summaryrefslogtreecommitdiffstats
path: root/compiler-rt/test/ubsan_minimal
Commit message (Collapse)AuthorAgeFilesLines
* [UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined ↵Roman Lebedev2019-10-101-0/+22
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
* compiler-rt: Remove .cc from all lit config filesNico Weber2019-08-051-1/+1
| | | | | | All cc files have been renamed to cpp now. llvm-svn: 367911
* [ubsan][test] Don't disable ubsan testing on 64-bit Solaris/x86Rainer Orth2019-07-081-3/+0
| | | | | | | | | | | Unlike asan, which isn't supported yet on 64-bit Solaris/x86, there's no reason to disable ubsan. This patch does that, but keeps the 64-bit ubsan-with-asan tests disabled. Tested on x86_64-pc-solaris2.11. Differential Revision: https://reviews.llvm.org/D63982 llvm-svn: 365302
* [compiler-rt] Rename lit.*.cfg.* -> lit.*.cfg.py.*Reid Kleckner2019-06-273-4/+4
| | | | | | | | | | | | | These lit configuration files are really Python source code. Using the .py file extension helps editors and tools use the correct language mode. LLVM and Clang already use this convention for lit configuration, this change simply applies it to all of compiler-rt. Reviewers: vitalybuka, dberris Differential Revision: https://reviews.llvm.org/D63658 llvm-svn: 364591
* [compiler-rt][UBSan] Sanitization for alignment assumptions.Roman Lebedev2019-01-151-0/+17
| | | | | | | | | | | | | | | | | | | | | | | | | | | Summary: This is the compiler-rt part. The clang part is D54589. This is a second commit, the original one was r351106, which was mass-reverted in r351159 because 2 compiler-rt tests were failing. Now, i have fundamentally changed the testing approach: i malloc a few bytes, intentionally mis-align the pointer (increment it by one), and check that. Also, i have decreased the expected alignment. This hopefully should be enough to pacify all the bots. If not, i guess i might just drop the two 'bad' tests. Reviewers: filcab, vsk, #sanitizers, vitalybuka, rsmith, morehouse Reviewed By: morehouse Subscribers: rjmccall, krytarowski, rsmith, kcc, srhines, kubamracek, dberris, llvm-commits Tags: #sanitizers Differential Revision: https://reviews.llvm.org/D54590 llvm-svn: 351178
* Revert alignment assumptions changesVlad Tsyrklevich2019-01-151-11/+0
| | | | | | | Revert r351104-6, r351109, r351110, r351119, r351134, and r351153. These changes fail on the sanitizer bots. llvm-svn: 351159
* [compiler-rt][UBSan] Sanitization for alignment assumptions.Roman Lebedev2019-01-141-0/+11
| | | | | | | | | | | | | | | | | | Summary: This is the compiler-rt part. The clang part is D54589. Reviewers: filcab, vsk, #sanitizers, vitalybuka, rsmith, morehouse Reviewed By: morehouse Subscribers: rjmccall, krytarowski, rsmith, kcc, srhines, kubamracek, dberris, llvm-commits Tags: #sanitizers Differential Revision: https://reviews.llvm.org/D54590 llvm-svn: 351106
* [compiler-rt][ubsan] Implicit Conversion Sanitizer - integer sign change - ↵Roman Lebedev2018-10-302-0/+34
| | | | | | | | | | | | | | | | | | | | | | compiler-rt part Summary: This is a compiler-rt part. The clang part is D50250. See [[ https://bugs.llvm.org/show_bug.cgi?id=21530 | PR21530 ]], https://github.com/google/sanitizers/issues/940. Reviewers: vsk, filcab, #sanitizers Reviewed By: filcab, #sanitizers Subscribers: mclow.lists, srhines, kubamracek, dberris, rjmccall, rsmith, llvm-commits, regehr Tags: #sanitizers Differential Revision: https://reviews.llvm.org/D50251 llvm-svn: 345659
* [compiler-rt][ubsan] Split Implicit Integer Truncation Sanitizer into ↵Roman Lebedev2018-10-112-1/+26
| | | | | | | | | | | | | | | | unsigned and signed checks Summary: This is compiler-rt part. clang part is D50901. Reviewers: rsmith, vsk, filcab, Sanitizers Reviewed by: filcab Differential Revision: https://reviews.llvm.org/D50902 llvm-svn: 344231
