clayborg requested changes to this revision.
clayborg added a comment.
This revision now requires changes to proceed.
One minor IsValid() fix in DWARFDIE to avoid crashes and this is good to go.
Comment at: source/Plugins/SymbolFile/DWARF/DWARFDIE.cpp:28-35
@@ -27,1 +27,10 @@
tberghammer added inline comments.
Comment at: source/Plugins/SymbolFile/DWARF/DIERef.cpp:34-44
@@ +33,13 @@
+
+DIERef::DIERef(const DWARFFormValue& form_value) :
+cu_offset(DW_INVALID_OFFSET),
+die_offset(DW_INVALID_OFFSET)
+{
+if (form_value.IsValid())
+{
+
clayborg requested changes to this revision.
clayborg added a comment.
This revision now requires changes to proceed.
A few changes.
I think this might create problems for DWARF in .o files on MacOSX. Are you
able to test on MacOSX? We used to store the compile unit index in the upper 32
bits o
tberghammer added inline comments.
Comment at: source/Plugins/SymbolFile/DWARF/DWARFCompileUnit.cpp:316-328
@@ +315,15 @@
+
+const char* dwo_name = cu_die.GetAttributeValueAsString(m_dwarf2Data,
+this,
+
clayborg requested changes to this revision.
clayborg added a comment.
This revision now requires changes to proceed.
See inlined comments.
Main fix themes:
- It would be great to try and avoid any manual function calls where we supply
the compile unit and a DIE offset that we assume comes from
tberghammer added a comment.
Added some inline comments to explain some implementation decisions
Comment at: include/lldb/Symbol/ObjectFile.h:372
@@ -371,3 +371,3 @@
virtual SectionList *
-GetSectionList ();
+GetSectionList (bool update_module_section_list = true);
tberghammer updated the summary for this revision.
tberghammer updated this revision to Diff 33676.
tberghammer added a comment.
[RFC] DO NOT COMMIT!
Update the design based on the discussion for the previous diff.
The current version have ~35 failures with split dwarf (most of them hit an
asse
clayborg added a comment.
I got my DWARFDIE stuff in with:
% svn commit
Sendinginclude/lldb/Core/dwarf.h
Sendinginclude/lldb/Symbol/ClangASTContext.h
Sendinginclude/lldb/Symbol/TypeSystem.h
Sendinglldb.xcodeproj/project.pbxproj
Sending
source/Pl
clayborg added a comment.
In http://reviews.llvm.org/D12291#233274, @tberghammer wrote:
> I though a bit more about the abstraction you plan to introduce (DWARFDIE)
> and I started to believe we don't need it at all. If we use lldb::user_id_t
> in the NameToDIE indexes where the first 4 byte is
tberghammer added a comment.
I though a bit more about the abstraction you plan to introduce (DWARFDIE) and
I started to believe we don't need it at all. If we use lldb::user_id_t in the
NameToDIE indexes where the first 4 byte is the compile unit offset (for dwo)
or the compile unit index (for
clayborg added a comment.
Let me get the DWARFDIE abstraction in and we can see where we are after I get
this in as we will see more of what is possible to abstract when this is done.
http://reviews.llvm.org/D12291
___
lldb-commits mailing list
lld
tberghammer added a comment.
If I understand you correctly then you are suggesting to create a class like
SymbolFileDWARFDebugMap for handling object files with dwo files. I think that
approach is practically have the same code flow as the current one (one symbol
file which one stores a list of
clayborg added a comment.
For our DWARF in .o files, I have SymbolFileDWARFDebugMap which loads the DWARF
from .o files. Each .o file is loaded in a completely unchanged version of
SymbolFileDWARF. Any lldb::user_id_t that are generated use:
lldb::user_id_t
SymbolFileDWARF::MakeUserID (dw_o
tberghammer added a comment.
Sorry, first I manage to submit my comments without finishing them.
In http://reviews.llvm.org/D12291#232191, @clayborg wrote:
> I also question why Symbo
>
> In http://reviews.llvm.org/D12291#231523, @tberghammer wrote:
>
> > In the current version of the patch the
tberghammer added a comment.
In http://reviews.llvm.org/D12291#232191, @clayborg wrote:
> I also question why Symbo
>
> In http://reviews.llvm.org/D12291#231523, @tberghammer wrote:
>
> > In the current version of the patch the compile units in the main object
> > file hands out only the compile
clayborg added a comment.
I also question why Symbo
In http://reviews.llvm.org/D12291#231523, @tberghammer wrote:
> In the current version of the patch the compile units in the main object file
> hands out only the compile unit DIE with the information what is available in
> the main object fi
tberghammer updated this revision to Diff 33073.
tberghammer marked 10 inline comments as done.
tberghammer added a comment.
Fix the minor refactors requested in the review.
I haven't changed the approach to return all DIEs from the dwo file when
indexing the main compile unit because I would li
abidh added inline comments.
Comment at: source/Plugins/SymbolFile/DWARF/DWARFCompileUnit.cpp:348
@@ -292,1 +347,3 @@
+}
+}
You are not using the DW_AT_GNU_dwo_id. Is this intentional or an oversight?
http://reviews.llvm.org/D12291
_
tberghammer added a comment.
In the current version of the patch the compile units in the main object file
hands out only the compile unit DIE with the information what is available in
the main object file. I considered the other approach (hand out all DIEs by the
DWARF compile unit in the main
clayborg requested changes to this revision.
clayborg added a comment.
This revision now requires changes to proceed.
I see the need for a lot of this, but I feel like there are way too many places
where we do this:
SymbolFileDWARFDwo* dwo_symbol_file = dwarf_cu->GetDwoSymbolFile();
if (dwo_
tberghammer created this revision.
tberghammer added reviewers: labath, clayborg.
tberghammer added a subscriber: lldb-commits.
Add split dwarf support to SymbolFileDWARF
* Create new dwo symbol file class
* Add handling for .dwo sections
* Propagate queries from SymbolFileDWARF to the dwo symbol
21 matches
Mail list logo
