1 Release 3.21.0 (?? Apr 2023)
2 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
5 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
6 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
7 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
8 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
9 AMD64/macOS 10.13 and nanoMIPS/Linux.
11 * ==================== CORE CHANGES ===================
13 * Make the address space limit on FreeBSD amd64 128Gbytes
14 (the same as Linux and Solaris, it was 32Gbytes)
16 * ==================== TOOL CHANGES ===================
19 - The option ---history-backtrace-size=<number> allows to configure
20 the number of entries to record in the stack traces of "old"
21 accesses. Previously, this number was hardcoded to 8.
23 * ==================== FIXED BUGS ====================
25 The following bugs have been fixed or resolved. Note that "n-i-bz"
26 stands for "not in bugzilla" -- that is, a bug that was reported to us
27 but never got a bugzilla entry. We encourage you to file bugs in
28 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
29 than mailing the developers (or mailing lists) directly -- bugs that
30 are not entered into bugzilla tend to get forgotten about or ignored.
32 170510 Don't warn about ioctl of size 0 without direction hint
33 351857 confusing error message about valid command line option
34 392331 Spurious lock not held error from inside pthread_cond_timedwait
35 400793 pthread_rwlock_timedwrlock false positive
36 444110 priv/guest_ppc_toIR.c:36198:31: warning: duplicated 'if' condition.
37 444488 Use glibc.pthread.stack_cache_size tunable
38 444568 drd/tests/pth_barrier_thr_cr fails on Fedora 38
39 459476 vgdb: allow address reuse to avoid "address already in use" errorsuse" errors
40 462830 WARNING: unhandled amd64-freebsd syscall: 474
41 463027 broken check for MPX instruction support in assembler
43 To see details of a given bug, visit
44 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
45 where XXXXXX is the bug number as listed above.
47 (3.21.0.RC1: ?? Apr 2023)
49 Release 3.20.0 (24 Oct 2022)
50 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
52 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
53 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
54 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
55 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
56 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
57 AMD64/macOS 10.13 and nanoMIPS/Linux.
59 * ==================== CORE CHANGES ===================
61 * The option "--vgdb-stop-at=event1,event2,..." accepts the new value abexit.
62 This indicates to invoke gdbserver when your program exits abnormally
63 (i.e. with a non zero exit code).
64 * Fix Rust v0 name demangling.
65 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
66 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
67 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with
69 * The option --enable-debuginfod=<no|yes> [default: yes] has been added on
71 * More DWARF5 support as generated by clang14.
73 * ==================== FIXED BUGS ====================
75 The following bugs have been fixed or resolved. Note that "n-i-bz"
76 stands for "not in bugzilla" -- that is, a bug that was reported to us
77 but never got a bugzilla entry. We encourage you to file bugs in
78 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
79 than mailing the developers (or mailing lists) directly -- bugs that
80 are not entered into bugzilla tend to get forgotten about or ignored.
82 131186 writev reports error in (vector[...])
83 434764 iconv_open causes ld.so v2.28+ to use optimised strncmp
84 446754 Improve error codes from alloc functions under memcheck
85 452274 memcheck crashes with Assertion 'sci->status.what == SsIdle' failed
86 452779 Valgrind fails to build on FreeBSD 13.0 with llvm-devel (15.0.0)
87 453055 shared_timed_mutex drd test fails with "Lock shared failed" message
88 453602 Missing command line option to enable/disable debuginfod
89 452802 Handle lld 9+ split RW PT_LOAD segments correctly
90 454040 s390x: False-positive memcheck:cond in memmem on arch13 systems
91 456171 [PATCH] FreeBSD: Don't record address errors when accessing the 'kern.ps_strings' sysctl struct
92 n-i-bz Implement vgdb invoker on FreeBSD
93 458845 PowerPC: The L field for the dcbf and sync instruction should be
95 458915 Remove register cache to fix 458915 gdbserver causes wrong syscall return
96 459031 Documentation on --error-exitcode incomplete
97 459477 XERROR messages lacks ending '\n' in vgdb
98 462007 Implicit int in none/tests/faultstatus.c
100 To see details of a given bug, visit
101 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
102 where XXXXXX is the bug number as listed above.
104 (3.20.0.RC1: 20 Oct 2022)
107 Release 3.19.0 (11 Apr 2022)
108 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
110 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
111 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
112 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
113 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
114 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
115 AMD64/macOS 10.13 and nanoMIPS/Linux.
117 * ==================== CORE CHANGES ===================
119 * Fix Rust v0 name demangling.
120 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
121 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
122 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with HardenedBSD
124 * ================== PLATFORM CHANGES =================
127 - ignore the "v8.x" architecture levels, only look at actual CPU features
128 present. Fixes mismatch detected between RDMA and atomics features
129 preventing startup on some QEMU configurations.
130 - Implement LD{,A}XP and ST{,L}XP
131 - Fix incorrect code emitted for doubleword CAS.
134 - Fix sys_ipc semtimedop syscall
135 - Fix VFLRX and WFLRX instructions
136 - Fix EXRL instruction with negative offset
139 - Reimplement the vbpermq instruction support to generate less Iops and
140 avoid overflowing internal buffers.
141 - Fix checking for scv support to avoid "Facility 'SCV' unavailable (12),
142 exception" messages in dmsg.
143 - Fix setting condition code for Vector Compare quad word instructions.
144 - Fix fix lxsibzx, lxsihzx and lxsihzx instructions so they only load
145 their respective sized data.
146 - Fix the prefixed stq instruction in PC relative mode.
148 * ==================== TOOL CHANGES ===================
151 - Speed up --track-origins=yes for large (in the range of hundreds to
152 thousands of megabytes) mmap/munmaps.
154 - Several fixes for new versions of libstd++ using new posix try_lock
157 * ==================== FIXED BUGS ====================
159 The following bugs have been fixed or resolved. Note that "n-i-bz"
160 stands for "not in bugzilla" -- that is, a bug that was reported to us
161 but never got a bugzilla entry. We encourage you to file bugs in
162 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
163 than mailing the developers (or mailing lists) directly -- bugs that
164 are not entered into bugzilla tend to get forgotten about or ignored.
166 403802 leak_cpp_interior fails with some reachable blocks different than expected
167 435732 memcheck/tests/leak_cpp_interior fails with gcc11
168 444242 s390x: Valgrind crashes on EXRL with negative offset
169 444399 arm64: unhandled instruction 0xC87F2D89 (LD{,A}XP and ST{,L}XP).
171 444481 gdb_server test failures on s390x
172 444495 dhat/tests/copy fails on s390x
173 444552 memcheck/tests/sem fails on s390x with glibc 2.34
174 444571 PPC, fix the lxsibzx and lxsihzx so they only load their respective
176 444836 PPC, pstq instruction for R=1 is not storing to the correct address.
177 444925 fexecve syscall wrapper not properly implemented
178 445032 valgrind/memcheck crash with SIGSEGV when SIGVTALRM timer used and
180 445211 Fix out of tree builds
181 445300 [PATCH] Fix building tests with Musl
182 445011 SIGCHLD is sent when valgrind uses debuginfod-find
183 445354 arm64 backend: incorrect code emitted for doubleword CAS
184 445415 arm64 front end: alignment checks missing for atomic instructions
185 445504 Using C++ condition_variable results in bogus "mutex is locked simultaneously by two threads" warning
186 445607 Unhandled amd64-freebsd syscall: 247
187 445668 Inline stack frame generation is broken for Rust binaries
188 445916 Demangle Rust v0 symbols with .llvm suffix
189 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false positives
190 446138 DRD/Helgrind with std::timed_mutex::try_lock_until false positives
191 446281 Add a DRD suppression for fwrite
192 446103 Memcheck: `--track-origins=yes` causes extreme slowdowns for large mmap/munmap
193 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false
194 446251 TARGET_SIGNAL_THR added to enum target_signal
195 446823 FreeBSD - missing syscalls when using libzm4
196 447991 s390x: Valgrind indicates illegal instruction on wflrx
197 447995 Valgrind segfault on power10 due to hwcap checking code
198 449483 Powerpc: vcmpgtsq., vcmpgtuq,, vcmpequq. instructions not setting the
199 condition code correctly.
200 449672 ppc64 --track-origins=yes failures because of bad cmov addHRegUse
201 449838 sigsegv liburing the 'impossible' happened for io_uring_setup
202 450025 Powerc: ACC file not implemented as a logical overlay of the VSR
204 450437 Warn for execve syscall with argv or argv[0] being NULL
205 450536 Powerpc: valgrind throws 'facility scv unavailable exception'
206 451626 Syscall param bpf(attr->raw_tracepoint.name) points to unaddressable byte(s)
207 451827 [ppc64le] VEX temporary storage exhausted with several vbpermq instructions
208 451843 valgrind fails to start on a FreeBSD system which enforces W^X
210 To see details of a given bug, visit
211 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
212 where XXXXXX is the bug number as listed above.
214 (3.19.0.RC1: 02 Apr 2022)
215 (3.19.0.RC2: 08 Apr 2022)
218 Release 3.18.0 (15 Oct 2021)
219 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
221 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
222 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
223 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
224 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
225 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
226 AMD64/macOS 10.13 and nanoMIPS/Linux.
228 * ==================== CORE CHANGES ===================
230 * The libiberty demangler has been updated, which brings support for
231 Rust v0 name demangling. [Update: alas, due to a bug, this support
232 isn't working in 3.18.0.]
234 * __libc_freeres isn't called anymore after the program recieves a
235 fatal signal. Causing some internal glibc resources to hang around,
236 but preventing any crashes after the program has ended.
238 * The DWARF reader is now very much faster at startup when just
239 --read-inline-info=yes (the default in most cases) is given.
241 * glibc 2.34, which moved various functions from libpthread.so into
242 libc.so, is now supported.
244 * ================== PLATFORM CHANGES =================
248 - v8.2 scalar and vector FABD, FACGE, FACGT and FADD.
249 - v8.2 FP compare & conditional compare instructions.
250 - Zero variants of v8.2 FP compare instructions.
254 - Support the miscellaneous-instruction-extensions facility 3 and
255 the vector-enhancements facility 2. This enables programs
256 compiled with "-march=arch13" or "-march=z15" to be executed
261 - ISA 3.1 support is now complete
262 - ISA 3.0 support for the darn instruction added.
263 - ISA 3.0 support for the vector system call instruction scv added.
264 - ISA 3.0 support for the copy, paste and cpabort instructions added.
266 * Support for X86/FreeBSD and AMD64/FreeBSD has been added.
268 * ==================== OTHER CHANGES ====================
270 * Memcheck on amd64: minor fixes to remove some false positive
273 * ==================== FIXED BUGS ====================
275 The following bugs have been fixed or resolved. Note that "n-i-bz"
276 stands for "not in bugzilla" -- that is, a bug that was reported to us
277 but never got a bugzilla entry. We encourage you to file bugs in
278 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
279 than mailing the developers (or mailing lists) directly -- bugs that
280 are not entered into bugzilla tend to get forgotten about or ignored.
282 208531 [PATCH]: FreeBSD support for valgrind
283 368960 WARNING: unhandled amd64-linux syscall: 163 (acct)
284 407589 [Linux] Add support for C11 aligned_alloc() and GNU reallocarray()
285 423963 Error in child thread when CLONE_PIDFD is used
286 426148 crash with "impossible happened" when running BPF CO-RE programs
287 429375 PPC ISA 3.1 support is missing, part 9
288 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
289 431306 Update demangler to support Rust v0 name mangling
290 432387 s390x: z15 instructions support
291 433437 FreeBSD support, part 1
292 433438 FreeBSD support, part 2
293 433439 FreeBSD support, part 3
294 433469 FreeBSD support, part 4
295 433473 FreeBSD support, part 5
296 433477 FreeBSD support, part 6
297 433479 FreeBSD support, part 7
298 433504 FreeBSD support, part 8
299 433506 FreeBSD support, part 9
300 433507 FreeBSD support, part 10
301 433508 FreeBSD support, part 11
302 433510 FreeBSD support, part 12
303 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
304 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
305 434296 s390x: False-positive memcheck diagnostics from vector string
307 434840 PPC64 darn instruction not supported
308 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
309 435908 valgrind tries to fetch from deubginfod for files which already
310 have debug information
311 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2
312 439046 valgrind is unusably large when linked with lld
313 439090 Implement close_range(2)
314 439326 Valgrind 3.17.0 won't compile with Intel 2021 oneAPI compilers
315 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
316 440670 unhandled ppc64le-linux syscall: 252 statfs64 and 253 fstatfs64
317 440906 Fix impossible constraint issue in P10 testcase.
318 441512 Remove a unneeded / unnecessary prefix check.
319 441534 Update the expected output for test_isa_3_1_VRT.
320 442061 very slow execution under Fedora 34 (readdwarf3)
321 443031 Gcc -many change requires explicit .machine directives
322 443033 Add support for the ISA 3.0 mcrxrx instruction
323 443034 Sraw, srawi, srad, sradi, mfs
324 443178 Powerpc, test jm-mfspr expected output needs to be updated.
325 443179 Need new test for the lxvx and stxvx instructions on ISA 2.07 and
327 443180 The subnormal test and the ISA 3.0 test generate compiler warnings
328 443314 In the latest GIT version, Valgrind with "--trace-flags" crashes
330 443605 Don't call final_tidyup (__libc_freeres) on FatalSignal
332 To see details of a given bug, visit
333 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
334 where XXXXXX is the bug number as listed below.
336 (3.18.0.RC1: 12 Oct 2021)
337 (3.18.0: 15 Oct 2021)
341 Release 3.17.0 (19 Mar 2021)
342 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
344 3.17.0 fixes a number of bugs and adds some functional changes: support for
345 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
346 some new instructions for Arm64, S390 and POWER. There are also some tool
349 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
350 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
351 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
352 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
353 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
355 * ==================== CORE CHANGES ===================
357 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
360 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
361 debugging information. When a debuginfo file cannot be found locally,
362 Valgrind is able to query debuginfod servers for the file using its
363 build-id. See the user manual for more information about debuginfod support.
365 * ================== PLATFORM CHANGES =================
369 - Inaccuracies resulting from double-rounding in the simulation of
370 floating-point multiply-add/subtract instructions have been fixed. These
371 should now behave exactly as the hardware does.
373 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
374 ongoing. Support for the half-word variants of at least the following
375 instructions has been added:
377 FABS <Vd>.<T>, <Vn>.<T>
379 FNEG <Vd>.<T>, <Vn>.<T>
381 FSQRT <Vd>.<T>, <Vn>.<T>
386 - Implement the new instructions/features that were added to z/Architecture
387 with the vector-enhancements facility 1. Also cover the instructions from
388 the vector-packed-decimal facility that are defined outside the chapter
389 "Vector Decimal Instructions", but not the ones from that chapter itself.
391 For a detailed list of newly supported instructions see the updates to
392 `docs/internals/s390-opcodes.csv'.
394 Since the miscellaneous instruction extensions facility 2 was already
395 added in Valgrind 3.16.0, this completes the support necessary to run
396 general programs built with `--march=z14' under Valgrind. The
397 vector-packed-decimal facility is currently not exploited by the standard
398 toolchain and libraries.
402 - Various bug fixes. Fix for the sync field to limit setting just two of
403 the two bits in the L-field. Fix the write size for the stxsibx and
404 stxsihx instructions. Fix the modsw and modsd instructions.
406 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
407 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
408 conversion instructions are still missing.
410 * ==================== TOOL CHANGES ====================
412 * General tool changes
414 - All the tools and their vgpreload libraries are now installed under
415 libexec because they cannot be executed directly and should be run through
416 the valgrind executable. This should be an internal, not user visible,
417 change, but might impact valgrind packagers.
419 - The --track-fds option now respects -q, --quiet and won't output anything
420 if no file descriptors are leaked. It also won't report the standard stdin
421 (0), stdout (1) or stderr (2) descriptors as being leaked with
422 --trace-fds=yes anymore. To track whether the standard file descriptors
423 are still open at the end of the program run use --trace-fds=all.
427 - DHAT has been extended, with two new modes of operation. The new
428 --mode=copy flag triggers copy profiling, which records calls to memcpy,
429 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
430 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
431 the new dhat/dhat.h file. This is useful for learning more about hot code
432 paths. See the user manual for more information about the new modes.
434 - Because of these changes, DHAT's file format has changed. DHAT output
435 files produced with earlier versions of DHAT will not work with this
436 version of DHAT's viewer, and DHAT output files produced with this version
437 of DHAT will not work with earlier versions of DHAT's viewer.
439 * ==================== FIXED BUGS ====================
441 The following bugs have been fixed or resolved. Note that "n-i-bz"
442 stands for "not in bugzilla" -- that is, a bug that was reported to us
443 but never got a bugzilla entry. We encourage you to file bugs in
444 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
445 than mailing the developers (or mailing lists) directly -- bugs that
446 are not entered into bugzilla tend to get forgotten about or ignored.
448 To see details of a given bug, visit
449 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
450 where XXXXXX is the bug number as listed below.
452 140178 open("/proc/self/exe", ...); doesn't quite work
453 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
454 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
455 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
456 345077 linux syscall execveat support (linux 3.19)
457 361770 Missing F_ADD_SEALS
458 369029 handle linux syscalls sched_getattr and sched_setattr
459 384729 __libc_freeres inhibits cross-platform valgrind
460 388787 Support for C++17 new/delete
461 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
462 396656 Warnings while reading debug info
463 397605 ioctl FICLONE mishandled
464 401416 Compile failure with openmpi 4.0
465 408663 Suppression file for musl libc
466 404076 s390x: z14 vector instructions not implemented
467 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
468 (actually affected all x86 and nanomips regardless of host bitness)
469 413547 regression test does not check for Arm 64 features.
470 414268 Enable AArch64 feature detection and decoding for v8.x instructions
471 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
472 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
473 422261 platform selection fails for unqualified client name
474 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
475 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
476 423195 PPC ISA 3.1 support is missing, part 1
477 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
478 424012 crash with readv/writev having invalid but not NULL arg2 iovec
479 424298 amd64: Implement RDSEED
480 425232 PPC ISA 3.1 support is missing, part 2
481 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
482 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
483 426123 PPC ISA 3.1 support is missing, part 3
484 426144 Fix "condition variable has not been initialized" on Fedora 33.
485 427400 PPC ISA 3.1 support is missing, part 4
486 427401 PPC ISA 3.1 support is missing, part 5
487 427404 PPC ISA 3.1 support is missing, part 6
488 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
489 427787 Support new faccessat2 linux syscall (439)
490 427969 debuginfo section duplicates a section in the main ELF file
491 428035 drd: Unbreak the musl build
492 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
493 428716 cppcheck detects potential leak in VEX/useful/smchash.c
494 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
495 429352 PPC ISA 3.1 support is missing, part 7
496 429354 PPC ISA 3.1 support is missing, part 8
497 429692 unhandled ppc64le-linux syscall: 147 (getsid)
498 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
500 429952 Errors when building regtest with clang
501 430354 ppc stxsibx and stxsihx instructions write too much data
502 430429 valgrind.h doesn't compile on s390x with clang
503 430485 expr_is_guardable doesn't handle Iex_Qop
504 431556 Complete arm64 FADDP v8.2 instruction support
505 432102 Add support for DWARF5 as produced by GCC11
506 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
507 432381 drd: Process STACK_REGISTER client requests
508 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
509 432672 vg_regtest: test-specific environment variables not reset between tests
510 432809 VEX should support REX.W + POPF
511 432861 PPC modsw and modsd give incorrect results for 1 mod 12
512 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
513 432215 Add debuginfod functionality
514 433323 Use pkglibexecdir as vglibdir
515 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
516 433629 valgrind/README has type "abd" instead of "and"
517 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
518 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
519 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
520 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
521 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
523 (3.17.0.RC1: 13 Mar 2021)
524 (3.17.0.RC2: 17 Mar 2021)
525 (3.17.0: 19 Mar 2021)
529 Release 3.16.1 (22 June 2020)
530 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
532 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
533 fixes character encoding problems in the documentation HTML.
