This reverts commit r221445. This change leads to false positives
reports from -fsanitize=vptr. See original commit thread for more
details.
llvm-svn: 224972
The change in r224819 started using internal_unlink in a sanitizer_common unit test. For some reason, internal_unlink is not defined in sanitizer_mac.cc, fixing that.
llvm-svn: 224910
This increases the limit from 4M locations to 16M, reserving
64Mb virtual memory. Chrome has >5M unique coverage locations with coverage=3.
llvm-svn: 224855
This is a re-commit of r224838 + r224839, previously reverted in r224850.
Test failures were likely (still can not reproduce) caused by two lit tests
using the same name for an intermediate build target.
llvm-svn: 224853
The current code leaves the first event in the trace part uninitialized
(from the previous thread). It can cause unpredictable behavior
during stack/mutexset restoration.
Initialize the first event to a fake harmless memory access.
llvm-svn: 224834
This is mostly useful for testing, as the only other way of specifying
activation options (Android system property) is system-wide and affects
concurrently running tests.
llvm-svn: 224824
Summary:
Protect CommonFlags singleton by adding const qualifier to
common_flags() accessor. The only ways to modify the flags are
SetCommonFlagsDefaults(), ParseCommonFlagsFromString() and
OverrideCommonFlags() functions, which are only supposed to be
called during initialization.
Test Plan: regression test suite
Reviewers: kcc, eugenis, glider
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D6741
llvm-svn: 224736
As mentioned in
https://code.google.com/p/address-sanitizer/issues/detail?id=365, when the
re-exec that adds the required DYLD_INSERT_LIBRARIES variable fails, ASan
currently continues to run, but things are broken (some memory can be
overwritten, interceptors don't work, ...). This patch aborts if the execv()
fails and prints an error message that DYLD_INSERT_LIBRARIES is required. It
also removes the "alllow_reexec" flag, since using it causes the same issues.
Reviewed at http://reviews.llvm.org/D6752
llvm-svn: 224712
NULL handler
Per
https://developer.apple.com/library/mac/documentation/Performance/Reference/GCD_libdispatch_Ref/index.html,
the dispatch_source_set_cancel_handler() API *can* be called with a NULL
handler. In that case, the libdispatch removes an already existing cancellation
handler, if there was one. ASan's interceptor always creates a new block that
always tries to call the original handler. In case the original block is NULL,
a segmentation fault happens. Let's fix that by not wrapping a NULL-block at
all.
It looks like all the other libdispatch APIs (which we intercept) do *not*
allow NULL. So it's really only the dispatch_source_set_cancel_handler one that
needs this fix.
Reviewed at http://reviews.llvm.org/D6747
llvm-svn: 224711
Add CommonFlags::SetDefaults() and CommonFlags::ParseFromString(),
so that this object can be easily tested. Enforce
that ParseCommonFlagsFromString() and SetCommonFlagsDefaults()
work only with singleton CommonFlags, shared across all sanitizer
runtimes.
llvm-svn: 224617
Now ASan deactivation doesn't modify common or ASan-specific runtime
flags. Flags stay constant after initialization, and "deactivation"
instead stashes initialized runtime state, and deactivates the
runtime. Activation then just restores the original state (possibly,
overriden by some activation flags provided in system property on
Android).
llvm-svn: 224614
Summary:
Reduce the dependency of allocator code on runtime flags. Instead,
pass a bunch of options that configure allocator behavior at
initialization or re-initialization. That would allow us to
cleaner modify allocator behavior during a program execution
when ASan is activated or de-activated.
Test Plan: regression test suite
Reviewers: kcc
Subscribers: llvm-commits, eugenis
Differential Revision: http://reviews.llvm.org/D6711
llvm-svn: 224605
Summary:
Change the way ASan flag parsing is performed at activation: instead
of overwriting all ASan and common flags from the activation string,
ASan will now only override a handful of whitelisted flags.
