Processor shows incorrect source file name if a frame have an inlined
frame and their source files are different.
Consider this example:
FILE 0 /tmp/a.h
FILE 1 /tmp/a.cpp
INLINE_ORIGIN 0 0 foo()
FUNC 1110 a 0 main
INLINE 0 22 0 1110 7
1110 7 3 0
1117 3 23 1
When querying the address 0x1110, we know this line 0x1110 corresponds
to /tmp/a.h line 3 and it's inside a inlined function foo() which is
defined at /tmp/a.h and called at line 22. But we don't know at which
file it's being called at line 22. So, we will get stacks like this:
void foo() /tmp/a.h:3
int main() /tmp/a.h:22
The correct stacks should be this:
void foo() /tmp/a.h:3
int main() /tmp/a.cpp:22
In this change:
1. Remove file_id field for INLINE_ORIGIN record.
2. Add call_site_file_id for INLINE record to represents the file where
this call being inlined.
After adding call_site_file_id to it (as third field), it looks like
this:
FILE 0 /tmp/a.h
FILE 1 /tmp/a.cpp
INLINE_ORIGIN 0 foo()
FUNC 1110 a 0 main
INLINE 0 22 1 0 1110 7
1110 7 3 0
1117 3 23 1
Bug: 1190878
Change-Id: Ibbb697d2f7e1b6ac3208cac6fae4353c8743198d
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/3232838
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
- Added StringView which is used as a reference to a string, but
doesn't own the string.
- Removed the old string pool in DwarfCUToModule::FilePrivate, since
it's doing string copy.
- Added a string pool in Module to store functions/inline origins'
names (mangled and demangled).
- The peak memory usage drops from 20.6 GB to 12.5 GB when disabling
inline records and drops from 36 GB to 20.3 GB when enabling inline records.
Bug: chromium:1246974, chromium:1250351
Change-Id: Ie7e9740ea10c1930a0fc58c6becaae2d718b83b8
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/3189410
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
It moves InlineOriginMap to module.h. Let Module keeps the global InlineOriginMap to easily get all referenced InlineOrigin when emitting. And release allocated memory inside its destructor.
Verified that the symbol file with inline records for chrome is the same before and after this change.
Change-Id: I7541aa05d3d2df0b9d52d670cab58241baecf20d
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/3171638
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
This change makes sure dump_syms process DW_TAG_inlined_subroutine only when -d flag is given, which save memory and time when -d is not given. Before this, it always processes DW_TAG_inlined_subroutine and -d determines whether or not to emit INLINE records.
Bug: chromium:1250351, chromium:1246974
Change-Id: I54725ba1e513cafe17268ca389ff8acc9c11b25e
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/3166674
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Dwarf generated by Clang -g1 will not have DW_AT_inline attribute for some DW_TAG_subprograms even if they are inlined. This warning recently increased a lot (~ 3 million) due to DW_TAG_inlined_subroutine also complains about unknown abstract origin. It caused infra failure in building bots.
Bug: 1241579
Change-Id: I9b5135925b71aa915760c140bcf73fc603bb77d3
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/3111782
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Building fails for some people because configure requires c++11 but make_unique is a c++14 feature.
Change-Id: I23ce689fc92e9e90a95e7643ff29602f6b32ccbb
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/3107784
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
The size of symbol file for chrome binary increased from 577 MB to
1205 MB. There are 7,453,748 INLINE records and 1,268,493 INLINE_ORIGIN
records.
Bug: 1190878
Change-Id: I802ec1b4574c14f74ff80d0f69daf3c81085778a
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/2915828
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Before this change, forward_ref_die_to_func is always empty.
Because forward reference DIEs could cross compilation units,
forward_ref_die_to_func is moved into FilePrivate.
By using "grep "name omitted" sym | wc -l", the number of functions
with no name is reduced from 26951 to 203 if forward reference only is
allowed within CU. It's reduced from 203 to 6 if crossing compilation
units is allowed.
Change-Id: Ie2a457abfc0c4d8e68fe0fa595a27ea4abf33a76
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/2904093
Reviewed-by: Mike Frysinger <vapier@chromium.org>
Third of 5 small patches to fix various breakpad issues found
while testing dump_syms on DWARF v5 in ChromeOS.