* [NFC] Some small test updates for Implicit Conversion sanitizer.Roman Lebedev2018-08-171-1/+2
| | | | | | Split off from D50251. llvm-svn: 339996
* [compiler-rt][ubsan] Implicit Conversion Sanitizer - integer truncation - ↵Roman Lebedev2018-07-301-0/+24
| | | | | | | | | | | | | | | | | | | | | | compiler-rt part Summary: This is a compiler-rt part. The clang part is D48958. See [[ https://bugs.llvm.org/show_bug.cgi?id=21530 | PR21530 ]], https://github.com/google/sanitizers/issues/940. Reviewers: #sanitizers, samsonov, vsk, rsmith, pcc, eugenis, kcc, filcab Reviewed By: #sanitizers, vsk, filcab Subscribers: llvm-commits, eugenis, filcab, kubamracek, dberris, #sanitizers, regehr Tags: #sanitizers Differential Revision: https://reviews.llvm.org/D48959 llvm-svn: 338287
* [UBsan] minimal unit tests removing OpenBSD spacial caseDavid Carlier2018-06-281-6/+0
| | | | | | | | | | | | The OpenBSD driver now handles sanitisers with ld.lld automatically. Reviewers: vitalybuka Reviewed By: vitalybuka Differential Revision: https://reviews.llvm.org/D48740 llvm-svn: 335923
* [UBsan] Enable ubsan minimal unit tests on OpenBSDDavid Carlier2018-06-251-1/+7
| | | | | | | | | | | | | OpenBSD needs lld linker for sanitisers. Disabling lint checking as some symbols cannot be defined and block the proper unit tests launch. Reviewers: lebedev.ri, vitalybuka Reviewed By: vitalybuka Differential Revision: https://reviews.llvm.org/D48528 llvm-svn: 335524
* [Sanitizers, test] Fix sanitizer tests on Solaris (PR 33274)Kamil Rytarowski2018-01-171-0/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Summary: This patch (on top of the previous two (https://reviews.llvm.org/D40898 and https://reviews.llvm.org/D40899) complete the compiler-rt side of the the Solaris sanitizer port. It contains the following sets of changes: * For the time being, the port is for 32-bit x86 only, so reject the various tests on x86_64. * When compiling as C++, <setjmp.h> resp. <iso/setjmp_iso.h> only declares _setjmp and _longjmp inside namespace std. * MAP_FILE is a Windows feature. While e.g. Linux <sys/mman.h> provides a no-op compat define, Solaris does not. * test/asan/TestCases/Posix/coverage.cc was initially failing like this: /vol/gcc/src/llvm/llvm/local/projects/compiler-rt/lib/sanitizer_common/scripts/sancov.py: 4 files merged; 2 PCs total rm: cannot remove '/var/gcc/llvm/local/projects/compiler-rt/test/asan/I386SunOSConfig/TestCases/Posix/Output/coverage': Invalid argument Further digging revealed that the rm was trying to remove the running test's working directory which failed as observed. cd'ing out of the dir before let the test pass. * Two tests needed a declaration of alloca. I've now copied the existing code from test/asan/TestCases/alloca_constant_size.cc, but it may be more profitable and maintainable to have a common testsuite header where such code is collected. * Similarly, Solaris' printf %p format doesn't include the leading 0x. * In test/asan/TestCases/malloc-no-intercept.c, I had to undef __EXTENSIONS__ (predefined by clang for no apparent reason) to avoid conflicting declarations for memalign. * test/ubsan/TestCases/Float/cast-overflow.cpp has different platform dependent ways to define BYTE_ORDER and friends. Why not just use __BYTE_ORDER__ and friends as predefined by clang and gcc? Patch by Rainer Orth. Reviewers: kcc, alekseyshl Reviewed By: alekseyshl Subscribers: srhines, kubamracek, mgorny, krytarowski, fedor.sergeev, JDevlieghere, llvm-commits, #sanitizers Tags: #sanitizers Differential Revision: https://reviews.llvm.org/D40900 llvm-svn: 322635