535 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
536 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
538 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
542 Release 3.16.0 (27 May 2020)
543 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
545 3.16.0 is a feature release with many improvements and the usual collection of
548 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
549 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
550 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
551 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
552 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
554 * ==================== CORE CHANGES ===================
556 * It is now possible to dynamically change the value of many command line
557 options while your program (or its children) are running under Valgrind.
559 To see the list of dynamically changeable options, run
560 "valgrind --help-dyn-options".
562 You can change the options from the shell by using vgdb to launch
563 the monitor command "v.clo <clo option>...".
564 The same monitor command can be used from a gdb connected
565 to the valgrind gdbserver.
566 Your program can also change the dynamically changeable options using
567 the client request VALGRIND_CLO_CHANGE(option).
569 * ================== PLATFORM CHANGES =================
571 * MIPS: preliminary support for nanoMIPS instruction set has been added.
573 * ==================== TOOL CHANGES ====================
577 - The implicit memcpy done by each call to realloc now counts towards the
578 read and write counts of resized heap blocks, making those counts higher
583 - cg_annotate's --auto and --show-percs options now default to 'yes', because
584 they are usually wanted.
588 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
589 because they are usually wanted.
591 - The command option --collect-systime has been enhanced to specify
592 the unit used to record the elapsed time spent during system calls.
593 The command option now accepts the values no|yes|msec|usec|nsec,
594 where yes is a synonym of msec. When giving the value nsec, the
595 system cpu time of system calls is also recorded.
599 - Several memcheck options are now dynamically changeable.
600 Use valgrind --help-dyn-options to list them.
602 - The release 3.15 introduced a backward incompatible change for
603 some suppression entries related to preadv and pwritev syscalls.
604 When reading a suppression entry using the unsupported 3.14 format,
605 valgrind will now produce a warning to say the suppression entry will not
606 work, and suggest the needed change.
608 - Significantly fewer false positive errors on optimised code generated by
609 Clang and GCC. In particular, Memcheck now deals better with the
610 situation where the compiler will transform C-level "A && B" into "B && A"
611 under certain circumstances (in which the transformation is valid).
612 Handling of integer equality/non-equality checks on partially defined
613 values is also improved on some architectures.
617 - The exprimental Stack and Global Array Checking tool has been removed.
618 It only ever worked on x86 and amd64, and even on those it had a
619 high false positive rate and was slow. An alternative for detecting
620 stack and global array overruns is using the AddressSanitizer (ASAN)
621 facility of the GCC and Clang compilers, which require you to rebuild
622 your code with -fsanitize=address.
624 * ==================== OTHER CHANGES ====================
626 * New and modified GDB server monitor features:
628 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
630 - The gdbserver monitor commands that require an address and an optional
631 length argument now accepts the alternate 'C like' syntax "address[length]".
632 For example, the memcheck command "monitor who_points_at 0x12345678 120"
633 can now also be given as "monitor who_points_at 0x12345678[120]".
635 * ==================== FIXED BUGS ====================
637 The following bugs have been fixed or resolved. Note that "n-i-bz"
638 stands for "not in bugzilla" -- that is, a bug that was reported to us
639 but never got a bugzilla entry. We encourage you to file bugs in
640 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
641 than mailing the developers (or mailing lists) directly -- bugs that
642 are not entered into bugzilla tend to get forgotten about or ignored.
644 To see details of a given bug, visit
645 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
646 where XXXXXX is the bug number as listed below.
648 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
649 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
650 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
651 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
652 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
653 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
654 400872 Add nanoMIPS support to Valgrind
655 403212 drd/tests/trylock hangs on FreeBSD
656 404406 s390x: z14 miscellaneous instructions not implemented
657 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
658 406561 mcinfcallWSRU gdbserver_test fails on ppc64
659 406824 Unsupported baseline
660 407218 Add support for the copy_file_range syscall
661 407307 Intercept stpcpy also in ld.so for arm64
662 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
664 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
665 408009 Expose rdrand and f16c even on avx if host cpu supports them
666 408091 Missing pkey syscalls
667 408414 Add support for missing for preadv2 and pwritev2 syscalls
668 409141 Valgrind hangs when SIGKILLed
669 409206 Support for Linux PPS and PTP ioctls
670 409367 exit_group() after signal to thread waiting in futex() causes hangs
671 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
672 409780 References to non-existent configure.in
673 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
674 410599 Non-deterministic behaviour of pth_self_kill_15_other test
675 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
676 411134 Allow the user to change a set of command line options during execution
677 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
678 412344 Problem setting mips flags with specific paths
679 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
680 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
681 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
682 413603 callgrind_annotate/cg_annotate truncate function names at '#'
683 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
684 415136 ARMv8.1 Compare-and-Swap instructions are not supported
685 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
686 416239 valgrind crashes when handling clock_adjtime
687 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
688 416286 DRD reports "conflicting load" error on std::mutex::lock()
689 416301 s390x: "compare and signal" not supported
690 416387 finit_module and bpf syscalls are unhandled on arm64
691 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
692 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
693 416753 new 32bit time syscalls for 2038+
694 417075 pwritev(vector[...]) suppression ignored
695 417075 is not fixed, but incompatible supp entries are detected
696 and a warning is produced for these.
697 417187 [MIPS] Conditional branch problem since 'grail' changes
698 417238 Test memcheck/tests/vbit-test fails on mips64 BE
699 417266 Make memcheck/tests/linux/sigqueue usable with musl
700 417281 s390x: /bin/true segfaults with "grail" enabled
701 417427 commit to fix vki_siginfo_t definition created numerous regression
703 417452 s390_insn_store_emit: dst->tag for HRcVec128
704 417578 Add suppressions for glibc DTV leaks
705 417906 clone with CLONE_VFORK and no CLONE_VM fails
706 418004 Grail code additions break ppc64.
707 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
708 418997 s390x: Support Iex_ITE for float and vector types
709 419503 s390x: Avoid modifying registers returned from isel functions
710 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
711 421570 std_mutex fails on Arm v8.1 h/w
712 434035 vgdb might crash if valgrind is killed
713 n-i-bz Fix minor one time leaks in dhat.
714 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
715 n-i-bz Add support for the Linux io_uring system calls
716 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
717 n-i-bz Fix non-glibc build of test suite with s390x_features
718 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
719 423195 PPC ISA 3.1 support is missing, part 1
721 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
722 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
723 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
727 Release 3.15.0 (12 April 2019)
728 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
730 3.15.0 is a feature release with many improvements and the usual collection of
733 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
734 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
735 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
736 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
737 support for X86/macOS 10.13 and AMD64/macOS 10.13.
739 * ==================== CORE CHANGES ===================
741 * The XTree Massif output format now makes use of the information obtained
742 when specifying --read-inline-info=yes.
744 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
746 * ==================== TOOL CHANGES ====================
750 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
751 it has been promoted from an experimental tool to a regular tool. Run it
752 with --tool=dhat instead of --tool=exp-dhat.
754 - DHAT now prints only minimal data when the program ends, instead writing
755 the bulk of the profiling data to a file. As a result, the --show-top-n
756 and --sort-by options have been removed.
758 - Profile results can be viewed with the new viewer, dh_view.html. When
759 a run ends, a short message is printed, explaining how to view the result.
761 - See the documentation for more details.
765 - cg_annotate has a new option, --show-percs, which prints percentages next
770 - callgrind_annotate has a new option, --show-percs, which prints percentages
771 next to all event counts.
773 - callgrind_annotate now inserts commas in call counts, and
774 sort the caller/callee lists in the call tree.
778 - The default value for --read-inline-info is now "yes" on
779 Linux/Android/Solaris. It is still "no" on other OS.
783 - The option --xtree-leak=yes (to output leak result in xtree format)
784 automatically activates the option --show-leak-kinds=all, as xtree
785 visualisation tools such as kcachegrind can in any case select what kind
786 of leak to visualise.
788 - There has been further work to avoid false positives. In particular,
789 integer equality on partially defined inputs (C == and !=) is now handled
792 * ==================== OTHER CHANGES ====================
794 * The new option --show-error-list=no|yes displays, at the end of the run, the
795 list of detected errors and the used suppressions. Prior to this change,
796 showing this information could only be done by specifying "-v -v", but that
797 also produced a lot of other possibly-non-useful messages. The option -s is
798 equivalent to --show-error-list=yes.
800 * ==================== FIXED BUGS ====================
802 The following bugs have been fixed or resolved. Note that "n-i-bz"
803 stands for "not in bugzilla" -- that is, a bug that was reported to us
804 but never got a bugzilla entry. We encourage you to file bugs in
805 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
806 than mailing the developers (or mailing lists) directly -- bugs that
807 are not entered into bugzilla tend to get forgotten about or ignored.
809 To see details of a given bug, visit
810 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
811 where XXXXXX is the bug number as listed below.
813 385411 s390x: z13 vector floating-point instructions not implemented
814 397187 z13 vector register support for vgdb gdbserver
815 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
816 398870 Please add support for instruction vcvtps2ph
817 399287 amd64 front end: Illegal Instruction vcmptrueps
818 399301 Use inlined frames in Massif XTree output.
819 399322 Improve callgrind_annotate output
820 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
821 400164 helgrind test encounters mips x-compiler warnings and assembler error
822 400490 s390x: VRs allocated as if separate from FPRs
823 400491 s390x: Operand of LOCH treated as unsigned integer
824 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
825 options, which specify a mips64 processor
826 401112 LLVM 5.0 generates comparison against partially initialized data
827 401277 More bugs in z13 support
828 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
829 401578 drd: crashes sometimes on fork()
830 401627 memcheck errors with glibc avx2 optimized wcsncmp
831 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
832 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
833 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
835 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
836 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
837 402123 invalid assembler opcodes for mips32r2
838 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
839 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
840 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
841 402351 mips64 libvexmultiarch_test fails on s390x
843 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
844 402480 Do not use %rsp in clobber list
845 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
846 402515 Implement new option --show-error-list=no|yes / -s
847 402519 POWER 3.0 addex instruction incorrectly implemented
848 402781 Redo the cache used to process indirect branch targets
849 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
850 403552 s390x: wrong facility bit checked for vector facility
851 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
852 404638 Add VG_(replaceIndexXA)
853 404843 s390x: backtrace sometimes ends prematurely
854 404888 autotools cleanup series
855 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
856 405182 Valgrind fails to build with Clang
857 405205 filter_libc: remove the line holding the futex syscall error entirely
858 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
859 the upper and lower 32-bits of the 64-bit result
860 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
861 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
862 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
864 405403 s390x disassembler cannot be used on x86
865 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
866 405458 MIPS mkFormVEC arguments swapped?
867 405716 drd: Fix an integer overflow in the stack margin calculation
868 405722 Support arm64 core dump
869 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
870 of the 64-bit destination field.
871 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
872 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
873 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
875 406256 PPC64, vector floating point instructions don't handle subnormal
876 according to VSCR[NJ] bit setting.
877 406352 cachegrind/callgrind fails ann tests because of missing a.c
878 406354 dhat is broken on x86 (32bit)
879 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
880 406357 gdbserver_tests fails because of gdb output change
881 406360 memcheck/tests/libstdc++.supp needs more supression variants
882 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
883 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
884 407340 PPC64, does not support the vlogefp, vexptefp instructions.
885 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
886 n-i-bz Fix callgrind_annotate non deterministic order for equal total
887 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
888 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
889 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
891 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
892 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
893 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
897 Release 3.14.0 (9 October 2018)
898 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
900 3.14.0 is a feature release with many improvements and the usual collection of
903 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
904 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
905 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
906 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
907 support for X86/macOS 10.13, AMD64/macOS 10.13.
909 * ==================== CORE CHANGES ===================
911 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
912 debug info for unloaded code. This allows saved stack traces (e.g. for
913 memory leaks) to include file/line info for code that has been dlclose'd (or
914 similar). See the user manual for more information and known limitations.
916 * Ability to specify suppressions based on source file name and line number.
918 * Majorly overhauled register allocator. No end-user changes, but the JIT
919 generates code a bit more quickly now.
921 * ================== PLATFORM CHANGES =================
923 * Preliminary support for macOS 10.13 has been added.
925 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
927 * mips: support for MIPS SIMD architecture (MSA) has been added.
929 * mips: support for MIPS N32 ABI has been added.
931 * s390: partial support for vector instructions (integer and string) has been
934 * ==================== TOOL CHANGES ====================
936 * Helgrind: Addition of a flag
937 --delta-stacktrace=no|yes [yes on linux amd64/x86]
938 which specifies how full history stack traces should be computed.
939 Setting this to =yes can speed up Helgrind by 25% when using
940 --history-level=full.
942 * Memcheck: reduced false positive rate for optimised code created by Clang 6
943 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
944 blocks more carefully to determine where it can avoid expensive definedness
945 checks without loss of precision. This is controlled by the flag
946 --expensive-definedness-checks=no|auto|yes [auto].
948 * ==================== OTHER CHANGES ====================
950 * Valgrind is now buildable with link-time optimisation (LTO). A new
951 configure option --enable-lto=yes allows building Valgrind with LTO. If the
952 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
953 Note that if you are doing Valgrind development, --enable-lto=yes massively
954 slows down the build process.
956 * ==================== FIXED BUGS ====================
958 The following bugs have been fixed or resolved. Note that "n-i-bz"
959 stands for "not in bugzilla" -- that is, a bug that was reported to us
960 but never got a bugzilla entry. We encourage you to file bugs in
961 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
962 than mailing the developers (or mailing lists) directly -- bugs that
963 are not entered into bugzilla tend to get forgotten about or ignored.
965 To see details of a given bug, visit
966 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
967 where XXXXXX is the bug number as listed below.
969 79362 Debug info is lost for .so files when they are dlclose'd
970 208052 strlcpy error when n = 0
971 255603 exp-sgcheck Assertion '!already_present' failed
972 338252 building valgrind with -flto (link time optimisation) fails
973 345763 MIPS N32 ABI support
974 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
975 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
976 372347 Replacement problem of the additional c++14/c++17 new/delete operators
977 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
978 376257 helgrind history full speed up using a cached stack
979 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
981 379748 Fix missing pselect syscall (OS X 10.11)
982 379754 Fix missing syscall ulock_wait (OS X 10.12)
983 380397 s390x: __GI_strcspn() replacemenet needed
984 381162 possible array overrun in VEX register allocator
985 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
986 381274 powerpc too chatty even with --sigill-diagnostics=no
987 381289 epoll_pwait can have a NULL sigmask
988 381553 VEX register allocator v3
989 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
990 381769 Use ucontext_t instead of struct ucontext
991 381805 arm32 needs ld.so index hardwire for new glibc security fixes
992 382256 gz compiler flag test doesn't work for gold
993 382407 vg_perf needs "--terse" command line option
994 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
995 382563 MIPS MSA ASE support
996 382998 xml-socket doesn't work
997 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
998 383723 Fix missing kevent_qos syscall (macOS 10.11)
999 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
1000 384096 Mention AddrCheck at Memcheck's command line option [..]
1001 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
1002 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
1003 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
1004 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
1005 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
1006 384337 performance improvements to VEX register allocator v2 and v3
1007 384526 reduce number of spill insns generated by VEX register allocator v3
1008 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
1009 384631 Sanitise client args as printed with -v
1010 384633 Add a simple progress-reporting facility
1011 384987 VEX regalloc: allocate caller-save registers for short lived vregs
1012 385055 PPC VEX temporary storage exhausted
1013 385182 PPC64 is missing support for the DSCR
1014 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
1015 385207 PPC64, generate_store_FPRF() generates too many Iops
1016 385208 PPC64, xxperm instruction exhausts temporary memory
1017 385210 PPC64, vpermr instruction could exhaust temporary memory
1018 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
1019 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
1020 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
1021 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
1022 385408 s390x: z13 vector "support" instructions not implemented
1023 385409 s390x: z13 vector integer instructions not implemented
1024 385410 s390x: z13 vector string instructions not implemented
1025 385412 s390x: new non-vector z13 instructions not implemented
1026 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
1027 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
1028 385939 Optionally exit on the first error
1029 386318 valgrind.org/info/tools.html is missing SGCheck
1030 386425 running valgrind + wine on armv7l gives illegal opcode
1031 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
1032 387410 MIPSr6 support
1033 387664 Memcheck: make expensive-definedness-checks be the default
1034 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
1035 387766 asm shifts cause false positive "Conditional jump or move depends
1036 on uninitialised value"
1037 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
1038 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
1039 388786 Support bpf syscall in amd64 Linux
1040 388862 Add replacements for wmemchr and wcsnlen on Linux
1041 389065 valgrind meets gcc flag -Wlogical-op
1042 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
1043 390471 suppression by specification of source-file line number
1044 390723 make xtree dump files world wide readable, similar to log files
1045 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
1046 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
1047 392118 unhandled amd64-linux syscall: 332 (statx)
1048 392449 callgrind not clearing the number of calls properly
1049 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
1050 lxssp, stxssp and stxvl instructions.
1051 393023 callgrind_control risks using the wrong vgdb
1052 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
1053 393099 posix_memalign() invalid write if alignment == 0
1054 393146 failing assert "is_DebugInfo_active(di)"
1055 395709 PPC64 is missing support for the xvnegsp instruction
1056 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
1058 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
1059 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
1060 396839 s390x: Trap instructions not implemented
1061 396887 arch_prctl should return EINVAL on unknown option
1062 == 397286 crash before launching binary (Unsupported arch_prctl option)
1063 == 397393 valgrind: the 'impossible' happened: (Archlinux)
1064 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
1065 396906 compile tests failure on mips32-linux: broken inline asm in tests on
1067 397012 glibc ld.so uses arch_prctl on i386
1068 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
1069 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
1070 397424 glibc 2.27 and gdb_server tests
1071 398028 Assertion `cfsi_fits` failing in simple C program
1072 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
1074 n-i-bz Fix missing workq_ops operations (macOS)
1075 n-i-bz fix bug in strspn replacement
1076 n-i-bz Add support for the Linux BLKFLSBUF ioctl
1077 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
1078 n-i-bz Fix possible stack trashing by semctl syscall wrapping
1079 n-i-bz Add support for the Linux membarrier() system call
1080 n-i-bz x86 front end: recognise and handle UD2 correctly
1081 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
1082 correctly aligned before entering the handler.
1084 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
1085 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
1086 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
1090 Release 3.13.0 (15 June 2017)
1091 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1093 3.13.0 is a feature release with many improvements and the usual collection of
1096 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1097 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1098 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1099 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
1101 * ==================== CORE CHANGES ===================
1103 * The translation cache size has been increased to keep up with the demands of
1104 large applications. The maximum number of sectors has increased from 24 to
1105 48. The default number of sectors has increased from 16 to 32 on all
1106 targets except Android, where the increase is from 6 to 12.
1108 * The amount of memory that Valgrind can use has been increased from 64GB to
1109 128GB. In particular this means your application can allocate up to about
1110 60GB when running on Memcheck.
1112 * Valgrind's default load address has been changed from 0x3800'0000 to
1113 0x5800'0000, so as to make it possible to load larger executables. This
1114 should make it possible to load executables of size at least 1200MB.
1116 * A massive spaceleak caused by reading compressed debuginfo files has been
1117 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
1120 * The C++ demangler has been updated.
1122 * Support for demangling Rust symbols has been added.