This is a first step towards making runtime flags immutable after
initialization. I plan to re-factor the activation stragegy to
the following one:
- Parse commandline flags. Override the defaults from compile
definition, env var, etc. Flags are immutable at this point.
- Initiailize the runtime from commandline flags.
- If ASan needs to be started deactivated, stash the runtime
state in "asan_deactivated_flags" and deactivate the runtime.
- When ASan is activated, override "asan_deactivated_flags" with
activation flags, and use it to re-activate the runtime.
Test Plan: regression test suite
Reviewers: eugenis, kcc
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D6709
llvm-svn: 224601
clang does not like the definition of builtins. In order to work around this,
we use a SUN CC to redefine the generated name. However, this requires that we
account for the user label prefix. Rather than hard coding that into the file,
rely on the compiler to tell us the information and use the preprocessor to
generate the name as we do in the assembly routines. NFC.
llvm-svn: 224597
signal handler reads sa_sigaction when a concurrent sigaction call can modify it
as the result in could try to call SIG_DFL or a partially overwritten function pointer
llvm-svn: 224530
Summary:
The numbers in /proc/self/statm are in pages, not in fixed 4k units.
This fixes Linux/hard_rss_limit_mb_test.cc on my PowerPC64 box which
has 64k pages.
Reviewers: kcc, willschm
Reviewed By: willschm
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D6717
llvm-svn: 224522
tctx==NULL crash observed during deadlock reporting.
There seems to be some bugs in the deadlock detector,
but it is still useful to be more robust during reporting.
llvm-svn: 224508
This commit changes the strategy for building shared ASan runtime
and the way we test it:
- COMPILER_RT_BUILD_SHARED_ASAN CMake option is removed. We now
always build shared ASan runtime (it is the default on Android,
Windows and Mac, and not the default on Linux and FreeBSD).
- Platforms, which use static runtime by default now have
"check-asan-dynamic" testsuite. This testsuite contains instrumented
unit tests, and ASan lit tests, and runs them with shared ASan
runtime. This testsuite is *not* a part of "check-asan" and
*not* a part of "check-all", as adding 1000 more test cases, which
duplicate existing ones is costly. However, you're welcome to
add this command to your buildbot.
llvm-svn: 224470
pthread_getspecific is not async-signal-safe.
MsanThread pointer is now stored in a TLS variable, and the TSD slot
is used only for its destructor, and never from a signal handler.
This should fix intermittent CHECK failures in MsanTSDSet.
llvm-svn: 224423
Introduce "Allocator" object, which contains all the bits and pieces
ASan allocation machinery actually use: allocator from sanitizer_common,
quarantine, fallback allocator and quarantine caches, fallback mutex.
This step is a preparation to adding more state to this object. We want
to reduce dependency of Allocator on commandline flags and be able to
"safely" modify its behavior (such as the size of the redzone) at
runtime.
llvm-svn: 224406
SetCanPoisonMemory()/CanPoisonMemory() functions are now used
instead of "poison_heap" flag to determine if ASan is allowed
to poison the shadow memory. This allows to hot-patch this
value in runtime (e.g. during ASan activation) without introducing
a data race.
llvm-svn: 224395
We recently had a broken version check because an newer OS X version is treated as MACOS_VERSION_UNKNOWN which is less than all the defined values. Let's have a separate enum value for unknown but newer versions, so the ">=" and "<=" version checks still work even in upcoming OS X releases.
Reviewed at http://reviews.llvm.org/D6137
llvm-svn: 224315
Summary:
Introduce __asan::malloc_context_size atomic that is used to determine
required malloc/free stack trace size. It is initialized with
common_flags()->malloc_context_size flag, but can later be overwritten
at runtime (e.g. when ASan is activated / deactivated).