The offset into the line table was being incorrectly added to the
.debug_string and debug_line_str sections in the code for reading
the line table. It was also skipping trying to read the line table
if the .debug_line_str section was present. This CL fixes these
issues.
Change-Id: If14543731016bcee201b8c33dca53e9520007222
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/2634548
Reviewed-by: Sterling Augustine <saugustine@google.com>
Second of 5 small patches to fix various breakpad issues found
while testing dump_syms on DWARF v5 in ChromeOS.
This patch adds code to properly find & parse the
DW_AT_str_offsets_base attribute, and use it to handle strings
of the forms DW_FORM_strx, DW_FORM_strx1..DW_FORM_strx4 This is the
largest of the DWARF5 fixes. It also includes a unittest to test
using a string offset.
Change-Id: I5d1def862d9d91cae4b2853578441e04ea85449d
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/2634547
Reviewed-by: Sterling Augustine <saugustine@google.com>
First of 5 small patches to fix various breakpad issues found
while testing dump_syms on DWARF v5 in ChromeOS.
This patch adds some missing DWARF enums, and their uses, and fixes
one small typo (was updating 'lineptr' instead of '*lineptr').
Change-Id: Ic674d5db29f29a69a3f6e370d0553eb4139c91de
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/2634546
Reviewed-by: Sterling Augustine <saugustine@google.com>
This is a big change. dwarf5 range lists are quite a bit more complicated
than dwarf 4 range lists, both in the contextual information required, and
in their own representation and interpretation.
The big design choice here is how to pass the CU information all the
way down to the reader. I chose a structure, because otherwise the
parameter list gets very long and error prone (and has to be passed
down several levels). This structure could be made a parto of the CU
context itself, or the range handler, so it wouldn't have to be
separately assembled at range-list read time, but both of those
solutions get even more invasive, and harder to follow.
I've tried to figure out how to break this into smaller changes, but it
affects nearly everything that has to do with a compilation unit's
own addresses and when decisions must be made about how to read them.
Dependency injection will do that to you.
It does add tests for range list reading, which did not exist before.
Change-Id: I923b9a2c3379a0f52609bc05310097de5cbb7227
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/2446635
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
This reverts commit 2b936b06c1.
After getting deep into the dwarf5 range reader, I realized that this
should be done a somewhat different way. So reverting in favor or
a better design, coming in a few minutes.
Change-Id: Ie0b2846e70b3df1e637831e96ea69fe093f4e712
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/2446011
Reviewed-by: Mark Mentovai <mark@chromium.org>
Dwarf5 introduces a new .debug_rngslist section, to take the place
of the Dwarf4 .debug_ranges. However, the dwarf version is CU-based,
and not file-based, so there can be both sections, and which section
the CU needs isn't known until the dwarf parser encounters either
DW_AT_ranges (dwarf 4 and lower) or DW_AT_rnglists_base (dwarf 5).
This change refactors the code around range lists and range list
readers to defer the decision of what section to parse until
the relevant attribute is found. It moves the range list section
reader from the range-list handler itself (which doesn't know which
section it will use) to the CU context, and then lets the handler
know when it encounters DW_AT_ranges.
I will add a reader for the new dwarf5 section, along with the code to
interpret the new section, and its forms and such in a subsequent patch.
Change-Id: Ie92e4c9daa3f0acb98d7ef74f6b9c2065db849b1
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/2433684
Reviewed-by: Mark Mentovai <mark@chromium.org>
We do this in a lot of places, but we're inconsistent.
Normalize the code to the Google C++ style guide.
Change-Id: Ic2aceab661ce8f6b993dda21b1cdf5d2198dcbbf
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/2262932
Reviewed-by: Sterling Augustine <saugustine@google.com>
Reviewed-by: Mark Mentovai <mark@chromium.org>
For the upcoming Dwarf5 LineInfo changes, we will need to get
several additional sections by name, but without a the Compilation
Unit. This change prepares for that.
Change-Id: I566855abb339a856110a2f7d243d3848fe2b3c18
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/2268861
Reviewed-by: Mark Mentovai <mark@chromium.org>
Given the almost nonexistent direct dwarfreader tests, I think
the best way to test these dwarf5 additions will be to add a full
dwarf5 compilation unit similar to the ones used incidentally in
the other tests. But I can't do that until enough dwarf5 is
correctly implemented.