1124 * A new representation of stack traces, the "XTree", has been added. An XTree
1125 is a tree of stacktraces with data associated with the stacktraces. This is
1126 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
1127 consumption of your program. Reporting is controlled by the new options
1128 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
1130 A report can also be produced on demand using the gdbserver monitor command
1131 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
1132 format' and 'massif format. The existing visualisers for these formats (e.g.
1133 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
1134 analyse these reports.
1136 Memcheck can also produce XTree leak reports using the Callgrind file
1137 format. For more details, see the user manual.
1139 * ================== PLATFORM CHANGES =================
1141 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
1143 * amd64: fixes for JIT failure problems on long AVX2 code blocks
1145 * amd64 and x86: support for CET prefixes has been added
1147 * arm32: a few missing ARMv8 instructions have been implemented
1149 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
1150 Store-Conditional instructions has been added. This is to deal with
1151 processor implementations that implement the LL/SC specifications strictly
1152 and as a result cause Valgrind to hang in certain situations. The
1153 alternative implementation is automatically enabled at startup, as required.
1154 You can use the option --sim-hints=fallback-llsc to force-enable it if you
1157 * Support for OSX 10.12 has been improved.
1159 * On Linux, clone handling has been improved to honour CLONE_VFORK that
1160 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
1161 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
1162 CLONE_VM exact semantics will (still) not work.
1164 * The TileGX/Linux port has been removed because it appears to be both unused
1167 * ==================== TOOL CHANGES ====================
1171 - Memcheck should give fewer false positives when running optimised
1172 Clang/LLVM generated code.
1174 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1176 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
1177 to produce the end of execution leak report in a xtree callgrind format
1180 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
1181 the leak report in an xtree file.
1185 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1187 - For some workloads (typically, for big applications), Massif memory
1188 consumption and CPU consumption has decreased significantly.
1192 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1194 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
1195 for Ada gnat compiled applications.
1197 * ==================== OTHER CHANGES ====================
1199 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
1200 append the inner guest stacktrace to the inner host stacktrace. This helps
1201 to investigate the errors reported by the outer, when they are caused by the
1202 inner guest program (such as an inner regtest). See README_DEVELOPERS for
1205 * To allow fast detection of callgrind files by desktop environments and file
1206 managers, the format was extended to have an optional first line that
1207 uniquely identifies the format ("# callgrind format"). Callgrind creates
1208 this line now, as does the new xtree functionality.
1210 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
1211 have a new %n format letter that is replaced by a sequence number.
1213 * "--version -v" now shows the SVN revision numbers from which Valgrind was
1216 * ==================== FIXED BUGS ====================
1218 The following bugs have been fixed or resolved. Note that "n-i-bz"
1219 stands for "not in bugzilla" -- that is, a bug that was reported to us
1220 but never got a bugzilla entry. We encourage you to file bugs in
1221 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1222 than mailing the developers (or mailing lists) directly -- bugs that
1223 are not entered into bugzilla tend to get forgotten about or ignored.
1225 To see details of a given bug, visit
1226 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1227 where XXXXXX is the bug number as listed below.
1229 162848 --log-file output isn't split when a program forks
1230 340777 Illegal instruction on mips (ar71xx)
1231 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
1232 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
1233 to a different stack.
1234 344139 x86 stack-seg overrides, needed by the Wine people
1235 344524 store conditional of guest applications always fail - observed on
1237 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
1238 352395 Please provide SVN revision info in --version -v
1239 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
1240 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
1241 INVALID_POSIX_THREADID' failed
1242 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
1243 barrier implementation
1244 358697 valgrind.h: Some code remains even when defining NVALGRIND
1245 359202 Add musl libc configure/compile
1246 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
1247 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
1248 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
1249 362223 assertion failed when .valgrindrc is a directory instead of a file
1250 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
1251 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
1252 368507 can't malloc chunks larger than about 34GB
1253 368529 Android arm target link error, missing atexit and pthread_atfork
1254 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
1255 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
1256 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
1257 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
1258 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
1259 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
1260 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
1261 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
1262 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
1263 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
1264 370028 Reduce the number of compiler warnings on MIPS platforms
1265 370635 arm64 missing syscall getcpu
1266 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
1267 371227 Clean AArch64 syscall table
1268 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
1269 371471 Valgrind complains about non legit memory leaks on placement new (C++)
1270 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
1271 371503 disInstr(arm64): unhandled instruction 0xF89F0000
1272 371869 support '%' in symbol Z-encoding
1273 371916 execution tree xtree concept
1274 372120 c++ demangler demangles symbols which are not c++
1275 372185 Support of valgrind on ARMv8 with 32 bit executable
1276 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1277 372195 Power PC, xxsel instruction is not always recognized.
1278 372504 Hanging on exit_group
1279 372600 process loops forever when fatal signals are arriving quickly
1280 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1281 373046 Stacks registered by core are never deregistered
1282 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1283 373086 Implement additional Xen hypercalls
1284 373192 Calling posix_spawn in glibc 2.24 completely broken
1285 373488 Support for fanotify API on ARM64 architecture
1286 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1287 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1288 373938 const IRExpr arguments for matchIRExpr()
1289 374719 some spelling fixes
1290 374963 increase valgrind's load address to prevent mmap failure
1291 375514 valgrind_get_tls_addr() does not work in case of static TLS
1292 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1293 for ML_(find_rx_mapping)()
1294 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1295 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1296 == 377159 "vex: the `impossible' happened" still present
1297 == 375150 Assertion 'tres.status == VexTransOK' failed
1298 == 378068 valgrind crashes on AVX2 function in FFmpeg
1299 376142 Segfaults on MIPS Cavium Octeon boards
1300 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1301 376455 Solaris: unhandled syscall lgrpsys(180)
1302 376518 Solaris: unhandled fast trap getlgrp(6)
1303 376611 ppc64 and arm64 don't know about prlimit64 syscall
1304 376729 PPC64, remove R2 from the clobber list
1306 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1307 to be wrongly marked as addressable
1308 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1309 PIE enabled by default
1310 377376 memcheck/tests/linux/getregset fails with glibc2.24
1311 377427 PPC64, lxv instruction failing on odd destination register
1312 377478 PPC64: ISA 3.0 setup fixes
1313 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1314 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1315 and 2 args for FUTEX_TRYLOCK_PI
1316 377717 Fix massive space leak when reading compressed debuginfo sections
1317 377891 Update Xen 4.6 domctl wrappers
1318 377930 fcntl syscall wrapper is missing flock structure check
1319 378524 libvexmultiarch_test regression on s390x and ppc64
1320 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1321 378673 Update libiberty demangler
1322 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1323 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1324 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1325 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1326 (task_register_dyld_image_infos)
1327 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1328 (task_register_dyld_shared_cache_image_info)
1329 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1330 379473 MIPS: add support for rdhwr cycle counter register
1331 379504 remove TileGX/Linux port
1332 379525 Support more x86 nop opcodes
1333 379838 disAMode(x86): not an addr!
1334 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1335 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1336 379895 clock_gettime does not execute POST syscall wrapper
1337 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1338 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1339 380200 xtree generated callgrind files refer to files without directory name
1340 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1341 380397 s390x: __GI_strcspn() replacement needed
1342 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1344 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1345 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1346 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1350 Release 3.12.0 (20 October 2016)
1351 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1353 3.12.0 is a feature release with many improvements and the usual
1354 collection of bug fixes.
1356 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1357 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1358 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1359 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1360 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1361 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1363 * ================== PLATFORM CHANGES =================
1365 * POWER: Support for ISA 3.0 has been added
1367 * mips: support for O32 FPXX ABI has been added.
1368 * mips: improved recognition of different processors
1369 * mips: determination of page size now done at run time
1371 * amd64: Partial support for AMD FMA4 instructions.
1373 * arm, arm64: Support for v8 crypto and CRC instructions.
1375 * Improvements and robustification of the Solaris port.
1377 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1379 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1380 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1381 where possible. Valgrind's support for 32-bit x86 has stagnated in
1382 recent years and has fallen far behind that for 64-bit x86
1383 instructions. By contrast 64-bit x86 is well supported, up to and
1386 * ==================== TOOL CHANGES ====================
1390 - Added meta mempool support for describing a custom allocator which:
1391 - Auto-frees all chunks assuming that destroying a pool destroys all
1393 - Uses itself to allocate other memory blocks
1395 - New flag --ignore-range-below-sp to ignore memory accesses below
1396 the stack pointer, if you really have to. The related flag
1397 --workaround-gcc296-bugs=yes is now deprecated. Use
1398 --ignore-range-below-sp=1024-1 as a replacement.
1402 - Improved thread startup time significantly on non-Linux platforms.
1406 - Added collection of the metric "tot-blocks-allocd"
1408 * ==================== OTHER CHANGES ====================
1410 * Replacement/wrapping of malloc/new related functions is now done not just
1411 for system libraries by default, but for any globally defined malloc/new
1412 related function (both in shared libraries and statically linked alternative
1413 malloc implementations). The dynamic (runtime) linker is excluded, though.
1414 To only intercept malloc/new related functions in
1415 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1416 "nouserintercepts" can be any non-existing library name).
1417 This new functionality is not implemented for MacOS X.
1419 * The maximum number of callers in a suppression entry is now equal to
1420 the maximum size for --num-callers (500).
1421 Note that --gen-suppressions=yes|all similarly generates suppressions
1422 containing up to --num-callers frames.
1424 * New and modified GDB server monitor features:
1426 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1427 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1430 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1431 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1434 * Valgrind is able to read compressed debuginfo sections in two formats:
1435 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1436 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1438 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1439 for the most common use case (x86_64-linux, Memcheck) has been
1442 * Improved performance for programs that do a lot of discarding of
1443 instruction address ranges of 8KB or less.
1445 * The C++ symbol demangler has been updated.
1447 * More robustness against invalid syscall parameters on Linux.
1449 * ==================== FIXED BUGS ====================
1451 The following bugs have been fixed or resolved. Note that "n-i-bz"
1452 stands for "not in bugzilla" -- that is, a bug that was reported to us
1453 but never got a bugzilla entry. We encourage you to file bugs in
1454 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1455 than mailing the developers (or mailing lists) directly -- bugs that
1456 are not entered into bugzilla tend to get forgotten about or ignored.
1458 To see details of a given bug, visit
1459 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1460 where XXXXXX is the bug number as listed below.
1462 191069 Exiting due to signal not reported in XML output
1463 199468 Suppressions: stack size limited to 25
1464 while --num-callers allows more frames
1465 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1466 278744 cvtps2pd with redundant RexW
1467 303877 valgrind doesn't support compressed debuginfo sections.
1468 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1469 348345 Assertion fails for negative lineno
1470 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1471 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1472 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1473 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1474 352197 mips: mmap2() not wrapped correctly for page size > 4096
1475 353083 arm64 doesn't implement various xattr system calls
1476 353084 arm64 doesn't support sigpending system call
1477 353137 www: update info for Supported Platforms
1478 353138 www: update "The Valgrind Developers" page
1479 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1482 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1483 353398 WARNING: unhandled amd64-solaris syscall: 207
1484 353660 XML in auxwhat tag not escaping reserved symbols properly
1485 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1486 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1487 353802 ELF debug info reader confused with multiple .rodata sections
1488 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1489 353917 unhandled amd64-solaris syscall fchdir(120)
1490 353920 unhandled amd64-solaris syscall: 170
1491 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1492 354392 unhandled amd64-solaris syscall: 171
1493 354797 Vbit test does not include Iops for Power 8 instruction support
1494 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1498 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1499 355188 valgrind should intercept all malloc related global functions
1500 355454 do not intercept malloc related symbols from the runtime linker
1501 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1502 356044 Dwarf line info reader misinterprets is_stmt register
1503 356112 mips: replace addi with addiu
1504 356393 valgrind (vex) crashes because isZeroU happened
1507 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1508 356678 arm64-linux: unhandled syscall 232 (mincore)
1509 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1510 356823 Unsupported ARM instruction: stlex
1511 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1512 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1513 357673 crash if I try to run valgrind with a binary link with libcurl
1514 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1515 357871 pthread_spin_destroy not properly wrapped
1516 357887 Calls to VG_(fclose) do not close the file descriptor
1517 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1518 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1519 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1520 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1521 359181 Buffer Overflow during Demangling
1522 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1523 359289 s390x: popcnt (B9E1) not implemented
1524 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1525 359503 Add missing syscalls for aarch64 (arm64)
1526 359645 "You need libc6-dbg" help message could be more helpful
1527 359703 s390: wire up separate socketcalls system calls
1528 359724 getsockname might crash - deref_UInt should call safe_to_deref
1529 359733 amd64 implement ld.so strchr/index override like x86
1530 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1531 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1533 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1534 359871 Incorrect mask handling in ppoll
1535 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1536 360008 Contents of Power vr registers contents is not printed correctly when
1537 the --vgdb-shadow-registers=yes option is used
1538 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1539 non-zero shadow bits
1540 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1541 360425 arm64 unsupported instruction ldpsw
1543 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1544 360571 Error about the Android Runtime reading below the stack pointer on ARM
1545 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1546 360749 kludge for multiple .rodata sections on Solaris no longer needed
1547 360752 raise the number of reserved fds in m_main.c from 10 to 12
1548 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1549 361226 s390x: risbgn (EC59) not implemented
1550 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1551 361354 ppc64[le]: wire up separate socketcalls system calls
1552 361615 Inconsistent termination for multithreaded process terminated by signal
1553 361926 Unhandled Solaris syscall: sysfs(84)
1554 362009 V dumps core on unimplemented functionality before threads are created
1555 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1556 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1557 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1558 362953 Request for an update to the Valgrind Developers page
1559 363680 add renameat2() support
1560 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1561 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1562 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1563 364058 clarify in manual limitations of array overruns detections
1564 364413 pselect sycallwrapper mishandles NULL sigmask
1565 364728 Power PC, missing support for several HW registers in
1566 get_otrack_shadow_offset_wrk()
1567 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1568 365273 Invalid write to stack location reported after signal handler runs
1569 365912 ppc64BE segfault during jm-insns test (RELRO)
1570 366079 FPXX Support for MIPS32 Valgrind
1571 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1572 366344 Multiple unhandled instruction for Aarch64
1573 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1574 367995 Integration of memcheck with custom memory allocator
1575 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1576 368412 False positive result for altivec capability check
1577 368416 Add tc06_two_races_xml.exp output for ppc64
1578 368419 Perf Events ioctls not implemented
1579 368461 mmapunmap test fails on ppc64
1580 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1581 369000 AMD64 fma4 instructions unsupported.
1582 369169 ppc64 fails jm_int_isa_2_07 test
1583 369175 jm_vec_isa_2_07 test crashes on ppc64
1584 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1585 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1586 369359 msghdr_foreachfield can crash when handling bad iovec
1587 369360 Bad sigprocmask old or new sets can crash valgrind
1588 369361 vmsplice syscall wrapper crashes on bad iovec
1589 369362 Bad sigaction arguments crash valgrind
1590 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1591 369402 Bad set/get_thread_area pointer crashes valgrind
1592 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1593 369446 valgrind crashes on unknown fcntl command
1594 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1595 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1596 370265 ISA 3.0 HW cap stuff needs updating
1597 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1598 372195 Power PC, xxsel instruction is not always recognized
1600 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1601 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1602 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1603 n-i-bz Fix ppoll_alarm exclusion on OS X
1604 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1605 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1606 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1607 n-i-bz amd64: memcheck false positive with shr %edx
1608 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1609 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1610 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1611 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1613 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1614 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1615 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1619 Release 3.11.0 (22 September 2015)
1620 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1622 3.11.0 is a feature release with many improvements and the usual
1623 collection of bug fixes.
1625 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1626 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1627 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1628 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1629 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1630 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1632 * ================== PLATFORM CHANGES =================
1634 * Support for Solaris/x86 and Solaris/amd64 has been added.
1636 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1638 * Preliminary support for the Tilera TileGX architecture has been added.
1640 * s390x: It is now required for the host to have the "long displacement"
1641 facility. The oldest supported machine model is z990.
1643 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1644 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1646 * The JIT's register allocator is significantly faster, making the JIT
1647 as a whole somewhat faster, so JIT-intensive activities, for example
1648 program startup, are modestly faster, around 5%.
1650 * There have been changes to the default settings of several command
1651 line flags, as detailed below.
1653 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1654 capable hosts, the simulated CPUID will now indicate AVX2 support.
1656 * ==================== TOOL CHANGES ====================
1660 - The default value for --leak-check-heuristics has been changed from
1661 "none" to "all". This helps to reduce the number of possibly
1662 lost blocks, in particular for C++ applications.
1664 - The default value for --keep-stacktraces has been changed from
1665 "malloc-then-free" to "malloc-and-free". This has a small cost in
1666 memory (one word per malloc-ed block) but allows Memcheck to show the
1667 3 stacktraces of a dangling reference: where the block was allocated,
1668 where it was freed, and where it is acccessed after being freed.
1670 - The default value for --partial-loads-ok has been changed from "no" to
1671 "yes", so as to avoid false positive errors resulting from some kinds
1672 of vectorised loops.
1674 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1675 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1676 than get_vbits when you need to associate byte data value with
1677 their corresponding validity bits.
1679 - The 'block_list' monitor command has been enhanced:
1680 o it can print a range of loss records
1681 o it now accepts an optional argument 'limited <max_blocks>'
1682 to control the number of blocks printed.
1683 o if a block has been found using a heuristic, then
1684 'block_list' now shows the heuristic after the block size.
1685 o the loss records/blocks to print can be limited to the blocks
1686 found via specified heuristics.
1688 - The C helper functions used to instrument loads on
1689 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1690 replaced by handwritten assembly sequences. This gives speedups
1691 in the region of 0% to 7% for those targets only.
1693 - A new command line option, --expensive-definedness-checks=yes|no,
1694 has been added. This is useful for avoiding occasional invalid
1695 uninitialised-value errors in optimised code. Watch out for
1696 runtime degradation, as this can be up to 25%. As always, though,
1697 the slowdown is highly application specific. The default setting
1702 - A new monitor command 'all_snapshots <filename>' dumps all
1703 snapshots taken so far.
1707 - Significant memory reduction and moderate speedups for
1708 --history-level=full for applications accessing a lot of memory
1709 with many different stacktraces.
1711 - The default value for --conflict-cache-size=N has been doubled to
1712 2000000. Users that were not using the default value should
1713 preferably also double the value they give.
1715 The default was changed due to the changes in the "full history"
1716 implementation. Doubling the value gives on average a slightly more
1717 complete history and uses similar memory (or significantly less memory
1718 in the worst case) than the previous implementation.
1720 - The Helgrind monitor command 'info locks' now accepts an optional
1721 argument 'lock_addr', which shows information about the lock at the
1724 - When using --history-level=full, the new Helgrind monitor command
1725 'accesshistory <addr> [<len>]' will show the recorded accesses for
1726 <len> (or 1) bytes at <addr>.
1728 * ==================== OTHER CHANGES ====================
1730 * The default value for the --smc-check option has been changed from
1731 "stack" to "all-non-file" on targets that provide automatic D-I
1732 cache coherence (x86, amd64 and s390x). The result is to provide,
1733 by default, transparent support for JIT generated and self-modifying
1734 code on all targets.
1736 * Mac OS X only: the default value for the --dsymutil option has been
1737 changed from "no" to "yes", since any serious usage on Mac OS X
1738 always required it to be "yes".
1740 * The command line options --db-attach and --db-command have been removed.
1741 They were deprecated in 3.10.0.
1743 * When a process dies due to a signal, Valgrind now shows the signal
1744 and the stacktrace at default verbosity (i.e. verbosity 1).
1746 * The address description logic used by Memcheck and Helgrind now
1747 describes addresses in anonymous segments, file mmap-ed segments,
1748 shared memory segments and the brk data segment.
