Commit Graph

139 Commits

Author SHA1 Message Date
Kostya Serebryany e88161626f [lsan] disable lsan if wordsize is not 64
llvm-svn: 209283
2014-05-21 08:30:18 +00:00
Alexey Samsonov 78a8435fd6 [CMake] Rename add_compiler_rt_static_runtime to add_compiler_rt_runtime.
Soon there will be an option to build compiler-rt parts as shared libraries
on Linux. Extracted from http://llvm-reviews.chandlerc.com/D3042
by Yuri Gribov.

llvm-svn: 205183
2014-03-31 13:45:36 +00:00
Rafael Espindola 5e46070516 Avoid aliases to weak aliases in interceptors.
The interceptors had code that after macro expansion ended up looking like

extern "C" void memalign()
    __attribute__((weak, alias("__interceptor_memalign")));
extern "C" void __interceptor_memalign() {}
extern "C" void __interceptor___libc_memalign()
    __attribute__((alias("memalign")));

That is,
* __interceptor_memalign is a function
* memalign is a weak alias to __interceptor_memalign
* __interceptor___libc_memalign is an alias to memalign

Both gcc and clang produce assembly that look like

__interceptor_memalign:
...
        .weak   memalign
memalign = __interceptor_memalign
        .globl  __interceptor___libc_memalign
__interceptor___libc_memalign = memalign

What it means in the end is that we have 3 symbols pointing to the
same position in the file, one of which is weak:

     8: 0000000000000000     1 FUNC    GLOBAL DEFAULT    1
__interceptor_memalign
     9: 0000000000000000     1 FUNC    WEAK   DEFAULT    1 memalign
    10: 0000000000000000     1 FUNC    GLOBAL DEFAULT    1
__interceptor___libc_memalign

In particular, note that __interceptor___libc_memalign will always
point to __interceptor_memalign, even if we do link in a strong symbol
for memalign. In fact, the above code produces exactly the same binary
as

extern "C" void memalign()
    __attribute__((weak, alias("__interceptor_memalign")));
extern "C" void __interceptor_memalign() {}
extern "C" void __interceptor___libc_memalign()
    __attribute__((alias("__interceptor_memalign")));

If nothing else, this patch makes it more obvious what is going on.

llvm-svn: 204823
2014-03-26 15:48:59 +00:00
Alexander Potapenko 1296436cbf [libsanitizer] Introduce flag descriptions.
Extend ParseFlag to accept the |description| parameter, add dummy values for all existing flags.
As the flags are parsed their descriptions are stored in a global linked list.
The tool can later call __sanitizer::PrintFlagDescriptions() to dump all the flag names and their descriptions.
Add the 'help' flag and make ASan, TSan and MSan print the flags if 'help' is set to 1.

llvm-svn: 204339
2014-03-20 12:52:52 +00:00
Kostya Serebryany cf5d8e4f29 AdjustStackSizeLinux() is used in Lsan, Tsan and Msan non-Linux-specific code so it seems it should have more generic name and moved to a common scope.
Renamed to AdjustStackSize.
Patch by Viktor Kutuzov.

llvm-svn: 202011
2014-02-24 08:53:26 +00:00
Alexey Samsonov 63a4af7346 [CMake] Add top-level target for each compiler-rt library, and add 'compiler-rt' target encompassing them all.
llvm-svn: 201556
2014-02-18 09:33:45 +00:00
Alexey Samsonov b73db72a17 [CMake] Simplify setting compile flag disabling RTTI
llvm-svn: 201547
2014-02-18 07:52:40 +00:00
Alexey Samsonov f1b7d05558 LSan: Print warning about dynamic linker only in verbose mode
llvm-svn: 201421
2014-02-14 15:12:46 +00:00
Alexey Samsonov 7eeef85bd4 Move LSan test suite under test/
llvm-svn: 201408
2014-02-14 12:26:05 +00:00
Alexey Samsonov 127e93e4dc Delete LSan unit tests
llvm-svn: 201406
2014-02-14 12:13:59 +00:00
Alexey Samsonov 81a2b466e9 Move shared configs for lit test suites to test/ and unittests/ directories
llvm-svn: 201399
2014-02-14 11:00:07 +00:00
Evgeniy Stepanov 2629e575db [sanitizer] Fix build.
llvm-svn: 201152
2014-02-11 13:45:01 +00:00
Nick Lewycky af2064f7ab Add throw() specifiers to more redeclarations of operator delete and operator delete[].
llvm-svn: 201016
2014-02-08 01:42:08 +00:00
Kostya Serebryany b965a2fa32 [lsan] simplify use_tls_dynamic.cc test
llvm-svn: 200748
2014-02-04 09:07:16 +00:00
Renato Golin 66e2b70e59 Enable compilation of RT on ARM
Adding the ARM RT sources to the CMake files, and enabling some
sanitizers to also build on ARM. This is far from supported or
production quality, but enabling it to build will get us errors
that we can actually fix.