* [ubsan] Re-commit: lit changes for lld testing, future lto testing.Roman Lebedev2017-12-011-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Summary: As discussed in https://github.com/google/oss-fuzz/issues/933, it would be really awesome to be able to use ThinLTO for fuzzing. However, as @kcc has pointed out, it is currently undefined (untested) whether the sanitizers actually function properly with LLD and/or LTO. This patch is inspired by the cfi test, which already do test with LTO (and/or LLD), since LTO is required for CFI to function. I started with UBSan, because it's cmakelists / lit.* files appeared to be the cleanest. This patch adds the infrastructure to easily add LLD and/or LTO sub-variants of the existing lit test configurations. Also, this patch adds the LLD flavor, that explicitly does use LLD to link. The check-ubsan does pass on my machine. And to minimize the [initial] potential buildbot breakage i have put some restrictions on this flavour. Please review carefully, i have not worked with lit/sanitizer tests before. The original attempt, r319525 was reverted in r319526 due to the failures in compiler-rt standalone builds. Reviewers: eugenis, vitalybuka Reviewed By: eugenis Subscribers: #sanitizers, pcc, kubamracek, mgorny, llvm-commits, mehdi_amini, inglorion, kcc Differential Revision: https://reviews.llvm.org/D39508 llvm-svn: 319575
* Revert "[ubsan] lit changes for lld testing, future lto testing."Roman Lebedev2017-12-011-1/+1
| | | | | | | | | | | | | | | | | | | | This reverts commit r319525. This change has introduced a problem with the Lit tests build for compiler-rt using Gold: http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux/builds/6047/steps/test%20standalone%20compiler-rt/logs/stdio llvm-lit: /b/sanitizer-x86_64-linux/build/llvm/utils/lit/lit/TestingConfig.py:101: fatal: unable to parse config file '/b/sanitizer-x86_64-linux/build/llvm/projects/compiler-rt/test/profile/Linux/lit.local.cfg', traceback: Traceback (most recent call last): File "/b/sanitizer-x86_64-linux/build/llvm/utils/lit/lit/TestingConfig.py", line 88, in load_from_path exec(compile(data, path, 'exec'), cfg_globals, None) File "/b/sanitizer-x86_64-linux/build/llvm/projects/compiler-rt/test/profile/Linux/lit.local.cfg", line 37, in <module> if root.host_os not in ['Linux'] or not is_gold_linker_available(): File "/b/sanitizer-x86_64-linux/build/llvm/projects/compiler-rt/test/profile/Linux/lit.local.cfg", line 27, in is_gold_linker_available stderr = subprocess.PIPE) File "/usr/lib/python2.7/subprocess.py", line 390, in __init__ errread, errwrite) File "/usr/lib/python2.7/subprocess.py", line 1024, in _execute_child raise child_exception OSError: [Errno 2] No such file or directory llvm-svn: 319529
* [ubsan] lit changes for lld testing, future lto testing.Roman Lebedev2017-12-011-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Summary: As discussed in https://github.com/google/oss-fuzz/issues/933, it would be really awesome to be able to use ThinLTO for fuzzing. However, as @kcc has pointed out, it is currently undefined (untested) whether the sanitizers actually function properly with LLD and/or LTO. This patch is inspired by the cfi test, which already do test with LTO (and/or LLD), since LTO is required for CFI to function. I started with UBSan, because it's cmakelists / lit.* files appeared to be the cleanest. This patch adds the infrastructure to easily add LLD and/or LTO sub-variants of the existing lit test configurations. Also, this patch adds the LLD flavor, that explicitly does use LLD to link. The check-ubsan does pass on my machine. And to minimize the [initial] potential buildbot breakage i have put some restrictions on this flavour. Please review carefully, i have not worked with lit/sanitizer tests before. Reviewers: eugenis, vitalybuka Reviewed By: eugenis Subscribers: #sanitizers, pcc, kubamracek, mgorny, llvm-commits, mehdi_amini, inglorion, kcc Differential Revision: https://reviews.llvm.org/D39508 llvm-svn: 319525
* [ubsan] Support ubsan-minimal tests on Android.Evgeniy Stepanov2017-10-111-1/+1
| | | | llvm-svn: 315507
* [ubsan-minimal] Test exported symbol set against RTUBsanVedant Kumar2017-09-191-0/+16