1750 * The new option --error-markers=<begin>,<end> can be used to mark the
1751 begin/end of errors in textual output mode, to facilitate
1752 searching/extracting errors in output files that mix valgrind errors
1753 with program output.
1755 * The new option --max-threads=<number> can be used to change the number
1756 of threads valgrind can handle. The default is 500 threads which
1757 should be more than enough for most applications.
1759 * The new option --valgrind-stacksize=<number> can be used to change the
1760 size of the private thread stacks used by Valgrind. This is useful
1761 for reducing memory use or increasing the stack size if Valgrind
1762 segfaults due to stack overflow.
1764 * The new option --avg-transtab-entry-size=<number> can be used to specify
1765 the expected instrumented block size, either to reduce memory use or
1766 to avoid excessive retranslation.
1768 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1770 * New and modified GDB server monitor features:
1772 - When a signal is reported in GDB, you can now use the GDB convenience
1773 variable $_siginfo to examine detailed signal information.
1775 - Valgrind's gdbserver now allows the user to change the signal
1776 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1777 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1778 continue without passing the signal to the process.
1780 - With GDB >= 7.10, the command 'target remote'
1781 will automatically load the executable file of the process running
1782 under Valgrind. This means you do not need to specify the executable
1783 file yourself, GDB will discover it itself. See GDB documentation about
1784 'qXfer:exec-file:read' packet for more info.
1786 * ==================== FIXED BUGS ====================
1788 The following bugs have been fixed or resolved. Note that "n-i-bz"
1789 stands for "not in bugzilla" -- that is, a bug that was reported to us
1790 but never got a bugzilla entry. We encourage you to file bugs in
1791 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1792 than mailing the developers (or mailing lists) directly -- bugs that
1793 are not entered into bugzilla tend to get forgotten about or ignored.
1795 To see details of a given bug, visit
1796 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1797 where XXXXXX is the bug number as listed below.
1799 116002 VG_(printf): Problems with justification of strings and integers
1800 155125 avoid cutting away file:lineno after long function name
1801 197259 Unsupported arch_prtctl PR_SET_GS option
1802 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
1803 201216 Fix Valgrind does not support pthread_sigmask() on OS X
1804 201435 Fix Darwin: -v does not show kernel version
1805 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
1806 211256 Fixed an outdated comment regarding the default platform.
1807 211529 Incomplete call stacks for code compiled by newer versions of MSVC
1808 211926 Avoid compilation warnings in valgrind.h with -pedantic
1809 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
1811 226609 Crediting upstream authors in man page
1812 231257 Valgrind omits path when executing script from shebang line
1813 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
1814 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
1815 269360 s390x: Fix addressing mode selection for compare-and-swap
1816 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
1818 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
1819 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
1820 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
1821 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
1822 330147 libmpiwrap PMPI_Get_count returns undefined value
1823 333051 mmap of huge pages fails due to incorrect alignment
1825 334802 valgrind does not always explain why a given option is bad
1826 335618 mov.w rN, pc/sp (ARM32)
1827 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
1832 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
1833 338602 AVX2 bit in CPUID missing
1834 338606 Strange message for scripts with invalid interpreter
1835 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
1836 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
1837 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
1839 339156 gdbsrv not called for fatal signal
1840 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
1841 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
1842 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
1843 339442 Fix testsuite build failure on OS X 10.9
1844 339542 Enable compilation with Intel's ICC compiler
1845 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
1846 339688 Mac-specific ASM does not support .version directive (cpuid,
1847 tronical and pushfpopf tests)
1848 339745 Valgrind crash when check Marmalade app (partial fix)
1849 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
1850 339778 Linux/TileGx platform support for Valgrind
1851 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
1852 339789 Fix none/tests/execve test on Mac OS X 10.9
1853 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
1854 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
1855 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
1856 340392 Allow user to select more accurate definedness checking in memcheck
1857 to avoid invalid complaints on optimised code
1858 340430 Fix some grammatical weirdness in the manual.
1859 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
1860 341419 Signal handler ucontext_t not filled out correctly on OS X
1861 341539 VG_(describe_addr) should not describe address as belonging to client
1862 segment if it is past the heap end
1863 341613 Enable building of manythreads and thread-exits tests on Mac OS X
1864 341615 Fix none/tests/darwin/access_extended test on Mac OS X
1865 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
1866 341789 aarch64: shmat fails with valgrind on ARMv8
1867 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
1868 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
1869 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
1870 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
1871 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
1872 342221 socket connect false positive uninit memory for unknown af family
1873 342353 Allow dumping full massif output while valgrind is still running
1874 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
1878 342603 Add I2C_SMBUS ioctl support
1879 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
1880 342683 Mark memory past the initial brk limit as unaddressable
1881 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
1882 342795 Internal glibc __GI_mempcpy call should be intercepted
1883 342841 s390x: Support instructions fiebr(a) and fidbr(a)
1884 343012 Unhandled syscall 319 (memfd_create)
1885 343069 Patch updating v4l2 API support
1886 343173 helgrind crash during stack unwind
1887 343219 fix GET_STARTREGS for arm
1888 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
1889 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
1890 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
1891 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
1892 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
1893 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
1894 343597 ppc64le: incorrect use of offseof macro
1895 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
1896 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
1897 343732 Unhandled syscall 144 (setgid) on aarch64
1898 343733 Unhandled syscall 187 (msgctl and related) on aarch64
1899 343802 s390x: False positive "conditional jump or move depends on [..]
1900 343902 --vgdb=yes doesn't break when --xml=yes is used
1901 343967 Don't warn about setuid/setgid/setcap executable for directories
1902 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
1903 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
1904 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
1905 344054 www - update info for Solaris/illumos
1906 344416 'make regtest' does not work cleanly on OS X
1907 344235 Remove duplicate include of pub_core_aspacemgr.h
1908 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
1909 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
1910 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
1911 344314 callgrind_annotate ... warnings about commands containing newlines
1912 344318 socketcall should wrap recvmmsg and sendmmsg
1913 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
1914 344416 Fix 'make regtest' does not work cleanly on OS X
1915 344499 Fix compilation for Linux kernel >= 4.0.0
1916 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
1917 unix:349 (__pthread_fchdir)
1918 344559 Garbage collection of unused segment names in address space manager
1919 344560 Fix stack traces missing penultimate frame on OS X
1920 344621 Fix memcheck/tests/err_disable4 test on OS X
1921 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
1922 344702 Fix missing libobjc suppressions on OS X 10.10
1924 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
1925 344939 Fix memcheck/tests/xml1 on OS X 10.10
1926 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
1927 345079 Fix build problems in VEX/useful/test_main.c
1928 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
1929 345177 arm64: prfm (reg) not implemented
1930 345215 Performance improvements for the register allocator
1931 345248 add support for Solaris OS in valgrind
1932 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
1933 345394 Fix memcheck/tests/strchr on OS X
1934 345637 Fix memcheck/tests/sendmsg on OS X
1935 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
1936 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
1937 345887 Fix an assertion in the address space manager
1938 345928 amd64: callstack only contains current function for small stacks
1939 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1940 345987 MIPS64: Implement cavium LHX instruction
1941 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1942 346185 Fix typo saving altivec register v24
1943 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1944 and LibVEX_GuestPPC64_get_CR()
1945 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1946 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1947 346307 fuse filesystem syscall deadlocks
1948 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1949 346411 MIPS: SysRes::_valEx handling is incorrect
1950 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1951 346474 PPC64 Power 8, spr TEXASRU register not supported
1952 346487 Compiler generates "note" about a future ABI change for PPC64
1953 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1954 and causing spurious "invalid read of size 8" warnings
1955 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1956 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1957 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1958 347322 Power PC regression test cleanup
1959 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1961 347389 unhandled syscall: 373 (Linux ARM syncfs)
1962 347686 Patch set to cleanup PPC64 regtests
1963 347978 Remove bash dependencies where not needed
1964 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1965 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1967 348102 Patch updating v4l2 API support
1968 348247 amd64 front end: jno jumps wrongly when overflow is not set
1969 348269 Improve mmap MAP_HUGETLB support.
1970 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1971 348345 Assertion fails for negative lineno
1972 348377 Unsupported ARM instruction: yield
1973 348565 Fix detection of command line option availability for clang
1974 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1975 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1976 348748 Fix redundant condition
1977 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1978 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1979 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1980 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1981 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1982 349626 Implemented additional Xen hypercalls
1983 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1984 349790 Clean up of the hardware capability checking utilities.
1985 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1986 349874 Fix typos in source code
1987 349879 memcheck: add handwritten assembly for helperc_LOADV*
1988 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1989 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1990 350202 Add limited param to 'monitor block_list'
1991 350290 s390x: Support instructions fixbr(a)
1992 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1993 350809 Fix none/tests/async-sigs for Solaris
1994 350811 Remove reference to --db-attach which has been removed.
1995 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1996 350854 hard-to-understand code in VG_(load_ELF)()
1997 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1998 351386 Solaris: Cannot run ld.so.1 under Valgrind
1999 351474 Fix VG_(iseqsigset) as obvious
2000 351531 Typo in /include/vki/vki-xen-physdev.h header guard
2001 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
2002 351858 ldsoexec support on Solaris
2003 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
2004 352130 helgrind reports false races for printfs using mempcpy on FILE* state
2005 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
2006 352320 arm64 crash on none/tests/nestedfs
2007 352765 Vbit test fails on Power 6
2008 352768 The mbar instruction is missing from the Power PC support
2009 352769 Power PC program priority register (PPR) is not supported
2010 n-i-bz Provide implementations of certain compiler builtins to support
2011 compilers that may not provide those
2012 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
2013 n-i-bz Fix compilation on distros with glibc < 2.5
2014 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
2015 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
2016 n-i-bz mremap did not work properly on shared memory
2017 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
2018 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
2020 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
2021 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
2022 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
2026 Release 3.10.1 (25 November 2014)
2027 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2028 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
2029 and backports fixes for all reported missing AArch64 ARMv8 instructions
2030 and syscalls from the trunk. If you package or deliver 3.10.0 for others
2031 to use, you might want to consider upgrading to 3.10.1 instead.
2033 The following bugs have been fixed or resolved. Note that "n-i-bz"
2034 stands for "not in bugzilla" -- that is, a bug that was reported to us
2035 but never got a bugzilla entry. We encourage you to file bugs in
2036 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2037 than mailing the developers (or mailing lists) directly -- bugs that
2038 are not entered into bugzilla tend to get forgotten about or ignored.
2040 To see details of a given bug, visit
2041 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2042 where XXXXXX is the bug number as listed below.
2044 335440 arm64: ld1 (single structure) is not implemented
2045 335713 arm64: unhanded instruction: prfm (immediate)
2046 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
2047 339182 ppc64: AvSplat ought to load destination vector register with [..]
2048 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
2049 339433 ppc64 lxvw4x instruction uses four 32-byte loads
2050 339645 Use correct tag names in sys_getdents/64 wrappers
2051 339706 Fix false positive for ioctl(TIOCSIG) on linux
2052 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
2053 339853 arm64 times syscall unknown
2054 339855 arm64 unhandled getsid/setsid syscalls
2055 339858 arm64 dmb sy not implemented
2056 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
2057 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
2058 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
2060 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
2061 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
2062 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
2063 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
2064 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
2065 340509 arm64: unhandled instruction fcvtas
2066 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
2067 340632 arm64: unhandled instruction fcvtas
2068 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
2069 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
2070 340788 warning: unhandled syscall: 318 (getrandom)
2071 340807 disInstr(arm): unhandled instruction: 0xEE989B20
2072 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
2073 340922 arm64: unhandled getgroups/setgroups syscalls
2074 350251 Fix typo in VEX utility program (test_main.c).
2075 350407 arm64: unhandled instruction ucvtf (vector, integer)
2076 350809 none/tests/async-sigs breaks when run under cron on Solaris
2077 350811 update README.solaris after r15445
2078 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
2079 350854 strange code in VG_(load_ELF)()
2080 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2081 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
2082 n-i-bz Add missing ]] to terminate CDATA.
2083 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
2084 n-i-bz Enable sys_fadvise64_64 on arm32.
2085 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
2086 n-i-bz Add test cases for all known arm64 load/store instructions.
2087 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
2088 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
2089 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
2090 n-i-bz arm64: Implement "fcvtpu w, s".
2091 n-i-bz arm64: implement ADDP and various others
2092 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
2093 n-i-bz arm64: enable FCVT{A,N}S X,S.
2095 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
2099 Release 3.10.0 (10 September 2014)
2100 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2102 3.10.0 is a feature release with many improvements and the usual
2103 collection of bug fixes.
2105 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
2106 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
2107 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
2108 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
2109 significantly improved relative to the 3.9.0 release.
2111 * ================== PLATFORM CHANGES =================
2113 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
2114 is mostly complete, and is usable, but some SIMD instructions are as
2117 * Support for little-endian variant of the 64-bit POWER architecture.
2119 * Support for Android on MIPS32.
2121 * Support for 64bit FPU on MIPS32 platforms.
2123 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
2125 * Configuration for and running on Android targets has changed.
2126 See README.android in the source tree for details.
2128 * ================== DEPRECATED FEATURES =================
2130 * --db-attach is now deprecated and will be removed in the next
2131 valgrind feature release. The built-in GDB server capabilities are
2132 superior and should be used instead. Learn more here:
2133 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
2135 * ==================== TOOL CHANGES ====================
2139 - Client code can now selectively disable and re-enable reporting of
2140 invalid address errors in specific ranges using the new client
2141 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
2142 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
2144 - Leak checker: there is a new leak check heuristic called
2145 "length64". This is used to detect interior pointers pointing 8
2146 bytes inside a block, on the assumption that the first 8 bytes
2147 holds the value "block size - 8". This is used by
2148 sqlite3MemMalloc, for example.
2150 - Checking of system call parameters: if a syscall parameter
2151 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
2152 several fields not initialised, an error is now reported for each
2153 field. Previously, an error was reported only for the first
2154 uninitialised field.
2156 - Mismatched alloc/free checking: a new flag
2157 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
2158 such checks if necessary.
2162 - Improvements to error messages:
2164 o Race condition error message involving heap allocated blocks also
2165 show the thread number that allocated the raced-on block.
2167 o All locks referenced by an error message are now announced.
2168 Previously, some error messages only showed the lock addresses.
2170 o The message indicating where a lock was first observed now also
2171 describes the address/location of the lock.
2173 - Helgrind now understands the Ada task termination rules and
2174 creates a happens-before relationship between a terminated task
2175 and its master. This avoids some false positives and avoids a big
2176 memory leak when a lot of Ada tasks are created and terminated.
2177 The interceptions are only activated with forthcoming releases of
2178 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
2180 - A new GDB server monitor command "info locks" giving the list of
2181 locks, their location, and their status.
2185 - callgrind_control now supports the --vgdb-prefix argument,
2186 which is needed if valgrind was started with this same argument.
2188 * ==================== OTHER CHANGES ====================
2190 * Unwinding through inlined function calls. Stack unwinding can now
2191 make use of Dwarf3 inlined-unwind information if it is available.
2192 The practical effect is that inlined calls become visible in stack
2193 traces. The suppression matching machinery has been adjusted
2194 accordingly. This is controlled by the new option
2195 --read-inline-info=yes|no. Currently this is enabled by default
2196 only on Linux and Android targets and only for the tools Memcheck,
2199 * Valgrind can now read EXIDX unwind information on 32-bit ARM
2200 targets. If an object contains both CFI and EXIDX unwind
2201 information, Valgrind will prefer the CFI over the EXIDX. This
2202 facilitates unwinding through system libraries on arm-android
2205 * Address description logic has been improved and is now common
2206 between Memcheck and Helgrind, resulting in better address
2207 descriptions for some kinds of error messages.
2209 * Error messages about dubious arguments (eg, to malloc or calloc) are
2210 output like other errors. This means that they can be suppressed
2211 and they have a stack trace.
2213 * The C++ demangler has been updated for better C++11 support.
2215 * New and modified GDB server monitor features:
2217 - Thread local variables/storage (__thread) can now be displayed.
2219 - The GDB server monitor command "v.info location <address>"
2220 displays information about an address. The information produced
2221 depends on the tool and on the options given to valgrind.
2222 Possibly, the following are described: global variables, local
2223 (stack) variables, allocated or freed blocks, ...
2225 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
2226 ask the GDB server to stop at the start of program execution, at
2227 the end of the program execution and on Valgrind internal errors.
2229 - A new monitor command "v.info stats" shows various Valgrind core
2230 and tool statistics.
2232 - A new monitor command "v.set hostvisibility" allows the GDB server
2233 to provide access to Valgrind internal host status/memory.
2235 * A new option "--aspace-minaddr=<address>" can in some situations
2236 allow the use of more memory by decreasing the address above which
2237 Valgrind maps memory. It can also be used to solve address
2238 conflicts with system libraries by increasing the default value.
2239 See user manual for details.
2241 * The amount of memory used by Valgrind to store debug info (unwind
2242 info, line number information and symbol data) has been
2243 significantly reduced, even though Valgrind now reads more
2244 information in order to support unwinding of inlined function calls.
2246 * Dwarf3 handling with --read-var-info=yes has been improved:
2248 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
2250 - Code compiled with
2251 -ffunction-sections -fdata-sections -Wl,--gc-sections
2252 no longer causes assertion failures.
2254 * Improved checking for the --sim-hints= and --kernel-variant=
2255 options. Unknown strings are now detected and reported to the user
2258 * The semantics of stack start/end boundaries in the valgrind.h
2259 VALGRIND_STACK_REGISTER client request has been clarified and
2260 documented. The convention is that start and end are respectively
2261 the lowest and highest addressable bytes of the stack.
2263 * ==================== FIXED BUGS ====================
2265 The following bugs have been fixed or resolved. Note that "n-i-bz"
2266 stands for "not in bugzilla" -- that is, a bug that was reported to us
2267 but never got a bugzilla entry. We encourage you to file bugs in
2268 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2269 than mailing the developers (or mailing lists) directly -- bugs that
2270 are not entered into bugzilla tend to get forgotten about or ignored.
2272 To see details of a given bug, visit
2273 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2274 where XXXXXX is the bug number as listed below.
2276 175819 Support for ipv6 socket reporting with --track-fds
2277 232510 make distcheck fails
2278 249435 Analyzing wine programs with callgrind triggers a crash
2279 278972 support for inlined function calls in stacktraces and suppression
2281 291310 FXSAVE instruction marks memory as undefined on amd64
2282 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2283 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2284 315199 vgcore file for threaded app does not show which thread crashed
2285 315952 tun/tap ioctls are not supported
2286 323178 Unhandled instruction: PLDW register (ARM)
2287 323179 Unhandled instruction: PLDW immediate (ARM)
2288 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2289 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2290 325124 [MIPSEL] Compilation error
2291 325477 Phase 4 support for IBM Power ISA 2.07
2292 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2293 325628 Phase 5 support for IBM Power ISA 2.07
2294 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2295 325751 Missing the two privileged Power PC Transactional Memory Instructions
2296 325816 Phase 6 support for IBM Power ISA 2.07
2297 325856 Make SGCheck fail gracefully on unsupported platforms
2298 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2299 326436 DRD: False positive in libstdc++ std::list::push_back
2300 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2301 326462 Refactor vgdb to isolate invoker stuff into separate module
2302 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2303 326623 DRD: false positive conflict report in a field assignment
2304 326724 Valgrind does not compile on OSX 1.9 Mavericks
2305 326816 Intercept for __strncpy_sse2_unaligned missing?
2306 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2307 326983 Clear direction flag after tests on amd64.
2308 327212 Do not prepend the current directory to absolute path names.