Having said that, the Compiler-RT and the Asan libraries are
know to work on some variations of ARM.

llvm-svn: 200546
2014-01-31 14:25:58 +00:00
Sergey Matveev 9b068a2c06 [lsan] Expand a comment to document our dynamic TLS hack.
llvm-svn: 199900
2014-01-23 15:10:35 +00:00
Kostya Serebryany 06222915cf [lsan] remove LeakSanitizerIsTurnedOffForTheCurrentProcess (this was a bad idea), leave __lsan_is_turned_off
llvm-svn: 199304
2014-01-15 08:04:21 +00:00
Kostya Serebryany 2b76278d38 [lsan] handle 'new T[0]' where T is a type with DTOR; fixes https://code.google.com/p/address-sanitizer/issues/detail?id=257
llvm-svn: 198932
2014-01-10 10:48:01 +00:00
Kostya Serebryany 27f5b8800d [lsan] rename __lsan_is_turned_off to LeakSanitizerIsTurnedOffForTheCurrentProcess (leave the old variant for now for compatibility)
llvm-svn: 198921
2014-01-10 07:58:25 +00:00
Alexey Samsonov 23347de6ef [Sanitizer] Add Symbolizer::GetModuleNameAndOffsetForPC() and use it in LSan suppression matching.
This allows us to avoid retrieving file/line info for suppressed modules.

llvm-svn: 198025
2013-12-25 20:15:46 +00:00
Sergey Matveev 7eab734554 [lsan] Minor code health stuff.
- introduce logging macros
- fix incorrect vector construction

llvm-svn: 198021
2013-12-25 17:14:40 +00:00
Sergey Matveev 27aea0b0b7 [lsan] Refactor the LeakReport class.
Those methods were too damn bloated.

llvm-svn: 197978
2013-12-24 12:42:15 +00:00
Sergey Matveev 625875d256 [lsan] Make the report_objects flag more useful.
Print the list of leaked objects after each leak report. Previously we
printed only a joint list of all leaked objects. As a bonus, suppressed objects
are no longer reported.

llvm-svn: 197977
2013-12-24 12:03:02 +00:00
Evgeniy Stepanov 42cebb00da [sanitizer] Use the new sanitizer_interception.h header in all interceptors.
llvm-svn: 197808
2013-12-20 13:17:31 +00:00
Timur Iskhodzhanov b5f983e4b0 [Sanitizers] Rename Symbolizer::SymbolizeCode to Symbolizer::SymbolizePC
llvm-svn: 197569
2013-12-18 14:17:39 +00:00
Sergey Matveev 149b794ec6 [lsan] Remove the LSan-specific verbosity flag.
It conflicted with the verbosity flag we had in common flags. We don't need an
LSan-specific flag anyway.

Also, shift some logging levels and remove some unnecessary code.

llvm-svn: 197512
2013-12-17 18:18:32 +00:00
Sergey Matveev b5769dbb86 [lsan] Introduce print_suppressions flag.
Introduce a flag to either always or never print matched suppressions.
Previously, matched suppressions were printed unconditionally if there were
unsuppressed leaks. Also, verbosity=1 no longer has the semantics of "always
print suppressions and summary".

llvm-svn: 197510
2013-12-17 18:01:45 +00:00
Alexey Samsonov 25d010a98d [Sanitizer] Always initialize a Symbolizer (even if 'symbolize' is false).
If 'symbolize' flag is not set, we still want to transform virtual address
to module+offset pair in the call to Symbolizer::SymbolizeCode().
See https://code.google.com/p/address-sanitizer/issues/detail?id=251 for
more details.

llvm-svn: 197491
2013-12-17 11:15:39 +00:00
Sergey Matveev 7237879926 [lsan] Introduce __lsan_(un)register_root_region().
Add an interface for telling LSan that a region of memory is to be treated as a
source of live pointers. Useful for code which stores pointers in mapped memory.

llvm-svn: 197489
2013-12-17 11:11:23 +00:00
Alexey Samsonov 5069790b24 [LSan] Add a real test for suppressions file instead of duplicated test case.
llvm-svn: 197488
2013-12-17 11:02:52 +00:00
Sergey Matveev b1b8d1aa47 [lsan] Make LSan ignore memory poisoned by ASan.
Summary:
No more (potenital) false negatives due to red zones or fake stack
frames.

Reviewers: kcc, samsonov

Reviewed By: samsonov

CC: llvm-commits, samsonov

Differential Revision: http://llvm-reviews.chandlerc.com/D2359

llvm-svn: 196778
2013-12-09 13:12:10 +00:00
Dmitry Vyukov 7bd319cc08 tsan: fix flags parsing
- running_on_valgrind was not parsed in some contexts
- refactor code a bit
- add comprehensive tests for flags parsing

llvm-svn: 195831
2013-11-27 09:54:10 +00:00
Sergey Matveev 5170bca73c [lsan] Unbreak lsan_testlib.cc.
Also, add missing logging output.

llvm-svn: 195652
2013-11-25 15:54:31 +00:00
Sergey Matveev 6b0f6af374 [lsan] Add a missing file.
llvm-svn: 195643
2013-11-25 14:30:37 +00:00
Sergey Matveev 2d3f8d7840 [lsan] Unbreak standalone LSan's initialization by making it more like ASan's.
No longer allow interceptors to be called during initialization, use the preinit
array (instead of initializing at the first call to an intercepted function) and
adopt the calloc() hack from ASan.

