: source/Plugins/SymbolFile/DWARF/DWARFUnit.cpp:905-908 ConstString demangled = mangled.GetDemangledName(cu_language); if (demangled) func_fullnames.Insert(demangled,

Re: [Lldb-commits] [RFC] Type lookup for template types is broken...

I think for display purposes, the type name should include all of the . If I have two variables, both using class C, but both have different template parameters, I want to see that in the class name and have that properly show up in the variable view. LLDB doesn't actually care about the name th

[Lldb-commits] [PATCH] D46580: Modernize and clean-up the Predicate class

clayborg accepted this revision. clayborg added a comment. This revision is now accepted and ready to land. Looks great! https://reviews.llvm.org/D46580 ___ lldb-commits mailing list lldb-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/

Re: [Lldb-commits] [RFC] Type lookup for template types is broken...

On Tue, 8 May 2018 at 17:12, via lldb-commits wrote: > So…. if the name in the type entry did not include the then the indexing would automatically do what you want; you would need to reconstruct the from the template parameter children of the DIE. Would that help? > On the LLVM side we have

[Lldb-commits] [PATCH] D46576: [DWARF] Align non-accelerated function fullname searching with the apple-tables path

clayborg added a comment. See inlined comments Comment at: source/Plugins/SymbolFile/DWARF/DWARFUnit.cpp:901-902 Mangled mangled(ConstString(mangled_cstr), true); +func_basenames.Insert(mangled.GetMangledName(), + DIERef

Re: [Lldb-commits] [RFC] Type lookup for template types is broken...

So…. if the name in the type entry did not include the then the indexing would automatically do what you want; you would need to reconstruct the from the template parameter children of the DIE. Would that help? On the LLVM side we have debated the merits of in the name back and forth. I wo

Re: [Lldb-commits] [RFC] Type lookup for template types is broken...

> On May 8, 2018, at 9:04 AM, Greg Clayton wrote: > > The only way for us to find all classes whose type is "C" is to add the entry > for all template classes named "C", so I would vote to add them as it is > accurate. Do we currently add one for "C<12, 16>”? Yes we do, not sure it is actual

Re: [Lldb-commits] [RFC] Type lookup for template types is broken...

The only way for us to find all classes whose type is "C" is to add the entry for all template classes named "C", so I would vote to add them as it is accurate. Do we currently add one for "C<12, 16>"? Greg > On May 8, 2018, at 8:58 AM, Frédéric Riss via lldb-commits > wrote: > > > >> On M

Re: [Lldb-commits] [RFC] Type lookup for template types is broken...

> On May 8, 2018, at 8:30 AM, paul.robin...@sony.com wrote: > > > >> -Original Message- >> From: lldb-commits [mailto:lldb-commits-boun...@lists.llvm.org >> ] On Behalf >> Of Pavel Labath via lldb-commits >> Sent: Tuesday, May 08, 2018 10:48

[Lldb-commits] [PATCH] D36977: Add 'break' into GDBRemoteClientBase::SendContinuePacketAndWaitForResponse to fix a warning

labath closed this revision. labath added a comment. Herald added a subscriber: llvm-commits. Looks like somebody already committed a patch like this. Repository: rL LLVM https://reviews.llvm.org/D36977 ___ lldb-commits mailing list lldb-commits@

Re: [Lldb-commits] [RFC] Type lookup for template types is broken...

> -Original Message- > From: lldb-commits [mailto:lldb-commits-boun...@lists.llvm.org] On Behalf > Of Pavel Labath via lldb-commits > Sent: Tuesday, May 08, 2018 10:48 AM > To: fr...@apple.com > Cc: lldb-commits@lists.llvm.org > Subject: Re: [Lldb-commits] [RFC] Type lookup for template t

Re: [Lldb-commits] [RFC] Type lookup for template types is broken...

Well.. it encodes some assumptions about how a class name looks like, which are probably valid for C++, but they don't have to hold for any language frontend LLVM supports. That said, I am not saying this is worth the trouble of adding a special "these are the additional names you are to insert int

Re: [Lldb-commits] [PATCH] D46551: Inject only relevant local variables in the expression evaluation context

> On May 8, 2018, at 1:32 AM, Pavel Labath via Phabricator > wrote: > > labath added a comment. > > One could probably concoct an example (using macros, token pasting and such) > where this would get it wrong, but that's probably not worth supporting. Yeah, I thought about this and I’m not

Re: [Lldb-commits] [RFC] Type lookup for template types is broken...

> On May 8, 2018, at 2:23 AM, Pavel Labath wrote: > > I am still building a picture for myself of how the accelerator tables and > our name lookup works, but from what I managed to learn so far, adding an > accelerator for "C" seems like a useful thing to do. However, this does go > beyond what

[Lldb-commits] [PATCH] D46580: Modernize and clean-up the Predicate class

labath created this revision. labath added reviewers: jingham, clayborg. Herald added a subscriber: mgorny. The comments on this class were out of date with the implementation, and the implementation itself was inconsistent with our usage of the Timeout class (I started converting everything to us

[Lldb-commits] [lldb] r331764 - [test] Re-enable TestUnicodeSymbols

Author: jdevlieghere Date: Tue May 8 06:28:34 2018 New Revision: 331764 URL: http://llvm.org/viewvc/llvm-project?rev=331764&view=rev Log: [test] Re-enable TestUnicodeSymbols Re-enable TestUnicodeSymbols now that we use the in-tree dsymutil. This was disabled because the hashing of unicode symbol

[Lldb-commits] [PATCH] D46576: [DWARF] Align non-accelerated function fullname searching with the apple-tables path

labath created this revision. labath added reviewers: clayborg, JDevlieghere. Before this patch the two paths were doing very different things - the apple path searched the .apple_names section, which contained mangled names, as well as basenames of all functions. It returned any name it found.

[Lldb-commits] [PATCH] D32167: Add support for type units (.debug_types) to LLDB in a way that is compatible with DWARF 5

labath added inline comments. Comment at: source/Plugins/SymbolFile/DWARF/DWARFDataExtractor.h:35-58 + //-- + /// Slide the data in the buffer so that access to the data will + /// start at offset \a offset. + //

[Lldb-commits] [PATCH] D46529: Add support to object files for accessing the .debug_types section

labath accepted this revision. labath added a comment. This revision is now accepted and ready to land. Thanks. Please also update the `lit/Modules/elf-section-types.yaml` test to include the new section type. https://reviews.llvm.org/D46529 ___ ll

Re: [Lldb-commits] [RFC] Type lookup for template types is broken...

I am still building a picture for myself of how the accelerator tables and our name lookup works, but from what I managed to learn so far, adding an accelerator for "C" seems like a useful thing to do. However, this does go beyond what the DWARF 5 spec says we should do (we are only required to add

[Lldb-commits] [PATCH] D46551: Inject only relevant local variables in the expression evaluation context

labath added a comment. One could probably concoct an example (using macros, token pasting and such) where this would get it wrong, but that's probably not worth supporting. Comment at: source/Expression/ExpressionSourceCode.cpp:193 +var_name == ConstString(".block_des