2309 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2310 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2311 327284 s390x: Fix translation of the risbg instruction
2312 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2313 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2314 327916 DW_TAG_typedef may have no name
2315 327943 s390x: add a redirection for the 'index' function
2316 328100 XABORT not implemented
2317 328205 Implement additional Xen hypercalls
2318 328454 add support Backtraces with ARM unwind tables (EXIDX)
2319 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2320 328711 valgrind.1 manpage "memcheck options" section is badly generated
2321 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2322 329612 Incorrect handling of AT_BASE for image execution
2323 329694 clang warns about using uninitialized variable
2324 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2325 330228 mmap must align to VKI_SHMLBA on mips32
2326 330257 LLVM does not support `-mno-dynamic-no-pic` option
2327 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2328 330459 --track-fds=yes doesn't track eventfds
2329 330469 Add clock_adjtime syscall support
2330 330594 Missing sysalls on PowerPC / uClibc
2331 330622 Add test to regression suite for POWER instruction: dcbzl
2332 330939 Support for AMD's syscall instruction on x86
2334 330941 Typo in PRE(poll) syscall wrapper
2335 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2336 331254 Fix expected output for memcheck/tests/dw4
2337 331255 Fix race condition in test none/tests/coolo_sigaction
2338 331257 Fix type of jump buffer in test none/tests/faultstatus
2339 331305 configure uses bash specific syntax
2340 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2341 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2342 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2343 331829 Unexpected ioctl opcode sign extension
2344 331830 ppc64: WARNING: unhandled syscall: 96/97
2345 331839 drd/tests/sem_open specifies invalid semaphore name
2346 331847 outcome of drd/tests/thread_name is nondeterministic
2347 332037 Valgrind cannot handle Thumb "add pc, reg"
2348 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2349 consistency checks enabled
2350 332263 intercepts for pthread_rwlock_timedrdlock and
2351 pthread_rwlock_timedwrlock are incorrect
2352 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2354 332276 Implement additional Xen hypercalls
2355 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2356 332765 Fix ms_print to create temporary files in a proper directory
2357 333072 drd: Add semaphore annotations
2358 333145 Tests for missaligned PC+#imm access for arm
2359 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2360 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2361 333248 WARNING: unhandled syscall: unix:443
2362 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2363 333501 cachegrind: assertion: Cache set count is not a power of two.
2366 333666 Recognize MPX instructions and bnd prefix.
2367 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2368 333817 Valgrind reports the memory areas written to by the SG_IO
2370 334049 lzcnt fails silently (x86_32)
2371 334384 Valgrind does not have support Little Endian support for
2373 334585 recvmmsg unhandled (+patch) (arm)
2374 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2375 334727 Build fails with -Werror=format-security
2376 334788 clarify doc about --log-file initial program directory
2377 334834 PPC64 Little Endian support, patch 2
2378 334836 PPC64 Little Endian support, patch 3 testcase fixes
2379 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2380 335034 Unhandled ioctl: HCIGETDEVLIST
2381 335155 vgdb, fix error print statement.
2382 335262 arm64: movi 8bit version is not supported
2383 335263 arm64: dmb instruction is not implemented
2384 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2385 335496 arm64: sbc/abc instructions are not implemented
2386 335554 arm64: unhandled instruction: abs
2387 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2388 335735 arm64: unhandled instruction: cnt
2389 335736 arm64: unhandled instruction: uaddlv
2390 335848 arm64: unhandled instruction: {s,u}cvtf
2391 335902 arm64: unhandled instruction: sli
2392 335903 arm64: unhandled instruction: umull (vector)
2393 336055 arm64: unhandled instruction: mov (element)
2394 336062 arm64: unhandled instruction: shrn{,2}
2395 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2396 336189 arm64: unhandled Instruction: mvn
2397 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2398 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2399 336772 Make moans about unknown ioctls more informative
2400 336957 Add a section about the Solaris/illumos port on the webpage
2401 337094 ifunc wrapper is broken on ppc64
2402 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2403 337528 leak check heuristic for block prefixed by length as 64bit number
2404 337740 Implement additional Xen hypercalls
2405 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2406 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2407 337871 deprecate --db-attach
2408 338023 Add support for all V4L2/media ioctls
2409 338024 inlined functions are not shown if DW_AT_ranges is used
2410 338106 Add support for 'kcmp' syscall
2411 338115 DRD: computed conflict set differs from actual after fork
2412 338160 implement display of thread local storage in gdbsrv
2413 338205 configure.ac and check for -Wno-tautological-compare
2414 338300 coredumps are missing one byte of every segment
2415 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2416 338499 --sim-hints parsing broken due to wrong order in tokens
2417 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2418 338681 Unable to unwind through clone thread created on i386-linux
2419 338698 race condition between gdbsrv and vgdb on startup
2420 338703 helgrind on arm-linux gets false positives in dynamic loader
2421 338791 alt dwz files can be relative of debug/main file
2422 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2423 338932 build V-trunk with gcc-trunk
2424 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2425 345079 Fix build problems in VEX/useful/test_main.c
2426 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2427 n-i-bz s390x: Fix memory corruption for multithreaded applications
2428 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2429 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2430 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2431 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2432 n-i-bz Add command-line option --default-suppressions
2433 n-i-bz Add support for BLKDISCARDZEROES ioctl
2434 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2435 n-i-bz Add support for sys_pivot_root and sys_unshare
2437 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2438 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2439 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2443 Release 3.9.0 (31 October 2013)
2444 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2445 3.9.0 is a feature release with many improvements and the usual
2446 collection of bug fixes.
2448 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2449 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2450 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2451 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2453 * ================== PLATFORM CHANGES =================
2455 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2456 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2458 * Support for MIPS DSP ASE on MIPS32 platforms.
2460 * Support for s390x Decimal Floating Point instructions on hosts that
2461 have the DFP facility installed.
2463 * Support for POWER8 (Power ISA 2.07) instructions
2465 * Support for Intel AVX2 instructions. This is available only on 64
2468 * Initial support for Intel Transactional Synchronization Extensions,
2471 * Initial support for Hardware Transactional Memory on POWER.
2473 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2474 run large GUI apps tolerably well.
2476 * ==================== TOOL CHANGES ====================
2480 - Improvements in handling of vectorised code, leading to
2481 significantly fewer false error reports. You need to use the flag
2482 --partial-loads-ok=yes to get the benefits of these changes.
2484 - Better control over the leak checker. It is now possible to
2485 specify which leak kinds (definite/indirect/possible/reachable)
2486 should be displayed, which should be regarded as errors, and which
2487 should be suppressed by a given leak suppression. This is done
2488 using the options --show-leak-kinds=kind1,kind2,..,
2489 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2490 "match-leak-kinds:" line in suppression entries, respectively.
2492 Note that generated leak suppressions contain this new line and
2493 are therefore more specific than in previous releases. To get the
2494 same behaviour as previous releases, remove the "match-leak-kinds:"
2495 line from generated suppressions before using them.
2497 - Reduced "possible leak" reports from the leak checker by the use
2498 of better heuristics. The available heuristics provide detection
2499 of valid interior pointers to std::stdstring, to new[] allocated
2500 arrays with elements having destructors and to interior pointers
2501 pointing to an inner part of a C++ object using multiple
2502 inheritance. They can be selected individually using the
2503 option --leak-check-heuristics=heur1,heur2,...
2505 - Better control of stacktrace acquisition for heap-allocated
2506 blocks. Using the --keep-stacktraces option, it is possible to
2507 control independently whether a stack trace is acquired for each
2508 allocation and deallocation. This can be used to create better
2509 "use after free" errors or to decrease Valgrind's resource
2510 consumption by recording less information.
2512 - Better reporting of leak suppression usage. The list of used
2513 suppressions (shown when the -v option is given) now shows, for
2514 each leak suppressions, how many blocks and bytes it suppressed
2515 during the last leak search.
2519 - False errors resulting from the use of statically initialised
2520 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2523 - False errors resulting from the use of pthread_cond_waits that
2524 timeout, have been removed.
2526 * ==================== OTHER CHANGES ====================
2528 * Some attempt to tune Valgrind's space requirements to the expected
2529 capabilities of the target:
2531 - The default size of the translation cache has been reduced from 8
2532 sectors to 6 on Android platforms, since each sector occupies
2533 about 40MB when using Memcheck.
2535 - The default size of the translation cache has been increased to 16
2536 sectors on all other platforms, reflecting the fact that large
2537 applications require instrumentation and storage of huge amounts
2538 of code. For similar reasons, the number of memory mapped
2539 segments that can be tracked has been increased by a factor of 6.
2541 - In all cases, the maximum number of sectors in the translation
2542 cache can be controlled by the new flag --num-transtab-sectors.
2544 * Changes in how debug info (line numbers, etc) is read:
2546 - Valgrind no longer temporarily mmaps the entire object to read
2547 from it. Instead, reading is done through a small fixed sized
2548 buffer. This avoids virtual memory usage spikes when Valgrind
2549 reads debuginfo from large shared objects.
2551 - A new experimental remote debug info server. Valgrind can read
2552 debug info from a different machine (typically, a build host)
2553 where debuginfo objects are stored. This can save a lot of time
2554 and hassle when running Valgrind on resource-constrained targets
2555 (phones, tablets) when the full debuginfo objects are stored
2556 somewhere else. This is enabled by the --debuginfo-server=
2559 - Consistency checking between main and debug objects can be
2560 disabled using the --allow-mismatched-debuginfo option.
2562 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2563 scanning can recover stack traces in some cases when the normal
2564 unwind mechanisms fail. Stack scanning is best described as "a
2565 nasty, dangerous and misleading hack" and so is disabled by default.
2566 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2569 * Detection and merging of recursive stack frame cycles. When your
2570 program has recursive algorithms, this limits the memory used by
2571 Valgrind for recorded stack traces and avoids recording
2572 uninteresting repeated calls. This is controlled by the command
2573 line option --merge-recursive-frame and by the monitor command
2574 "v.set merge-recursive-frames".
2576 * File name and line numbers for used suppressions. The list of used
2577 suppressions (shown when the -v option is given) now shows, for each
2578 used suppression, the file name and line number where the suppression
2581 * New and modified GDB server monitor features:
2583 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2584 that can be used to execute gdbserver monitor commands from the
2587 - A new monitor command, "v.info open_fds", that gives the list of
2588 open file descriptors and additional details.
2590 - An optional message in the "v.info n_errs_found" monitor command,
2591 for example "v.info n_errs_found test 1234 finished", allowing a
2592 comment string to be added to the process output, perhaps for the
2593 purpose of separating errors of different tests or test phases.
2595 - A new monitor command "v.info execontext" that shows information
2596 about the stack traces recorded by Valgrind.
2598 - A new monitor command "v.do expensive_sanity_check_general" to run
2599 some internal consistency checks.
2601 * New flag --sigill-diagnostics to control whether a diagnostic
2602 message is printed when the JIT encounters an instruction it can't
2603 translate. The actual behavior -- delivery of SIGILL to the
2604 application -- is unchanged.
2606 * The maximum amount of memory that Valgrind can use on 64 bit targets
2607 has been increased from 32GB to 64GB. This should make it possible
2608 to run applications on Memcheck that natively require up to about 35GB.
2610 * ==================== FIXED BUGS ====================
2612 The following bugs have been fixed or resolved. Note that "n-i-bz"
2613 stands for "not in bugzilla" -- that is, a bug that was reported to us
2614 but never got a bugzilla entry. We encourage you to file bugs in
2615 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2616 than mailing the developers (or mailing lists) directly -- bugs that
2617 are not entered into bugzilla tend to get forgotten about or ignored.
2619 To see details of a given bug, visit
2620 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2621 where XXXXXX is the bug number as listed below.
2623 123837 system call: 4th argument is optional, depending on cmd
2624 135425 memcheck should tell you where Freed blocks were Mallocd
2625 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2626 207815 Adds some of the drm ioctls to syswrap-linux.c
2627 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2628 252955 Impossible to compile with ccache
2629 253519 Memcheck reports auxv pointer accesses as invalid reads.
2630 263034 Crash when loading some PPC64 binaries
2631 269599 Increase deepest backtrace
2632 274695 s390x: Support "compare to/from logical" instructions (z196)
2633 275800 s390x: Autodetect cache info (part 2)
2634 280271 Valgrind reports possible memory leaks on still-reachable std::string
2635 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2636 289578 Backtraces with ARM unwind tables (stack scan flags)
2637 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2638 304832 ppc32: build failure
2639 305431 Use find_buildid shdr fallback for separate .debug files
2640 305728 Add support for AVX2 instructions
2641 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2642 306035 s390x: Fix IR generation for LAAG and friends
2643 306054 s390x: Condition code computation for convert-to-int/logical
2644 306098 s390x: alternate opcode form for convert to/from fixed
2645 306587 Fix cache line detection from auxiliary vector for PPC.
2646 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2647 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2648 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2649 307101 sys_capget second argument can be NULL
2650 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2651 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2652 307113 s390x: DFP support
2653 307141 valgrind does't work in mips-linux system
2654 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2655 307285 x86_amd64 feature test for avx in test suite is wrong
2656 307290 memcheck overlap testcase needs memcpy version filter
2657 307463 Please add "&limit=0" to the "all open bugs" link
2658 307465 --show-possibly-lost=no should reduce the error count / exit code
2659 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2660 307729 pkgconfig support broken valgrind.pc
2661 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2662 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2663 308089 Unhandled syscall on ppc64: prctl
2664 308135 PPC32 MPC8xx has 16 bytes cache size
2665 308321 testsuite memcheck filter interferes with gdb_filter
2667 308341 vgdb should report process exit (or fatal signal)
2668 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2669 308495 Remove build dependency on installed Xen headers
2670 308573 Internal error on 64-bit instruction executed in 32-bit mode
2672 308627 pmovmskb validity bit propagation is imprecise
2673 308644 vgdb command for having the info for the track-fds option
2674 308711 give more info about aspacemgr and arenas in out_of_memory
2675 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2676 308718 ARM implement SMLALBB family of instructions
2677 308886 Missing support for PTRACE_SET/GETREGSET
2678 308930 syscall name_to_handle_at (303 on amd64) not handled
2679 309229 V-bit tester does not report number of tests generated
2680 309323 print unrecognized instuction on MIPS
2681 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2682 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2683 309430 Self hosting ppc64 encounters a vassert error on operand type
2684 309600 valgrind is a bit confused about 0-sized sections
2685 309823 Generate errors for still reachable blocks
2686 309921 PCMPISTRI validity bit propagation is imprecise
2687 309922 none/tests/ppc64/test_dfp5 sometimes fails
2688 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2689 310424 --read-var-info does not properly describe static variables
2690 310792 search additional path for debug symbols
2691 310931 s390x: Message-security assist (MSA) instruction extension [..]
2692 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2693 311318 ARM: "128-bit constant is not implemented" error message
2694 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2695 311690 V crashes because it redirects branches inside of a redirected function
2696 311880 x86_64: make regtest hangs at shell_valid1
2697 311922 WARNING: unhandled syscall: 170
2699 312171 ppc: insn selection for DFP
2700 312571 Rounding mode call wrong for the DFP Iops [..]
2701 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2702 312913 Dangling pointers error should also report the alloc stack trace
2703 312980 Building on Mountain Lion generates some compiler warnings
2704 313267 Adding MIPS64/Linux port to Valgrind
2707 313811 Buffer overflow in assert_fail
2708 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2709 314269 ppc: dead code in insn selection
2710 314718 ARM: implement integer divide instruction (sdiv and udiv)
2711 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2712 315441 sendmsg syscall should ignore unset msghdr msg_flags
2713 315534 msgrcv inside a thread causes valgrind to hang (block)
2714 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2715 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2716 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2717 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2718 316144 valgrind.1 manpage contains unknown ??? strings [..]
2719 316145 callgrind command line options in manpage reference (unknown) [..]
2720 316145 callgrind command line options in manpage reference [..]
2721 316181 drd: Fixed a 4x slowdown for certain applications
2722 316503 Valgrind does not support SSE4 "movntdqa" instruction
2723 316535 Use of |signed int| instead of |size_t| in valgrind messages
2724 316696 fluidanimate program of parsec 2.1 stuck
2725 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2726 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2727 317186 "Impossible happens" when occurs VCVT instruction on ARM
2728 317318 Support for Threading Building Blocks "scalable_malloc"
2729 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2730 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2731 317463 bmi testcase IR SANITY CHECK FAILURE
2732 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2733 318050 libmpiwrap fails to compile with out-of-source build
2734 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2735 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2736 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2737 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2738 318932 Add missing PPC64 and PPC32 system call support
2739 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2740 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2741 319494 VEX Makefile-gcc standalone build update after r2702
2742 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2743 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2744 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2745 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2746 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2747 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2748 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2749 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2750 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2751 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2752 320895 add fanotify support (patch included)
2753 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2754 321065 Valgrind updates for Xen 4.3
2755 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2756 321363 Unhandled instruction: SSAX (ARM + Thumb)
2757 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2758 321466 Unhandled instruction: SHASX (ARM + Thumb)
2759 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2760 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2761 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2762 321620 Unhandled instruction: UASX (ARM + Thumb)
2763 321621 Unhandled instruction: USAX (ARM + Thumb)
2764 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2765 321693 Unhandled instruction: LDRSBT (Thumb)
2766 321694 Unhandled instruction: UQASX (ARM + Thumb)
2767 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2768 321697 Unhandled instruction: UHASX (ARM + Thumb)
2769 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2770 321704 Unhandled instruction: REVSH (ARM + Thumb)
2771 321730 Add cg_diff and cg_merge man pages
2772 321738 Add vgdb and valgrind-listener man pages
2774 321891 Unhandled instruction: LDRHT (Thumb)
2775 321960 pthread_create() then alloca() causing invalid stack write errors
2776 321969 ppc32 and ppc64 don't support [lf]setxattr
2777 322254 Show threadname together with tid if set by application
2778 322294 Add initial support for IBM Power ISA 2.07
2779 322368 Assertion failure in wqthread_hijack under OS X 10.8
2780 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2781 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2782 322851 0bXXX binary literal syntax is not standard
2783 323035 Unhandled instruction: LDRSHT(Thumb)
2784 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2785 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2786 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2787 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2788 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2789 323437 Phase 2 support for IBM Power ISA 2.07
2790 323713 Support mmxext (integer sse) subset on i386 (athlon)
2791 323803 Transactional memory instructions are not supported for Power
2792 323893 SSE3 not available on amd cpus in valgrind
2793 323905 Probable false positive from Valgrind/drd on close()
2794 323912 valgrind.h header isn't compatible for mingw64
2795 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
2796 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
2797 324181 mmap does not handle MAP_32BIT
2798 324227 memcheck false positive leak when a thread calls exit+block [..]
2799 324421 Support for fanotify API on ARM architecture
2800 324514 gdbserver monitor cmd output behaviour consistency [..]
2801 324518 ppc64: Emulation of dcbt instructions does not handle [..]
2802 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
2803 324582 When access is made to freed memory, report both allocation [..]
2804 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
2805 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
2806 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
2807 324834 Unhandled instructions in Microsoft C run-time for x86_64
2808 324894 Phase 3 support for IBM Power ISA 2.07
2809 326091 drd: Avoid false race reports from optimized strlen() impls
2810 326113 valgrind libvex hwcaps error on AMD64
2811 n-i-bz Some wrong command line options could be ignored
2812 n-i-bz patch to allow fair-sched on android
2813 n-i-bz report error for vgdb snapshot requested before execution
2814 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
2816 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
2820 Release 3.8.1 (19 September 2012)
2821 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2822 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
2823 that occur moderately frequently in real use cases, adds support for
2824 some missing instructions on ARM, and fixes a deadlock condition on
2825 MacOSX. If you package or deliver 3.8.0 for others to use, you might
2826 want to consider upgrading to 3.8.1 instead.