llvm-svn: 195642
2013-11-25 14:25:36 +00:00
Sergey Matveev 10548681c0 [lsan] Use real memset to clear memory in standalone LSan.
Performance improvement. Also, the allocator was using CompactSizeClassMap for
no good reason, so I switched it to DefaultSizeClassMap.

llvm-svn: 195570
2013-11-24 14:28:18 +00:00
Alexey Samsonov 6345150992 [Sanitizer] Specify a default value for each common runtime flag
llvm-svn: 194479
2013-11-12 13:59:08 +00:00
Alexander Potapenko 49034e3c33 [ASan] Add CMake configs for libclang_rt.asan_iossim_dynamic.dylib
CMake changes to build the ASan runtime for the iOS simulator. This is a universal library targeting the same architectures as the OSX ASan runtime does, thus the iossim version can't live in the same universal libclang_rt.asan_osx_dynamic.dylib

The difference between the OSX and iossim builds is in the -mios-simulator-version-min and -ios_simulator_version_min flags that tell Clang to compile and link iossim code.

The iossim runtime can only be built on a machine with both Xcode and the iOS Simulator SDK installed. If xcodebuild -version -sdk iphonesimulator Path returns a nonempty path, it is used when compiling and linking the iossim runtime.

llvm-svn: 194199
2013-11-07 10:08:19 +00:00
Alexey Samsonov 5dc6cff06a [Sanitizer] Unify summary reporting across all sanitizers.
This change unifies the summary printing across sanitizers:
now each tool uses specific version of ReportErrorSummary() method,
which deals with symbolization of the top frame and formatting a
summary message. This change modifies the summary line for ASan+LSan mode:
now the summary mentions "AddressSanitizer" instead of "LeakSanitizer".

llvm-svn: 193864
2013-11-01 17:02:14 +00:00
Alexey Samsonov 4708c5912b Consistently use StackTrace::PrintStack in ASan, LSan and MSan
llvm-svn: 193834
2013-11-01 00:19:46 +00:00
Alexey Samsonov a687d2593f [Sanitizer] Simplify StackTrace::PrintStack interface: prefer common flags to turn on/off the symbolization
llvm-svn: 193587
2013-10-29 05:31:25 +00:00
Peter Collingbourne 791e65dcfb Overhaul the symbolizer interface.
This moves away from creating the symbolizer object and initializing the
external symbolizer as separate steps.  Those steps now always take place
together.

Sanitizers with a legacy requirement to specify their own symbolizer path
should use InitSymbolizer to initialize the symbolizer with the desired
path, and GetSymbolizer to access the symbolizer.  Sanitizers with no
such requirement (e.g. UBSan) can use GetOrInitSymbolizer with no need for
initialization.

The symbolizer interface has been made thread-safe (as far as I can
tell) by protecting its member functions with mutexes.

Finally, the symbolizer interface no longer relies on weak externals, the
introduction of which was probably a mistake on my part.

Differential Revision: http://llvm-reviews.chandlerc.com/D1985

llvm-svn: 193448
2013-10-25 23:03:29 +00:00
Nick Lewycky 993f1f38fd Continue to keep 'SUMMARY: ' prefix in lsan. Whoops!
llvm-svn: 193232
2013-10-23 07:58:11 +00:00
Sergey Matveev fd10073aeb [lsan] When detect_leaks=false, be completely silent.
In particular, don't make a fuss if we're passed a malformed suppressions file,
or if we have trouble identifying ld.so. Also, make LSan interface functions
no-ops in this case.

llvm-svn: 193108
2013-10-21 19:35:00 +00:00
Sergey Matveev 27ef175ef0 [lsan] Fix bug when discovering indirectly leaked objects.
If an object contains pointers to itself, that doesn't make it indirectly
leaked. D'oh!

llvm-svn: 192716
2013-10-15 16:00:11 +00:00
Dmitry Vyukov 7502a3a90c tsan: use verbosity flag in sanitizer_common code directly
now it's available from common_flags()

llvm-svn: 192705
2013-10-15 14:12:26 +00:00
Sergey Matveev 43d90cbd86 [lsan] Support ASan's stack-use-after-return mode in LSan.
Treat the fake stack as live memory.

llvm-svn: 192593
2013-10-14 14:04:50 +00:00
Sergey Matveev da9f5e7e7f [asan] Improve thread lifetime tracking on POSIX systems.
Call AsanThread::Destroy() from a late-running TSD destructor.
Previously we called it before any user-registered TSD destructors, which caused
false positives in LeakSanitizer.

llvm-svn: 192585
2013-10-14 12:01:05 +00:00
Alexey Samsonov c129e65661 [Sanitizer] Turn GetStackTrace() into StackTrace::Unwind()
llvm-svn: 192533
2013-10-12 12:23:00 +00:00
Sergey Matveev dc75cf3368 [sanitizer] Move the PTHREAD_DESTRUCTOR_ITERATIONS constant to sanitizer_linux.h.
Add a test.

llvm-svn: 192442
2013-10-11 12:09:49 +00:00