Change-Id: I3418bda7212ae85c4b67232a2ab8fea9b9ca5d42
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/2258838
Reviewed-by: Mark Mentovai <mark@chromium.org>
Reviewed-by: Mike Frysinger <vapier@chromium.org>
Previously, dump_syms produced warnings whenever a DW_AT_specification
or DW_AT_abstract_origin attribute was a forward reference. 47cd498384
allowed those attributes to carry forward references, removing the
warnings altogether. It was not correct to remove the warnings entirely.
References that do not point to valid DIEs should still produce
warnings, whether a back reference or a forward reference.
This reintroduces those warnings as appropriate.
Bug: google-breakpad:813
Test: dumper_unittest SimpleCU.UnknownAbstractOrigin,Specifications.BadOffset
Change-Id: Ie7222c7a1886bab31423f27e2fbcce93e69625b3
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/2090103
Reviewed-by: Mike Frysinger <vapier@chromium.org>
DW_AT_specification and DW_AT_abstract_origin attributes carry
references to other DIEs. Nothing prevents the DIEs referred to from
appearing later in .debug_info than the DIE containing the referring
attribute, but dump_syms incompletly implemented its handling of these
references, and was only able to resolve them when they were
back-references.
This will fix the chronic warnings produced by dump_syms of the form:
dump_syms: the DIE at offset <offset> has a {DW_AT_specification,
DW_AT_abstract_origin} attribute referring to the die at offset
<offset>, which either was not marked as {a declaration, an inline}, or
comes later in the file
Patch by Greg Clayton
Bug: breakpad:441
Change-Id: I98957d64a234c22afb6d0153f1bdc09e6a600b1d
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/1946706
Reviewed-by: Mark Mentovai <mark@chromium.org>
The high_pc is an address and has already been read from .debug_addr
before being passed into FuncHandler::ProcessAttributeUnsigned.
Bug:870908
Change-Id: I950098e360b5193f26bf767b8fa0a5f9d59e66ce
Reviewed-on: https://chromium-review.googlesource.com/1178760
Reviewed-by: Mark Mentovai <mark@chromium.org>
This enables the DWARF reader to properly parse DW_AT_ranges attributes
in compilation units and functions. Code covered by a function is now
represented by a vector of ranges instead of a single contiguous range
and DW_AT_ranges entries are used to populate it. All the code and tests
that assumed functions to be contiguous entities has been updated to
reflect the change. DW_AT_ranges attributes found in compilation units
are parsed but no data is generated for them as it is not currently needed.
BUG=754
Change-Id: I310391b525aaba0dd329f1e3187486f2e0c6d442
Reviewed-on: https://chromium-review.googlesource.com/1124721
Reviewed-by: Ted Mielczarek <ted.mielczarek@gmail.com>
When DW_AT_MIPS_linkage_name doesn't demangle, breakpad currently throws
the symbol completely, but in some cases, there is no DW_AT_name or
DW_AT_abstract_origin to figure out a name, and the raw value from
DW_AT_MIPS_linkage_name is still better than nothing. Fall back to that
in when there is nothing else.
R=ted@mielczarek.org
Change-Id: I5cc7580244f2b99f5f1f279d09b904031cae1a37
Reviewed-on: https://chromium-review.googlesource.com/1082176
Reviewed-by: Ted Mielczarek <ted.mielczarek@gmail.com>
For more details take a look at common/using_std_string.h
BUG=
Change-Id: Ifebfc57f691ef3a3bef8cfed7106c567985edffc
Reviewed-on: https://chromium-review.googlesource.com/399738
Reviewed-by: Mark Mentovai <mark@chromium.org>
The Rust compiler uses GCC C++ name mangling, but it has another layer of
encoding so abi::cxa_demangle doesn't produce great results. This patch
changes dump_syms to dump unmangled names by default so that consumers can
demangle them after-the-fact.