2828 The following bugs have been fixed or resolved. Note that "n-i-bz"
2829 stands for "not in bugzilla" -- that is, a bug that was reported to us
2830 but never got a bugzilla entry. We encourage you to file bugs in
2831 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2832 than mailing the developers (or mailing lists) directly -- bugs that
2833 are not entered into bugzilla tend to get forgotten about or ignored.
2835 To see details of a given bug, visit
2836 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2837 where XXXXXX is the bug number as listed below.
2840 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
2841 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
2842 298281 wcslen causes false(?) uninitialised value warnings
2843 301281 valgrind hangs on OS X when the process calls system()
2844 304035 disInstr(arm): unhandled instruction 0xE1023053
2845 304867 implement MOVBE instruction in x86 mode
2846 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
2847 305042 amd64: implement 0F 7F encoding of movq between two registers
2848 305199 ARM: implement QDADD and QDSUB
2849 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
2850 305513 killed by fatal signal: SIGSEGV
2851 305690 DRD reporting invalid semaphore when sem_trywait fails
2852 305926 Invalid alignment checks for some AVX instructions
2853 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
2854 306310 3.8.0 release tarball missing some files
2855 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
2856 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
2857 n-i-bz shmat of a segment > 4Gb does not work
2858 n-i-bz simulate_control_c script wrong USR1 signal number on mips
2859 n-i-bz vgdb ptrace calls wrong on mips [...]
2860 n-i-bz Fixes for more MPI false positives
2861 n-i-bz exp-sgcheck's memcpy causes programs to segfault
2862 n-i-bz OSX build w/ clang: asserts at startup
2863 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
2864 n-i-bz fix a couple of union tag-vs-field mixups
2865 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
2867 The following bugs were fixed in 3.8.0 but not listed in this NEWS
2870 254088 Valgrind should know about UD2 instruction
2873 304754 NEWS blows TeX's little mind
2875 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
2879 Release 3.8.0 (10 August 2012)
2880 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2881 3.8.0 is a feature release with many improvements and the usual
2882 collection of bug fixes.
2884 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2885 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
2886 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
2887 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
2888 There is initial support for MacOSX 10.8, but it is not usable for
2889 serious work at present.
2891 * ================== PLATFORM CHANGES =================
2893 * Support for MIPS32 platforms running Linux. Valgrind has been
2894 tested on MIPS32 and MIPS32r2 platforms running different Debian
2895 Squeeze and MeeGo distributions. Both little-endian and big-endian
2896 cores are supported. The tools Memcheck, Massif and Lackey have
2897 been tested and are known to work. See README.mips for more details.
2899 * Preliminary support for Android running on x86.
2901 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
2903 * Support for Intel AVX instructions and for AES instructions. This
2904 support is available only for 64 bit code.
2906 * Support for POWER Decimal Floating Point instructions.
2908 * ==================== TOOL CHANGES ====================
2910 * Non-libc malloc implementations are now supported. This is useful
2911 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
2912 Using the new option --soname-synonyms, such tools can be informed
2913 that the malloc implementation is either linked statically into the
2914 executable, or is present in some other shared library different
2915 from libc.so. This makes it possible to process statically linked
2916 programs, and programs using other malloc libraries, for example
2917 TCMalloc or JEMalloc.
2919 * For tools that provide their own replacement for malloc et al, the
2920 option --redzone-size=<number> allows users to specify the size of
2921 the padding blocks (redzones) added before and after each client
2922 allocated block. Smaller redzones decrease the memory needed by
2923 Valgrind. Bigger redzones increase the chance to detect blocks
2924 overrun or underrun. Prior to this change, the redzone size was
2925 hardwired to 16 bytes in Memcheck.
2929 - The leak_check GDB server monitor command now can
2930 control the maximum nr of loss records to output.
2932 - Reduction of memory use for applications allocating
2933 many blocks and/or having many partially defined bytes.
2935 - Addition of GDB server monitor command 'block_list' that lists
2936 the addresses/sizes of the blocks of a leak search loss record.
2938 - Addition of GDB server monitor command 'who_points_at' that lists
2939 the locations pointing at a block.
2941 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2942 detect an invalid access of these redzones, by marking them
2943 noaccess. Similarly, if a redzone size is given for a memory
2944 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2945 This still allows to find some bugs if the user has forgotten to
2946 mark the pool superblock noaccess.
2948 - Performance of memory leak check has been improved, especially in
2949 cases where there are many leaked blocks and/or many suppression
2950 rules used to suppress leak reports.
2952 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2953 more precise analysis, which is important for LLVM/Clang
2954 generated code. This is at the cost of somewhat reduced
2955 performance. Note there is no change to analysis precision or
2956 costs on Linux targets.
2960 - Added even more facilities that can help finding the cause of a data
2961 race, namely the command-line option --ptrace-addr and the macro
2962 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2964 - Fixed a subtle bug that could cause false positive data race reports.
2966 * ==================== OTHER CHANGES ====================
2968 * The C++ demangler has been updated so as to work well with C++
2969 compiled by up to at least g++ 4.6.
2971 * Tool developers can make replacement/wrapping more flexible thanks
2972 to the new option --soname-synonyms. This was reported above, but
2973 in fact is very general and applies to all function
2974 replacement/wrapping, not just to malloc-family functions.
2976 * Round-robin scheduling of threads can be selected, using the new
2977 option --fair-sched= yes. Prior to this change, the pipe-based
2978 thread serialisation mechanism (which is still the default) could
2979 give very unfair scheduling. --fair-sched=yes improves
2980 responsiveness of interactive multithreaded applications, and
2981 improves repeatability of results from the thread checkers Helgrind
2984 * For tool developers: support to run Valgrind on Valgrind has been
2985 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2987 * gdbserver now shows the float shadow registers as integer
2988 rather than float values, as the shadow values are mostly
2989 used as bit patterns.
2991 * Increased limit for the --num-callers command line flag to 500.
2993 * Performance improvements for error matching when there are many
2994 suppression records in use.
2996 * Improved support for DWARF4 debugging information (bug 284184).
2998 * Initial support for DWZ compressed Dwarf debug info.
3000 * Improved control over the IR optimiser's handling of the tradeoff
3001 between performance and precision of exceptions. Specifically,
3002 --vex-iropt-precise-memory-exns has been removed and replaced by
3003 --vex-iropt-register-updates, with extended functionality. This
3004 allows the Valgrind gdbserver to always show up to date register
3007 * Modest performance gains through the use of translation chaining for
3010 * ==================== FIXED BUGS ====================
3012 The following bugs have been fixed or resolved. Note that "n-i-bz"
3013 stands for "not in bugzilla" -- that is, a bug that was reported to us
3014 but never got a bugzilla entry. We encourage you to file bugs in
3015 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3016 than mailing the developers (or mailing lists) directly -- bugs that
3017 are not entered into bugzilla tend to get forgotten about or ignored.
3019 To see details of a given bug, visit
3020 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3021 where XXXXXX is the bug number as listed below.
3023 197914 Building valgrind from svn now requires automake-1.10
3024 203877 increase to 16Mb maximum allowed alignment for memalign et al
3025 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
3026 247386 make perf does not run all performance tests
3027 270006 Valgrind scheduler unfair
3028 270777 Adding MIPS/Linux port to Valgrind
3029 270796 s390x: Removed broken support for the TS insn
3030 271438 Fix configure for proper SSE4.2 detection
3031 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
3032 273475 Add support for AVX instructions
3033 274078 improved configure logic for mpicc
3034 276993 fix mremap 'no thrash checks'
3035 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
3036 281482 memcheck incorrect byte allocation count in realloc() for silly argument
3037 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
3038 283413 Fix wrong sanity check
3039 283671 Robustize alignment computation in LibVEX_Alloc
3040 283961 Adding support for some HCI IOCTLs
3041 284124 parse_type_DIE: confused by: DWARF 4
3042 284864 == 273475 (Add support for AVX instructions)
3043 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
3044 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
3045 285725 == 273475 (Add support for AVX instructions)
3046 286261 add wrapper for linux I2C_RDWR ioctl
3047 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
3048 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
3049 286384 configure fails "checking for a supported version of gcc"
3050 286497 == 273475 (Add support for AVX instructions)
3051 286596 == 273475 (Add support for AVX instructions)
3052 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
3053 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
3054 287260 Incorrect conditional jump or move depends on uninitialised value(s)
3055 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
3056 287307 == 273475 (Add support for AVX instructions)
3057 287858 VG_(strerror): unknown error
3058 288298 (MacOSX) unhandled syscall shm_unlink
3059 288995 == 273475 (Add support for AVX instructions)
3060 289470 Loading of large Mach-O thin binaries fails.
3061 289656 == 273475 (Add support for AVX instructions)
3062 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
3063 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
3064 289839 s390x: Provide support for unicode conversion instructions
3065 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
3066 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
3067 290655 Add support for AESKEYGENASSIST instruction
3068 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
3069 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
3070 291253 ES register not initialised in valgrind simulation
3071 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
3072 291865 s390x: Support the "Compare Double and Swap" family of instructions
3073 292300 == 273475 (Add support for AVX instructions)
3074 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
3075 292493 == 273475 (Add support for AVX instructions)
3076 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
3077 292627 Missing support for some SCSI ioctls
3078 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
3079 292841 == 273475 (Add support for AVX instructions)
3080 292993 implement the getcpu syscall on amd64-linux
3081 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
3082 293088 Add some VEX sanity checks for ppc64 unhandled instructions
3083 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
3084 293754 PCMPxSTRx not implemented for 16-bit characters
3085 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
3086 293808 CLFLUSH not supported by latest VEX for amd64
3087 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
3088 294048 MPSADBW instruction not implemented
3089 294055 regtest none/tests/shell fails when locale is not set to C
3090 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
3091 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
3092 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
3093 294260 disInstr_AMD64: disInstr miscalculated next %rip
3094 294523 --partial-loads-ok=yes causes false negatives
3095 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
3096 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
3097 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
3098 295089 can not annotate source for both helgrind and drd
3099 295221 POWER Processor decimal floating point instruction support missing
3100 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
3101 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
3102 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
3103 295617 ARM - Add some missing syscalls
3104 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
3105 296229 Linux user input device ioctls missing wrappers
3106 296318 ELF Debug info improvements (more than one rx/rw mapping)
3107 296422 Add translation chaining support
3108 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
3109 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
3110 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
3111 297078 gdbserver signal handling problems [..]
3112 297147 drd false positives on newly allocated memory
3113 297329 disallow decoding of IBM Power DFP insns on some machines
3114 297497 POWER Processor decimal floating point instruction support missing
3115 297701 Another alias for strncasecmp_l in libc-2.13.so
3116 297911 'invalid write' not reported when using APIs for custom mem allocators.
3117 297976 s390x: revisit EX implementation
3118 297991 Valgrind interferes with mmap()+ftell()
3119 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
3120 297993 Fix compilation of valgrind with gcc -g3.
3121 298080 POWER Processor DFP support missing, part 3
3122 298227 == 273475 (Add support for AVX instructions)
3123 298335 == 273475 (Add support for AVX instructions)
3124 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
3125 298394 s390x: Don't bail out on an unknown machine model. [..]
3126 298421 accept4() syscall (366) support is missing for ARM
3127 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
3128 298732 valgrind installation problem in ubuntu with kernel version 3.x
3129 298862 POWER Processor DFP instruction support missing, part 4
3130 298864 DWARF reader mis-parses DW_FORM_ref_addr
3131 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
3132 299053 Support DWARF4 DW_AT_high_pc constant form
3133 299104 == 273475 (Add support for AVX instructions)
3134 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
3135 299629 dup3() syscall (358) support is missing for ARM
3136 299694 POWER Processor DFP instruction support missing, part 5
3137 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
3138 299803 == 273475 (Add support for AVX instructions)
3139 299804 == 273475 (Add support for AVX instructions)
3140 299805 == 273475 (Add support for AVX instructions)
3141 300140 ARM - Missing (T1) SMMUL
3142 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
3143 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
3144 300414 FCOM and FCOMP unimplemented for amd64 guest
3145 301204 infinite loop in canonicaliseSymtab with ifunc symbol
3146 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
3147 301265 add x86 support to Android build
3148 301984 configure script doesn't detect certain versions of clang
3149 302205 Fix compiler warnings for POWER VEX code and POWER test cases
3150 302287 Unhandled movbe instruction on Atom processors
3151 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
3152 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
3153 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
3154 302656 == 273475 (Add support for AVX instructions)
3155 302709 valgrind for ARM needs extra tls support for android emulator [..]
3156 302827 add wrapper for CDROM_GET_CAPABILITY
3157 302901 Valgrind crashes with dwz optimized debuginfo
3158 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
3159 303116 Add support for the POWER instruction popcntb
3160 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
3161 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
3162 303466 == 273475 (Add support for AVX instructions)
3163 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
3164 303963 strstr() function produces wrong results under valgrind callgrind
3165 304054 CALL_FN_xx macros need to enforce stack alignment
3166 304561 tee system call not supported
3167 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
3168 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
3169 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
3170 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
3171 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
3172 n-i-bz s390x: Shadow registers can now be examined using vgdb
3174 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
3175 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
3179 Release 3.7.0 (5 November 2011)
3180 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3181 3.7.0 is a feature release with many significant improvements and the
3182 usual collection of bug fixes.
3184 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3185 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
3186 Support for recent distros and toolchain components (glibc 2.14, gcc
3187 4.6, MacOSX 10.7) has been added.
3189 * ================== PLATFORM CHANGES =================
3191 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
3192 analyse 64-bit programs running on z/Architecture. Most user space
3193 instructions up to and including z10 are supported. Valgrind has
3194 been tested extensively on z9, z10, and z196 machines running SLES
3195 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
3196 known to work well. Callgrind, Helgrind, and DRD work reasonably
3197 well on z9 and later models. See README.s390 for more details.
3199 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
3200 64-bit processes are supported. Some complex threaded applications
3201 (Firefox) are observed to hang when run as 32 bit applications,
3202 whereas 64-bit versions run OK. The cause is unknown. Memcheck
3203 will likely report some false errors. In general, expect some rough
3204 spots. This release also supports MacOSX 10.6, but drops support
3207 * Preliminary support for Android (on ARM). Valgrind can now run
3208 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
3209 README.android for more details, plus instructions on how to get
3212 * Support for the IBM Power ISA 2.06 (Power7 instructions)
3214 * General correctness and performance improvements for ARM/Linux, and,
3215 by extension, ARM/Android.
3217 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
3218 instruction set support is under development but is not available in
3221 * Support for AIX5 has been removed.
3223 * ==================== TOOL CHANGES ====================
3225 * Memcheck: some incremental changes:
3227 - reduction of memory use in some circumstances
3229 - improved handling of freed memory, which in some circumstances
3230 can cause detection of use-after-free that would previously have
3233 - fix of a longstanding bug that could cause false negatives (missed
3234 errors) in programs doing vector saturated narrowing instructions.
3236 * Helgrind: performance improvements and major memory use reductions,
3237 particularly for large, long running applications which perform many
3238 synchronisation (lock, unlock, etc) events. Plus many smaller
3241 - display of locksets for both threads involved in a race
3243 - general improvements in formatting/clarity of error messages
3245 - addition of facilities and documentation regarding annotation
3246 of thread safe reference counted C++ classes
3248 - new flag --check-stack-refs=no|yes [yes], to disable race checking
3249 on thread stacks (a performance hack)
3251 - new flag --free-is-write=no|yes [no], to enable detection of races
3252 where one thread accesses heap memory but another one frees it,
3253 without any coordinating synchronisation event
3255 * DRD: enabled XML output; added support for delayed thread deletion
3256 in order to detect races that occur close to the end of a thread
3257 (--join-list-vol); fixed a memory leak triggered by repeated client
3258 memory allocatation and deallocation; improved Darwin support.
3260 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
3262 * exp-sgcheck: this tool has been reduced in scope so as to improve
3263 performance and remove checking that Memcheck does better.
3264 Specifically, the ability to check for overruns for stack and global
3265 arrays is unchanged, but the ability to check for overruns of heap
3266 blocks has been removed. The tool has accordingly been renamed to
3267 exp-sgcheck ("Stack and Global Array Checking").
3269 * ==================== OTHER CHANGES ====================
3271 * GDB server: Valgrind now has an embedded GDB server. That means it
3272 is possible to control a Valgrind run from GDB, doing all the usual
3273 things that GDB can do (single stepping, breakpoints, examining
3274 data, etc). Tool-specific functionality is also available. For
3275 example, it is possible to query the definedness state of variables
3276 or memory from within GDB when running Memcheck; arbitrarily large
3277 memory watchpoints are supported, etc. To use the GDB server, start
3278 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3281 * Improved support for unfriendly self-modifying code: a new option
3282 --smc-check=all-non-file is available. This adds the relevant
3283 consistency checks only to code that originates in non-file-backed
3284 mappings. In effect this confines the consistency checking only to
3285 code that is or might be JIT generated, and avoids checks on code
3286 that must have been compiled ahead of time. This significantly
3287 improves performance on applications that generate code at run time.
3289 * It is now possible to build a working Valgrind using Clang-2.9 on
3292 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3293 These enable and disable error reporting on a per-thread, and
3294 nestable, basis. This is useful for hiding errors in particularly
3295 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3296 now uses this facility.
3298 * Added the --mod-funcname option to cg_diff.
3300 * ==================== FIXED BUGS ====================
3302 The following bugs have been fixed or resolved. Note that "n-i-bz"
3303 stands for "not in bugzilla" -- that is, a bug that was reported to us
3304 but never got a bugzilla entry. We encourage you to file bugs in
3305 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3306 mailing the developers (or mailing lists) directly -- bugs that are
3307 not entered into bugzilla tend to get forgotten about or ignored.
3309 To see details of a given bug, visit
3310 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3311 where XXXXXX is the bug number as listed below.
3313 79311 malloc silly arg warning does not give stack trace
3314 210935 port valgrind.h (not valgrind) to win32 to support client requests
3315 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3316 243404 Port to zSeries
3317 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3318 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3319 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3320 253206 Some fixes for the faultstatus testcase
3321 255223 capget testcase fails when running as root
3322 256703 xlc_dbl_u32.c testcase broken
3323 256726 Helgrind tests have broken inline asm
3324 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3325 264800 testcase compile failure on zseries
3326 265762 make public VEX headers compilable by G++ 3.x
3327 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3328 266753 configure script does not give the user the option to not use QtCore
3329 266931 gen_insn_test.pl is broken
3330 266961 ld-linux.so.2 i?86-linux strlen issues
3331 266990 setns instruction causes false positive
3332 267020 Make directory for temporary files configurable at run-time.