Test Plan: regression test suite
Reviewers: kcc, eugenis
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D6645
llvm-svn: 224305
Summary:
Turn "allocator_may_return_null" common flag into an
Allocator::may_return_null bool flag. We want to make sure
that common flags are immutable after initialization. There
are cases when we want to change this flag in the allocator
at runtime: e.g. in unit tests and during ASan activation
on Android.
Test Plan: regression test suite, real-life applications
Reviewers: kcc, eugenis
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D6623
llvm-svn: 224148
Enabling COMPILER_RT_INCLUDE_TESTS and updating tests/sanitizer_allocator_test.cc to remove Allocator64 related tests for MIPS.
Reviewed By: samsonov
llvm-svn: 224101
This commit changes the way we get fake stack from ASan runtime
(to find use-after-return errors) and the way we represent local
variables:
- __asan_stack_malloc function now returns pointer to newly allocated
fake stack frame, or NULL if frame cannot be allocated. It doesn't
take pointer to real stack as an input argument, it is calculated
inside the runtime.
- __asan_stack_free function doesn't take pointer to real stack as
an input argument. Now this function is never called if fake stack
frame wasn't allocated.
- __asan_init version is bumped to reflect changes in the ABI.
- new flag "-asan-stack-dynamic-alloca" allows to store all the
function local variables in a dynamic alloca, instead of the static
one. It reduces the stack space usage in use-after-return mode
(dynamic alloca will not be called if the local variables are stored
in a fake stack), and improves the debug info quality for local
variables (they will not be described relatively to %rbp/%rsp, which
are assumed to be clobbered by function calls). This flag is turned
off by default for now, but I plan to turn it on after more
testing.
llvm-svn: 224063
Summary:
- Make sure mmap() is never called inside RawWrite function.
- Wrap a bunch of standalone globals in a ReportFile object.
- Make sure accesses to these globals are thread-safe.
- Fix report_path functionality on Windows, where
__sanitizer_set_report_path() would break program.
I've started this yak shaving in order to make
"CommonFlags::mmap_limit_mb" immutable. Currently we drop this flag
to zero before printing an error message.
Test Plan: regression test suite
Reviewers: kcc, glider
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D6595
llvm-svn: 224031
Each of the object formats use a different directive for selecting the constant
section. Use a macro to avoid the duplication across a number of files. Also
correct a small macro mismatch on the Windows case (HIDDEN_DIRECTIVE -> HIDDEN).
Patch by Vadim Chugunov!
llvm-svn: 223910
On Darwin, compiler_rt uses magic linker symbols to find the profile
counters in the __DATA segment. This is a reasonable method for
normal, hosted, userspace programs. However programs with custom
memory layouts, such as the kernel, will need to tell compiler_rt
explicitly where to find these sections.
Patch by Lawrence D'Anna. Thanks!
llvm-svn: 223840
In the current scheme of things, the call to ThreadStart() in the child
thread is not synchronized with the parent thread. So, if a pointer is passed to
pthread_create, there may be a window of time during which this pointer will not
be discoverable by LSan. I.e. the pthread_create interceptor has already
returneed and thus the pointer is no longer on the parent stack, but we don't
yet know the location of the child stack. This has caused bogus leak reports
(see http://llvm.org/bugs/show_bug.cgi?id=21621/).
This patch makes the pthread_create interceptor wait until the child thread is
properly registered before returning.
llvm-svn: 223419
Previously, all origin ids were "chained" origins, i.e values of
ChainedOriginDepot. This added a level of indirection for simple
stack and heap allocation, which were represented as chains of
length 1. This costs both RAM and CPU, but provides a joined 2**29
origin id space. It also made function (any instrumented function)
entry non-async-signal-safe, but that does not really matter because
memory stores in track-origins=2 mode are not async-signal-safe anyway.
With this change, the type of the origin is encoded in origin id.
See comment in msan_origin.h for more details. This reduces chained and stack
origin id range to 2**28 each, but leaves extra 2**31 for heap origins.
This change should not have any user-visible effects.
llvm-svn: 223233
Alexey Samsonov