It also adds a tiny bit of support for linking against a Rust library I wrote
that can demangle Rust symbols nicely:
https://github.com/luser/rust-demangle-capi
BUG=
Change-Id: I63a425035ebb7ac516f067fed2aa782849ea9604
Reviewed-on: https://chromium-review.googlesource.com/402308
Reviewed-by: Mark Mentovai <mark@chromium.org>
The DWARF data for Swift code has a top-level DW_TAG_module DIE as the
child of the DW_TAG_compile_unit DIE and the parent of the
DW_TAG_subprogram DIEs that dump_syms uses to locate functions.
dump_syms needs to process DW_TAG_module DIEs as introducing nested
scopes to make it work with Swift.
This also reworks demangling to be language-specific, so that the C++
demangler isn't invoked when processing Swift code. The DWARF data for
Swift code presents its mangled names in the same form as used for C++
(DW_AT_MIPS_linkage_name or DW_AT_linkage_name) but the mangling is
Swift-specific (beginning with _T instead of _Z). There is no
programmatic interface to a Swift name demangler as an analogue to C++'s
__cxa_demangle(), so mangled Swift names are exposed as-is. Xcode's
"xcrun swift-demangle" can be used to post-process these mangled Swift
names on macOS.
Support for mangled names presented in a DW_AT_linkage_name attribute,
as used by DWARF 4, is added. This supersedes the earlier use of
DW_AT_MIPS_linkage_name.
BUG=google-breakpad:702,google-breakpad:715
R=ted.mielczarek@gmail.com
Review URL: https://codereview.chromium.org/2147523005 .
Currently an inlined function in a namespace in DWARF will
be given a name comprised of just `namespace::`. This is due
to a logic error in ComputeQualifiedName, where it doesn't
handle an empty `unqualified_name` properly.
We apparently have a fair number of these in our Mac builds,
an example of the DWARF that's being mishandled looks like:
0x117eda40: TAG_namespace [5] *
AT_name( "js" )
AT_decl_file( "../../dist/include/js/Utility.h" )
AT_decl_line( 35 )
0x11808500: TAG_subprogram [251] *
AT_low_pc( 0x0000000002f12110 )
AT_high_pc( 0x0000000002f1216b )
AT_APPLE_omit_frame_ptr( 0x01 )
AT_frame_base( rsp )
AT_abstract_origin( {0x0000000011800a4f}"_ZN2js40TraceManuallyBarrieredGenericPointerEdgeEP8JSTracerPPNS_2gc4CellEPKc" )
AT_MIPS_linkage_name( "_ZN2js40TraceManuallyBarrieredGenericPointerEdgeEP8JSTracerPPNS_2gc4CellEPKc" )
AT_name( "TraceManuallyBarrieredGenericPointerEdge" )
AT_decl_file( "/builds/slave/rel-m-rel-m64_bld-000000000000/build/js/src/gc/Marking.cpp" )
AT_decl_line( 547 )
AT_external( 0x01 )
AT_APPLE_optimized( 0x01 )
AT_inline( DW_INL_inlined )
This turned a few instances of this in the file I was testing on into
`<name omitted>`, which seems to just be a symptom of the
"DW_AT_abstract_origin comes later in the file" issue. (Which is probably
also worth fixing given that it occurs some 29k times when dumping
symbols from Firefox's XUL binary, but it's a separate issue.)
R=mark@chromium.org
BUG=
Review URL: https://codereview.chromium.org/1887033002 .
Newer gcc versions default to -Werror=narrowing when using newer C++
standards (which we do). This causes issues when we try to stuff a
value like 0xea into a char -- the value is out of range for signed
char bytes. That's when gcc throws an error:
.../bytereader_unittest.cc: In member function 'virtual void Reader_DW_EH_PE_absptr4_Test::TestBody()':
.../bytereader_unittest.cc:400:55: error: narrowing conversion of '234' from 'int' to 'char' inside { } [-Wnarrowing]
BUG=chromium:579384
TEST=`make check` passes
R=mark@chromium.org
Review URL: https://codereview.chromium.org/1605153004 .
problems.
- The ordering of keys in stl containers cannot change. Make the relevant
members const to guarantee this assumption.
- Add handling and logging for demangle errors.
- Fix a potential double-delete bug if a function passed to AddFunction() is
already present.
BUG=chromium:449214
R=mark@chromium.org
Review URL: https://breakpad.appspot.com/10704002/
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@1415 4c0a9323-5329-0410-9bdc-e9ce6186880e