3333 267342 == 267997 (segmentation fault on Mac OS 10.6)
3334 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3335 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3336 267488 regtest: darwin support for 64-bit build
3337 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3338 267630 Add support for IBM Power ISA 2.06 -- stage 1
3339 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3340 267819 Add client request for informing the core about reallocation
3341 267925 laog data structure quadratic for a single sequence of lock
3342 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3343 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3344 268513 missed optimizations in fold_Expr
3345 268619 s390x: fpr - gpr transfer facility
3346 268620 s390x: reconsider "long displacement" requirement
3347 268621 s390x: improve IR generation for XC
3348 268715 s390x: FLOGR is not universally available
3349 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3350 268930 s390x: MHY is not universally available
3351 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3352 269079 Support ptrace system call on ARM
3353 269144 missing "Bad option" error message
3354 269209 conditional load and store facility (z196)
3355 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3356 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3357 269736 s390x: minor code generation tweaks
3358 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3359 269863 s390x: remove unused function parameters
3360 269864 s390x: tweak s390_emit_load_cc
3361 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3362 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3363 270115 s390x: rewrite some testcases
3364 270309 == 267997 (valgrind crash on startup)
3365 270320 add support for Linux FIOQSIZE ioctl() call
3366 270326 segfault while trying to sanitize the environment passed to execle
3367 270794 IBM POWER7 support patch causes regression in none/tests
3368 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3369 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3370 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3371 270959 s390x: invalid use of R0 as base register
3372 271042 VSX configure check fails when it should not
3373 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3374 271259 s390x: fix code confusion
3375 271337 == 267997 (Valgrind segfaults on MacOS X)
3376 271385 s390x: Implement Ist_MBE
3377 271501 s390x: misc cleanups
3378 271504 s390x: promote likely and unlikely
3379 271579 ppc: using wrong enum type
3380 271615 unhandled instruction "popcnt" (arch=amd10h)
3381 271730 Fix bug when checking ioctls: duplicate check
3382 271776 s390x: provide STFLE instruction support
3383 271779 s390x: provide clock instructions like STCK
3384 271799 Darwin: ioctls without an arg report a memory error
3385 271820 arm: fix type confusion
3386 271917 pthread_cond_timedwait failure leads to not-locked false positive
3387 272067 s390x: fix DISP20 macro
3388 272615 A typo in debug output in mc_leakcheck.c
3389 272661 callgrind_annotate chokes when run from paths containing regex chars
3390 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3391 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3392 272967 make documentation build-system more robust
3393 272986 Fix gcc-4.6 warnings with valgrind.h
3394 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3395 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3396 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3397 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3398 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3399 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3400 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3401 273778 exp-ptrcheck: unhandled sysno == 259
3402 274089 exp-ptrcheck: unhandled sysno == 208
3403 274378 s390x: Various dispatcher tweaks
3404 274447 WARNING: unhandled syscall: 340
3405 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3406 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3407 274926 valgrind does not build against linux-3
3408 275148 configure FAIL with glibc-2.14
3409 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3410 275168 Make Valgrind work for MacOSX 10.7 Lion
3411 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3412 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3413 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3414 275308 Fix implementation for ppc64 fres instruc
3415 275339 s390x: fix testcase compile warnings
3416 275517 s390x: Provide support for CKSM instruction
3417 275710 s390x: get rid of redundant address mode calculation
3418 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3419 275852 == 250101 (valgrind uses all swap space and is killed)
3420 276784 Add support for IBM Power ISA 2.06 -- stage 3
3421 276987 gdbsrv: fix tests following recent commits
3422 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3423 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3424 277471 Unhandled syscall: 340
3425 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3426 277653 ARM: support Thumb2 PLD instruction
3427 277663 ARM: NEON float VMUL by scalar incorrect
3428 277689 ARM: tests for VSTn with register post-index are broken
3429 277694 ARM: BLX LR instruction broken in ARM mode
3430 277780 ARM: VMOV.F32 (immediate) instruction is broken
3431 278057 fuse filesystem syscall deadlocks
3432 278078 Unimplemented syscall 280 on ppc32
3433 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3434 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3435 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3436 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3437 279027 Support for MVCL and CLCL instruction
3438 279027 s390x: Provide support for CLCL and MVCL instructions
3439 279062 Remove a redundant check in the insn selector for ppc.
3440 279071 JDK creates PTEST with redundant REX.W prefix
3441 279212 gdbsrv: add monitor cmd v.info scheduler.
3442 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3443 279698 memcheck discards valid-bits for packuswb
3444 279795 memcheck reports uninitialised values for mincore on amd64
3445 279994 Add support for IBM Power ISA 2.06 -- stage 3
3446 280083 mempolicy syscall check errors
3447 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3448 280710 s390x: config files for nightly builds
3449 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3450 280965 Valgrind breaks fcntl locks when program does mmap
3451 281138 WARNING: unhandled syscall: 340
3452 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3453 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3454 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3455 281468 s390x: handle do_clone and gcc clones in call traces
3456 281488 ARM: VFP register corruption
3457 281828 == 275284 (false memmove warning: "Source and destination overlap")
3458 281883 s390x: Fix system call wrapper for "clone".
3459 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3460 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3461 282238 SLES10: make check fails
3462 282979 strcasestr needs replacement with recent(>=2.12) glibc
3463 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3464 283243 Regression in ppc64 memcheck tests
3465 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3466 283427 re-connect epoll_pwait syscall on ARM linux
3467 283600 gdbsrv: android: port vgdb.c
3468 283709 none/tests/faultstatus needs to account for page size
3469 284305 filter_gdb needs enhancement to work on ppc64
3470 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3471 284472 Thumb2 ROR.W encoding T2 not implemented
3472 284621 XML-escape process command line in XML output
3473 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3474 that have non-power-of-2 sizes (also AMDs)
3475 n-i-bz don't be spooked by libraries mashed by elfhack
3476 n-i-bz don't be spooked by libxul.so linked with gold
3477 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3479 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3480 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3481 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3485 Release 3.6.1 (16 February 2011)
3486 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3487 3.6.1 is a bug fix release. It adds support for some SSE4
3488 instructions that were omitted in 3.6.0 due to lack of time. Initial
3489 support for glibc-2.13 has been added. A number of bugs causing
3490 crashing or assertion failures have been fixed.
3492 The following bugs have been fixed or resolved. Note that "n-i-bz"
3493 stands for "not in bugzilla" -- that is, a bug that was reported to us
3494 but never got a bugzilla entry. We encourage you to file bugs in
3495 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3496 mailing the developers (or mailing lists) directly -- bugs that are
3497 not entered into bugzilla tend to get forgotten about or ignored.
3499 To see details of a given bug, visit
3500 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3501 where XXXXXX is the bug number as listed below.
3503 188572 Valgrind on Mac should suppress setenv() mem leak
3504 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3505 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3506 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3507 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3508 254420 memory pool tracking broken
3509 254957 Test code failing to compile due to changes in memcheck.h
3510 255009 helgrind/drd: crash on chmod with invalid parameter
3511 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3512 255355 helgrind/drd: crash on threaded programs doing fork
3514 255418 (SSE4.x) rint call compiled with ICC
3515 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3516 255888 closing valgrindoutput tag outputted to log-stream on error
3517 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3518 255966 Slowness when using mempool annotations
3519 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3520 256600 super-optimized strcasecmp() false positive
3521 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3522 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3523 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3524 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3525 257276 Missing case in memcheck --track-origins=yes
3526 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3527 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3528 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3529 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3530 263099 callgrind_annotate counts Ir improperly [...]
3531 263877 undefined coprocessor instruction on ARMv7
3532 265964 configure FAIL with glibc-2.13
3533 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3534 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3535 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3536 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3537 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3539 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3543 Release 3.6.0 (21 October 2010)
3544 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3545 3.6.0 is a feature release with many significant improvements and the
3546 usual collection of bug fixes.
3548 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3549 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3550 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3552 -------------------------
3554 Here are some highlights. Details are shown further down:
3556 * Support for ARM/Linux.
3558 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3560 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3562 * Support for the SSE4.2 instruction set.
3564 * Enhancements to the Callgrind profiler, including the ability to
3565 handle CPUs with three levels of cache.
3567 * A new experimental heap profiler, DHAT.
3569 * A huge number of bug fixes and small enhancements.
3571 -------------------------
3573 Here are details of the above changes, together with descriptions of
3574 many other changes, and a list of fixed bugs.
3576 * ================== PLATFORM CHANGES =================
3578 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3579 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3580 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3582 This requires a CPU capable of running the ARMv7-A instruction set
3583 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3584 of the user space instruction set, including ARM and Thumb integer
3585 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3586 Cachegrind and Massif tools work properly; other tools work to
3589 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3590 with support for recent releases of the underlying toolchain
3591 components, notably gcc-4.5 and glibc-2.12.
3593 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3594 support also works much better on OS X 10.5, and is as solid as
3597 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3598 64-bit mode. In 32-bit mode, support is only available up to and
3599 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3600 supported in 64-bit mode, and 32-bit mode does in fact support the
3601 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3602 10.6 on 32-bit targets.
3604 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3605 and including version 2.05 is supported.
3607 * ==================== TOOL CHANGES ====================
3609 * Cachegrind has a new processing script, cg_diff, which finds the
3610 difference between two profiles. It's very useful for evaluating
3611 the performance effects of a change in a program.
3613 Related to this change, the meaning of cg_annotate's (rarely-used)
3614 --threshold option has changed; this is unlikely to affect many
3615 people, if you do use it please see the user manual for details.
3617 * Callgrind now can do branch prediction simulation, similar to
3618 Cachegrind. In addition, it optionally can count the number of
3619 executed global bus events. Both can be used for a better
3620 approximation of a "Cycle Estimation" as derived event (you need to
3621 update the event formula in KCachegrind yourself).
3623 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3624 rather than the L2 cache. This is to accommodate machines with
3625 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3626 cache configuration of such a machine it will run the simulation as
3627 if the L2 cache isn't present. This means the results are less
3628 likely to match the true result for the machine, but
3629 Cachegrind/Callgrind's results are already only approximate, and
3630 should not be considered authoritative. The results are still
3631 useful for giving a general idea about a program's locality.
3633 * Massif has a new option, --pages-as-heap, which is disabled by
3634 default. When enabled, instead of tracking allocations at the level
3635 of heap blocks (as allocated with malloc/new/new[]), it instead
3636 tracks memory allocations at the level of memory pages (as mapped by
3637 mmap, brk, etc). Each mapped page is treated as its own block.
3638 Interpreting the page-level output is harder than the heap-level
3639 output, but this option is useful if you want to account for every
3640 byte of memory used by a program.
3642 * DRD has two new command-line options: --free-is-write and
3643 --trace-alloc. The former allows to detect reading from already freed
3644 memory, and the latter allows tracing of all memory allocations and
3647 * DRD has several new annotations. Custom barrier implementations can
3648 now be annotated, as well as benign races on static variables.
3650 * DRD's happens before / happens after annotations have been made more
3651 powerful, so that they can now also be used to annotate e.g. a smart
3652 pointer implementation.
3654 * Helgrind's annotation set has also been drastically improved, so as
3655 to provide to users a general set of annotations to describe locks,
3656 semaphores, barriers and condition variables. Annotations to
3657 describe thread-safe reference counted heap objects have also been
3660 * Memcheck has a new command-line option, --show-possibly-lost, which
3661 is enabled by default. When disabled, the leak detector will not
3662 show possibly-lost blocks.
3664 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3665 has been added. DHAT keeps track of allocated heap blocks, and also
3666 inspects every memory reference to see which block (if any) is being
3667 accessed. This gives a lot of insight into block lifetimes,
3668 utilisation, turnover, liveness, and the location of hot and cold
3669 fields. You can use DHAT to do hot-field profiling.
3671 * ==================== OTHER CHANGES ====================
3673 * Improved support for unfriendly self-modifying code: the extra
3674 overhead incurred by --smc-check=all has been reduced by
3675 approximately a factor of 5 as compared with 3.5.0.
3677 * Ability to show directory names for source files in error messages.
3678 This is combined with a flexible mechanism for specifying which
3679 parts of the paths should be shown. This is enabled by the new flag
3682 * A new flag, --require-text-symbol, which will stop the run if a
3683 specified symbol is not found it a given shared object when it is
3684 loaded into the process. This makes advanced working with function
3685 intercepting and wrapping safer and more reliable.
3687 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3688 and control of Valgrind is now available for the tools Memcheck and
3689 Helgrind. XML output from Valgrind is available for Memcheck,
3690 Helgrind and exp-Ptrcheck.
3692 * More reliable stack unwinding on amd64-linux, particularly in the
3693 presence of function wrappers, and with gcc-4.5 compiled code.
3695 * Modest scalability (performance improvements) for massive
3696 long-running applications, particularly for those with huge amounts
3699 * Support for analyzing programs running under Wine with has been
3700 improved. The header files <valgrind/valgrind.h>,
3701 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3702 Windows-programs compiled with MinGW or one of the Microsoft Visual
3705 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3706 The 32-bit simulator was not affected. This did not occur often,
3707 but when it did would usually crash the program under test.
3710 * A large number of bugs were fixed. These are shown below.
3712 * A number of bugs were investigated, and were candidates for fixing,
3713 but are not fixed in 3.6.0, due to lack of developer time. They may
3714 get fixed in later releases. They are:
3716 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3717 212419 false positive "lock order violated" (A+B vs A)
3718 213685 Undefined value propagates past dependency breaking instruction
3719 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3720 237920 valgrind segfault on fork failure
3721 242137 support for code compiled by LLVM-2.8
3722 242423 Another unknown Intel cache config value
3723 243232 Inconsistent Lock Orderings report with trylock
3724 243483 ppc: callgrind triggers VEX assertion failure
3725 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3726 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3728 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3729 249435 Analyzing wine programs with callgrind triggers a crash
3730 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3731 250065 Handling large allocations
3732 250101 huge "free" memory usage due to m_mallocfree.c
3733 "superblocks fragmentation"
3734 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3735 252091 Callgrind on ARM does not detect function returns correctly
3736 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3737 254420 memory pool tracking broken
3738 n-i-bz support for adding symbols for JIT generated code
3741 The following bugs have been fixed or resolved. Note that "n-i-bz"
3742 stands for "not in bugzilla" -- that is, a bug that was reported to us
3743 but never got a bugzilla entry. We encourage you to file bugs in
3744 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3745 mailing the developers (or mailing lists) directly -- bugs that are
3746 not entered into bugzilla tend to get forgotten about or ignored.
3748 To see details of a given bug, visit
3749 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3750 where XXXXXX is the bug number as listed below.
3752 135264 dcbzl instruction missing
3754 153699 Valgrind should report unaligned reads with movdqa
3756 190429 Valgrind reports lost of errors in ld.so
3757 with x86_64 2.9.90 glibc
3758 197266 valgrind appears to choke on the xmms instruction
3760 197988 Crash when demangling very large symbol names
3761 202315 unhandled syscall: 332 (inotify_init1)
3762 203256 Add page-level profiling to Massif
3763 205093 dsymutil=yes needs quotes, locking (partial fix)
3764 205241 Snow Leopard 10.6 support (partial fix)
3765 206600 Leak checker fails to upgrade indirect blocks when their
3766 parent becomes reachable
3767 210935 port valgrind.h (not valgrind) to win32 so apps run under
3768 wine can make client requests
3769 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3770 within Linux ip-stack checksum functions
3771 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3773 213685 Undefined value propagates past dependency breaking instruction
3775 215914 Valgrind inserts bogus empty environment variable
3777 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3778 222545 shmat fails under valgind on some arm targets
3779 222560 ARM NEON support
3782 232509 Docs build fails with formatting inside <title></title> elements
3784 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3785 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3786 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3787 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3788 237485 symlink (syscall 57) is not supported on Mac OS
3789 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3791 238208 is_just_below_ESP doesn't take into account red-zone
3792 238345 valgrind passes wrong $0 when executing a shell script
3793 238679 mq_timedreceive syscall doesn't flag the reception buffer
3795 238696 fcntl command F_DUPFD_CLOEXEC not supported
3796 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3797 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3798 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
3800 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
3806 242606 unhandled syscall: setegid (in Ptrcheck)
3807 242814 Helgrind "Impossible has happened" during
3808 QApplication::initInstance();
3809 243064 Valgrind attempting to read debug information from iso
3810 243270 Make stack unwinding in Valgrind wrappers more reliable
3811 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
3812 sysno = 277 (mq_open)
3813 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
3814 244493 ARM VFP d16-d31 registers support
3815 244670 add support for audit_session_self syscall on Mac OS 10.6
3816 244921 The xml report of helgrind tool is not well format
3817 244923 In the xml report file, the <preamble> not escape the
3818 xml char, eg '<','&','>'
3819 245535 print full path names in plain text reports
3820 245925 x86-64 red zone handling problem
3821 246258 Valgrind not catching integer underruns + new [] s
3822 246311 reg/reg cmpxchg doesn't work on amd64
3823 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
3824 246888 Improve Makefile.vex.am
3825 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
3826 to [f]chmod_extended
3827 247526 IBM POWER6 (ISA 2.05) support is incomplete
3828 247561 Some leak testcases fails due to reachable addresses in
3830 247875 sizeofIRType to handle Ity_I128
3831 247894 [PATCH] unhandled syscall sys_readahead
3832 247980 Doesn't honor CFLAGS passed to configure
3833 248373 darwin10.supp is empty in the trunk
3834 248822 Linux FIBMAP ioctl has int parameter instead of long
3835 248893 [PATCH] make readdwarf.c big endianess safe to enable
3836 unwinding on big endian systems
3837 249224 Syscall 336 not supported (SYS_proc_info)
3839 249775 Incorrect scheme for detecting NEON capabilities of host CPU
3840 249943 jni JVM init fails when using valgrind
3841 249991 Valgrind incorrectly declares AESKEYGENASSIST support
3843 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
3844 250799 frexp$fenv_access_off function generates SIGILL
3845 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
3846 251251 support pclmulqdq insn
3847 251362 valgrind: ARM: attach to debugger either fails or provokes
3849 251674 Unhandled syscall 294
3852 254257 Add support for debugfiles found by build-id
3853 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
3854 254646 Wrapped functions cause stack misalignment on OS X
3855 (and possibly Linux)
3856 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
3858 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
3862 Release 3.5.0 (19 August 2009)
3863 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3864 3.5.0 is a feature release with many significant improvements and the
3865 usual collection of bug fixes. The main improvement is that Valgrind
3866 now works on Mac OS X.
3868 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
3869 and X86/Darwin. Support for recent distros and toolchain components
3870 (glibc 2.10, gcc 4.5) has been added.
3872 -------------------------
3874 Here is a short summary of the changes. Details are shown further
3877 * Support for Mac OS X (10.5.x).
3879 * Improvements and simplifications to Memcheck's leak checker.
3881 * Clarification and simplifications in various aspects of Valgrind's
3884 * XML output for Helgrind and Ptrcheck.
3886 * Performance and stability improvements for Helgrind and DRD.
3888 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
3890 * A new experimental tool, BBV, useful for computer architecture
3893 * Improved Wine support, including ability to read Windows PDB
3896 -------------------------
3898 Here are details of the above changes, followed by descriptions of
3899 many other minor changes, and a list of fixed bugs.
3902 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
3903 called "Darwin" because that is the name of the OS core, which is the
3904 level that Valgrind works at.)
3908 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
3909 because it would require work and 10.4 is only becoming less common.
3911 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
3912 fairly well. For 10.5.x, 32-bit programs are the default even on
3913 64-bit machines, so it handles most current programs.
3915 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
3916 officially supported, but simple programs at least will probably work.
3917 However, start-up is slow.
3919 - PowerPC machines are not supported.
3921 Things that don't work:
3923 - The Ptrcheck tool.
3925 - Objective-C garbage collection.
3929 - If you have Rogue Amoeba's "Instant Hijack" program installed,
3930 Valgrind will fail with a SIGTRAP at start-up. See
3931 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
3936 - You will likely find --dsymutil=yes a useful option, as error
3937 messages may be imprecise without it.
3939 - Mac OS X support is new and therefore will be less robust than the
3940 Linux support. Please report any bugs you find.
3942 - Threaded programs may run more slowly than on Linux.
3944 Many thanks to Greg Parker for developing this port over several years.
3947 * Memcheck's leak checker has been improved.
3949 - The results for --leak-check=summary now match the summary results
3950 for --leak-check=full. Previously they could differ because
3951 --leak-check=summary counted "indirectly lost" blocks and
3952 "suppressed" blocks as "definitely lost".
3954 - Blocks that are only reachable via at least one interior-pointer,
3955 but are directly pointed to by a start-pointer, were previously
3956 marked as "still reachable". They are now correctly marked as
3959 - The default value for the --leak-resolution option has been
3960 changed from "low" to "high". In general, this means that more
3961 leak reports will be produced, but each leak report will describe
3962 fewer leaked blocks.