| | | | | | | | | | | Check that the symbol sets exported by the minimal runtime and the full runtime match (making exceptions for special cases as needed). This test uses some possibly non-standard nm options, and needs to inspect the symbols in runtime dylibs. I haven't found a portable way to do this, so it's limited to x86-64/Darwin for now. llvm-svn: 313615
* [ubsan-minimal] Make the interface more compatible with RTUBSanVedant Kumar2017-09-191-1/+15
| | | | | | | | | | | | | | This eliminates a few inconsistencies between the symbol sets exported by RTUBSan and RTUBSan_minimal: * Handlers for nonnull_return were missing from the minimal RT, and are now added in. * The minimal runtime exported recoverable handlers for builtin_unreachable and missing_return. These are not supposed to exist, and are now removed. llvm-svn: 313614
* [ubsan-minimal] Filter targets to test by host arch on DarwinVedant Kumar2017-09-132-5/+3
| | | | | | | This reverts r313189, and adds a use of darwin_filter_host_archs() for ubsan-minimal. llvm-svn: 313206
* [ubsan-minimal] Temporarily disable x86_64h testing on DarwinVedant Kumar2017-09-131-0/+5
| | | | | | | | | | | | | | | We're seeing strange issues on the public GreenDragon Darwin bots which we don't understand. x86_64h tests are still being run on pre-Haswell bots despite the added checks in test/ubsan_minimal/lit.common.cfg, which were verified on our internal bots. I'm unable to ssh into the affected public bot, so for now am trying a more aggressive check which disables all x86_64h testing for ubsan-minimal on Darwin. rdar://problem/34409349 llvm-svn: 313189
* [ubsan-minimal] Fix the x86_64h config checkVedant Kumar2017-09-121-1/+1
| | | | | | | | | | | | Checking if config.target_arch is x86_64h doesn't work (the 'h' suffix is dropped here, and I didn't account for that). Instead, check to see if '-arch x86_64h' is in the cflags. Tested on a pre-Haswell bot. rdar://problem/34378605 llvm-svn: 313053
* [ubsan-minimal] Disable x86_64h tests when not on x86_64hVedant Kumar2017-09-121-0/+4
| | | | llvm-svn: 312982
* [ubsan-minimal] Enable on DarwinVedant Kumar2017-09-111-1/+1
| | | | | | | | Testing: check-ubsan-minimal Differential Revision: https://reviews.llvm.org/D37646 llvm-svn: 312959
* [ubsan] Give ubsan-minimal lit test suite a name.Evgeniy Stepanov2017-08-311-0/+1
| | | | | | | | | Otherwise llvm-lit -v prints this: PASS: <unnamed> :: TestCases/recover-dedup-limit.cpp (1 of 3) PASS: <unnamed> :: TestCases/recover-dedup.cpp (2 of 3) PASS: <unnamed> :: TestCases/uadd-overflow.cpp (3 of 3) llvm-svn: 312203
* Disable ubsan-minimal on Darwin.Evgeniy Stepanov2017-08-291-1/+1
| | | | | | | Should un-break this bot: http://green.lab.llvm.org/green//job/clang-stage1-configure-RA_build/38264/consoleFull llvm-svn: 312036
* Minimal runtime for UBSan.Evgeniy Stepanov2017-08-296-0/+145
Summary: An implementation of ubsan runtime library suitable for use in production. Minimal attack surface. * No stack traces. * Definitely no C++ demangling. * No UBSAN_OPTIONS=log_file=/path (very suid-unfriendly). And no UBSAN_OPTIONS in general. * as simple as possible Minimal CPU and RAM overhead. * Source locations unnecessary in the presence of (split) debug info. * Values and types (as in A+B overflows T) can be reconstructed from register/stack dumps, once you know what type of error you are looking at. * above two items save 3% binary size. When UBSan is used with -ftrap-function=abort, sometimes it is hard to reason about failures. This library replaces abort with a slightly more informative message without much extra overhead. Since ubsan interface in not stable, this code must reside in compiler-rt. Reviewers: pcc, kcc Subscribers: srhines, mgorny, aprantl, krytarowski, llvm-commits Differential Revision: https://reviews.llvm.org/D36810 llvm-svn: 312029
OpenPOWER on IntegriCloud