3964 - With --leak-check=full, "definitely lost" and "possibly lost"
3965 leaks are now considered as proper errors, ie. they are counted
3966 for the "ERROR SUMMARY" and affect the behaviour of
3967 --error-exitcode. These leaks are not counted as errors if
3968 --leak-check=summary is specified, however.
3970 - Documentation for the leak checker has been improved.
3973 * Various aspects of Valgrind's text output have changed.
3975 - Valgrind's start-up message has changed. It is shorter but also
3976 includes the command being run, which makes it easier to use
3977 --trace-children=yes. An example:
3979 - Valgrind's shut-down messages have also changed. This is most
3980 noticeable with Memcheck, where the leak summary now occurs before
3981 the error summary. This change was necessary to allow leaks to be
3982 counted as proper errors (see the description of the leak checker
3983 changes above for more details). This was also necessary to fix a
3984 longstanding bug in which uses of suppressions against leaks were
3985 not "counted", leading to difficulties in maintaining suppression
3986 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3988 - Behavior of -v has changed. In previous versions, -v printed out
3989 a mixture of marginally-user-useful information, and tool/core
3990 statistics. The statistics printing has now been moved to its own
3991 flag, --stats=yes. This means -v is less verbose and more likely
3992 to convey useful end-user information.
3994 - The format of some (non-XML) stack trace entries has changed a
3995 little. Previously there were six possible forms:
3997 0x80483BF: really (a.c:20)
3998 0x80483BF: really (in /foo/a.out)
4000 0x80483BF: (within /foo/a.out)
4001 0x80483BF: ??? (a.c:20)
4004 The third and fourth of these forms have been made more consistent
4005 with the others. The six possible forms are now:
4007 0x80483BF: really (a.c:20)
4008 0x80483BF: really (in /foo/a.out)
4009 0x80483BF: really (in ???)
4010 0x80483BF: ??? (in /foo/a.out)
4011 0x80483BF: ??? (a.c:20)
4014 Stack traces produced when --xml=yes is specified are different
4018 * Helgrind and Ptrcheck now support XML output, so they can be used
4019 from GUI tools. Also, the XML output mechanism has been
4022 - The XML format has been overhauled and generalised, so it is more
4023 suitable for error reporting tools in general. The Memcheck
4024 specific aspects of it have been removed. The new format, which
4025 is an evolution of the old format, is described in
4026 docs/internals/xml-output-protocol4.txt.
4028 - Memcheck has been updated to use the new format.
4030 - Helgrind and Ptrcheck are now able to emit output in this format.
4032 - The XML output mechanism has been overhauled. XML is now output
4033 to its own file descriptor, which means that:
4035 * Valgrind can output text and XML independently.
4037 * The longstanding problem of XML output being corrupted by
4038 unexpected un-tagged text messages is solved.
4040 As before, the destination for text output is specified using
4041 --log-file=, --log-fd= or --log-socket=.
4043 As before, XML output for a tool is enabled using --xml=yes.
4045 Because there's a new XML output channel, the XML output
4046 destination is now specified by --xml-file=, --xml-fd= or
4049 Initial feedback has shown this causes some confusion. To
4050 clarify, the two envisaged usage scenarios are:
4052 (1) Normal text output. In this case, do not specify --xml=yes
4053 nor any of --xml-file=, --xml-fd= or --xml-socket=.
4055 (2) XML output. In this case, specify --xml=yes, and one of
4056 --xml-file=, --xml-fd= or --xml-socket= to select the XML
4057 destination, one of --log-file=, --log-fd= or --log-socket=
4058 to select the destination for any remaining text messages,
4059 and, importantly, -q.
4061 -q makes Valgrind completely silent on the text channel,
4062 except in the case of critical failures, such as Valgrind
4063 itself segfaulting, or failing to read debugging information.
4064 Hence, in this scenario, it suffices to check whether or not
4065 any output appeared on the text channel. If yes, then it is
4066 likely to be a critical error which should be brought to the
4067 attention of the user. If no (the text channel produced no
4068 output) then it can be assumed that the run was successful.
4070 This allows GUIs to make the critical distinction they need to
4071 make (did the run fail or not?) without having to search or
4072 filter the text output channel in any way.
4074 It is also recommended to use --child-silent-after-fork=yes in
4078 * Improvements and changes in Helgrind:
4080 - XML output, as described above
4082 - Checks for consistent association between pthread condition
4083 variables and their associated mutexes are now performed.
4085 - pthread_spinlock functions are supported.
4087 - Modest performance improvements.
4089 - Initial (skeletal) support for describing the behaviour of
4090 non-POSIX synchronisation objects through ThreadSanitizer
4091 compatible ANNOTATE_* macros.
4093 - More controllable tradeoffs between performance and the level of
4094 detail of "previous" accesses in a race. There are now three
4097 * --history-level=full. This is the default, and was also the
4098 default in 3.4.x. It shows both stacks involved in a race, but
4099 requires a lot of memory and can be very slow in programs that
4100 do many inter-thread synchronisation events.
4102 * --history-level=none. This only shows the later stack involved
4103 in a race. This can be much faster than --history-level=full,
4104 but makes it much more difficult to find the other access
4105 involved in the race.
4107 The new intermediate setting is
4109 * --history-level=approx
4111 For the earlier (other) access, two stacks are presented. The
4112 earlier access is guaranteed to be somewhere in between the two
4113 program points denoted by those stacks. This is not as useful
4114 as showing the exact stack for the previous access (as per
4115 --history-level=full), but it is better than nothing, and it's
4116 almost as fast as --history-level=none.
4119 * New features and improvements in DRD:
4121 - The error messages printed by DRD are now easier to interpret.
4122 Instead of using two different numbers to identify each thread
4123 (Valgrind thread ID and DRD thread ID), DRD does now identify
4124 threads via a single number (the DRD thread ID). Furthermore
4125 "first observed at" information is now printed for all error
4126 messages related to synchronization objects.
4128 - Added support for named semaphores (sem_open() and sem_close()).
4130 - Race conditions between pthread_barrier_wait() and
4131 pthread_barrier_destroy() calls are now reported.
4133 - Added support for custom allocators through the macros
4134 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
4135 in <valgrind/valgrind.h>). An alternative for these two macros is
4136 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
4139 - Added support for annotating non-POSIX synchronization objects
4140 through several new ANNOTATE_*() macros.
4142 - OpenMP: added support for the OpenMP runtime (libgomp) included
4143 with gcc versions 4.3.0 and 4.4.0.
4147 - Added two new command-line options (--first-race-only and
4148 --segment-merging-interval).
4151 * Genuinely atomic support for x86/amd64/ppc atomic instructions
4153 Valgrind will now preserve (memory-access) atomicity of LOCK-
4154 prefixed x86/amd64 instructions, and any others implying a global
4155 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
4157 This means that Valgrinded processes will "play nicely" in
4158 situations where communication with other processes, or the kernel,
4159 is done through shared memory and coordinated with such atomic
4160 instructions. Prior to this change, such arrangements usually
4161 resulted in hangs, races or other synchronisation failures, because
4162 Valgrind did not honour atomicity of such instructions.
4165 * A new experimental tool, BBV, has been added. BBV generates basic
4166 block vectors for use with the SimPoint analysis tool, which allows
4167 a program's overall behaviour to be approximated by running only a
4168 fraction of it. This is useful for computer architecture
4169 researchers. You can run BBV by specifying --tool=exp-bbv (the
4170 "exp-" prefix is short for "experimental"). BBV was written by
4174 * Modestly improved support for running Windows applications under
4175 Wine. In particular, initial support for reading Windows .PDB debug
4176 information has been added.
4179 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
4180 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
4184 * The Valgrind client requests VALGRIND_PRINTF and
4185 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
4186 the string was always printed immediately on its own line. Now, the
4187 string will be added to a buffer but not printed until a newline is
4188 encountered, or other Valgrind output is printed (note that for
4189 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
4190 "other Valgrind output"). This allows you to use multiple
4191 VALGRIND_PRINTF calls to build up a single output line, and also to
4192 print multiple output lines with a single request (by embedding
4193 multiple newlines in the string).
4196 * The graphs drawn by Massif's ms_print program have changed slightly:
4198 - The half-height chars '.' and ',' are no longer drawn, because
4199 they are confusing. The --y option can be used if the default
4200 y-resolution is not high enough.
4202 - Horizontal lines are now drawn after the top of a snapshot if
4203 there is a gap until the next snapshot. This makes it clear that
4204 the memory usage has not dropped to zero between snapshots.
4207 * Something that happened in 3.4.0, but wasn't clearly announced: the
4208 option --read-var-info=yes can be used by some tools (Memcheck,
4209 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
4210 variable type and location information. This makes those tools
4211 start up more slowly and increases memory consumption, but
4212 descriptions of data addresses in error messages become more
4216 * exp-Omega, an experimental instantaneous leak-detecting tool, was
4217 disabled in 3.4.0 due to a lack of interest and maintenance,
4218 although the source code was still in the distribution. The source
4219 code has now been removed from the distribution. For anyone
4220 interested, the removal occurred in SVN revision r10247.
4223 * Some changes have been made to the build system.
4225 - VEX/ is now integrated properly into the build system. This means
4226 that dependency tracking within VEX/ now works properly, "make
4227 install" will work without requiring "make" before it, and
4228 parallel builds (ie. 'make -j') now work (previously a
4229 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
4230 was effectively ignored).
4232 - The --with-vex configure option has been removed. It was of
4233 little use and removing it simplified the build system.
4235 - The location of some install files has changed. This should not
4236 affect most users. Those who might be affected:
4238 * For people who use Valgrind with MPI programs, the installed
4239 libmpiwrap.so library has moved from
4240 $(INSTALL)/<platform>/libmpiwrap.so to
4241 $(INSTALL)/libmpiwrap-<platform>.so.
4243 * For people who distribute standalone Valgrind tools, the
4244 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
4245 have moved to $(INSTALL)/libcoregrind-<platform>.a.
4247 These changes simplify the build system.
4249 - Previously, all the distributed suppression (*.supp) files were
4250 installed. Now, only default.supp is installed. This should not
4251 affect users as the other installed suppression files were not
4252 read; the fact that they were installed was a mistake.
4255 * KNOWN LIMITATIONS:
4257 - Memcheck is unusable with the Intel compiler suite version 11.1,
4258 when it generates code for SSE2-and-above capable targets. This
4259 is because of icc's use of highly optimised inlined strlen
4260 implementations. It causes Memcheck to report huge numbers of
4261 false errors even in simple programs. Helgrind and DRD may also
4264 Versions 11.0 and earlier may be OK, but this has not been
4268 The following bugs have been fixed or resolved. Note that "n-i-bz"
4269 stands for "not in bugzilla" -- that is, a bug that was reported to us
4270 but never got a bugzilla entry. We encourage you to file bugs in
4271 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4272 mailing the developers (or mailing lists) directly -- bugs that are
4273 not entered into bugzilla tend to get forgotten about or ignored.
4275 To see details of a given bug, visit
4276 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4277 where XXXXXX is the bug number as listed below.
4279 84303 How about a LockCheck tool?
4280 91633 dereference of null ptr in vgPlain_st_basetype
4281 97452 Valgrind doesn't report any pthreads problems
4282 100628 leak-check gets assertion failure when using
4283 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4284 108528 NPTL pthread cleanup handlers not called
4285 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4286 110128 mallinfo is not implemented...
4287 110770 VEX: Generated files not always updated when making valgrind
4288 111102 Memcheck: problems with large (memory footprint) applications
4289 115673 Vex's decoder should never assert
4290 117564 False positive: Syscall param clone(child_tidptr) contains
4291 uninitialised byte(s)
4292 119404 executing ssh from inside valgrind fails
4293 133679 Callgrind does not write path names to sources with dwarf debug
4295 135847 configure.in problem with non gnu compilers (and possible fix)
4296 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4297 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4298 136230 memcheck reports "possibly lost", should be "still reachable"
4299 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4300 137904 Valgrind reports a memory leak when using POSIX threads,
4302 139076 valgrind VT_GETSTATE error
4303 142228 complaint of elf_dynamic_do_rela in trivial usage
4304 145347 spurious warning with USBDEVFS_REAPURB
4305 148441 (wine) can't find memory leak in Wine, win32 binary
4307 148742 Leak-check fails assert on exit
4308 149878 add (proper) check for calloc integer overflow
4309 150606 Call graph is broken when using callgrind control
4310 152393 leak errors produce an exit code of 0. I need some way to
4311 cause leak errors to result in a nonzero exit code.
4312 157154 documentation (leak-resolution doc speaks about num-callers
4313 def=4) + what is a loss record
4314 159501 incorrect handling of ALSA ioctls
4315 162020 Valgrinding an empty/zero-byte file crashes valgrind
4316 162482 ppc: Valgrind crashes while reading stabs information
4317 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4318 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4319 163560 VEX/test_main.c is missing from valgrind-3.3.1
4320 164353 malloc_usable_size() doesn't return a usable size
4321 165468 Inconsistent formatting in memcheck manual -- please fix
4322 169505 main.c:286 (endOfInstr):
4323 Assertion 'ii->cost_offset == *cost_offset' failed
4324 177206 Generate default.supp during compile instead of configure
4325 177209 Configure valt_load_address based on arch+os
4326 177305 eventfd / syscall 323 patch lost
4327 179731 Tests fail to build because of inlining of non-local asm labels
4328 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4329 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4330 181594 Bogus warning for empty text segment
4331 181707 dwarf doesn't require enumerations to have name
4332 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4333 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4334 Assertion '!already_present' failed.
4335 185359 exp-ptrcheck: unhandled syscall getresuid()
4336 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4337 185816 Valgrind is unable to handle debug info for files with split
4338 debug info that are prelinked afterwards
4339 185980 [darwin] unhandled syscall: sem_open
4340 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4341 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4342 186790 Suppression pattern used for leaks are not reported
4343 186796 Symbols with length>200 in suppression files are ignored
4344 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4345 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4346 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4347 188046 bashisms in the configure script
4348 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4349 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4350 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4351 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4352 assertion fail "!lock->heldBy"
4353 188427 Add support for epoll_create1 (with patch)
4354 188530 Support for SIOCGSTAMPNS
4355 188560 Include valgrind.spec in the tarball
4356 188572 Valgrind on Mac should suppress setenv() mem leak
4357 189054 Valgrind fails to build because of duplicate non-local asm labels
4358 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4359 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4360 189763 drd assertion failure: s_threadinfo[tid].is_recording
4361 190219 unhandled syscall: 328 (x86-linux)
4362 190391 dup of 181394; see above
4363 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4364 190820 No debug information on powerpc-linux
4365 191095 PATCH: Improve usbdevfs ioctl handling
4366 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4368 191189 --xml=yes should obey --gen-suppressions=all
4369 191192 syslog() needs a suppression on macosx
4370 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4371 191761 getrlimit on MacOSX
4372 191992 multiple --fn-skip only works sometimes; dependent on order
4373 192634 V. reports "aspacem sync_check_mapping_callback:
4374 segment mismatch" on Darwin
4375 192954 __extension__ missing on 2 client requests
4376 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4377 194474 "INSTALL" file has different build instructions than "README"
4378 194671 Unhandled syscall (sem_wait?) from mac valgrind
4379 195069 memcheck: reports leak (memory still reachable) for
4381 195169 drd: (vgDrd_barrier_post_wait):
4382 Assertion 'r->sg[p->post_iteration]' failed.
4383 195268 valgrind --log-file doesn't accept ~/...
4384 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4385 195860 WARNING: unhandled syscall: unix:223
4386 196528 need a error suppression for pthread_rwlock_init under os x?
4387 197227 Support aio_* syscalls on Darwin
4388 197456 valgrind should reject --suppressions=(directory)
4389 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4390 197591 unhandled syscall 27 (mincore)
4391 197793 Merge DCAS branch to the trunk == 85756, 142103
4392 197794 Avoid duplicate filenames in Vex
4393 197898 make check fails on current SVN
4394 197901 make check fails also under exp-ptrcheck in current SVN
4395 197929 Make --leak-resolution=high the default
4396 197930 Reduce spacing between leak reports
4397 197933 Print command line of client at start-up, and shorten preamble
4398 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4399 198395 add BBV to the distribution as an experimental tool
4400 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4401 198649 callgrind_annotate doesn't cumulate counters
4402 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4403 199977 Valgrind complains about an unrecognized instruction in the
4404 atomic_incs test program
4405 200029 valgrind isn't able to read Fedora 12 debuginfo
4406 200760 darwin unhandled syscall: unix:284
4407 200827 DRD doesn't work on Mac OS X
4408 200990 VG_(read_millisecond_timer)() does not work correctly
4409 201016 Valgrind does not support pthread_kill() on Mac OS
4410 201169 Document --read-var-info
4411 201323 Pre-3.5.0 performance sanity checking
4412 201384 Review user manual for the 3.5.0 release
4413 201585 mfpvr not implemented on ppc
4414 201708 tests failing because x86 direction flag is left set
4415 201757 Valgrind doesn't handle any recent sys_futex additions
4416 204377 64-bit valgrind can not start a shell script
4417 (with #!/path/to/shell) if the shell is a 32-bit executable
4418 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4419 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4420 about memory allocation events with memory access tracing enabled
4421 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4423 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4427 Release 3.4.1 (28 February 2009)
4428 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4429 3.4.1 is a bug-fix release that fixes some regressions and assertion
4430 failures in debug info reading in 3.4.0, most notably incorrect stack
4431 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4432 other debug info problems are also fixed. A number of bugs in the
4433 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4435 In view of the fact that 3.4.0 contains user-visible regressions
4436 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4437 encouraged to ship 3.4.1 in preference to 3.4.0.
4439 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4440 bugzilla" -- that is, a bug that was reported to us but never got a
4441 bugzilla entry. We encourage you to file bugs in bugzilla
4442 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4443 developers (or mailing lists) directly -- bugs that are not entered
4444 into bugzilla tend to get forgotten about or ignored.
4446 n-i-bz Fix various bugs reading icc-11 generated debug info
4447 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4448 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4449 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4450 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4451 179618 exp-ptrcheck crashed / exit prematurely
4452 179624 helgrind: false positive races with pthread_create and
4453 recv/open/close/read
4454 134207 pkg-config output contains @VG_PLATFORM@
4455 176926 floating point exception at valgrind startup with PPC 440EPX
4456 181594 Bogus warning for empty text segment
4457 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4458 181707 Dwarf3 doesn't require enumerations to have name
4459 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4460 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4461 Assertion '!already_present' failed.
4462 185359 exp-ptrcheck unhandled syscall getresuid()
4464 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4465 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4469 Release 3.4.0 (2 January 2009)
4470 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4471 3.4.0 is a feature release with many significant improvements and the
4472 usual collection of bug fixes. This release supports X86/Linux,
4473 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4474 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4476 3.4.0 brings some significant tool improvements. Memcheck can now
4477 report the origin of uninitialised values, the thread checkers
4478 Helgrind and DRD are much improved, and we have a new experimental
4479 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4480 global arrays. In detail:
4482 * Memcheck is now able to track the origin of uninitialised values.
4483 When it reports an uninitialised value error, it will try to show
4484 the origin of the value, as either a heap or stack allocation.
4485 Origin tracking is expensive and so is not enabled by default. To
4486 use it, specify --track-origins=yes. Memcheck's speed will be
4487 essentially halved, and memory usage will be significantly
4488 increased. Nevertheless it can drastically reduce the effort
4489 required to identify the root cause of uninitialised value errors,
4490 and so is often a programmer productivity win, despite running more
4493 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4494 3.4.0, will be released shortly.
4496 * Helgrind's race detection algorithm has been completely redesigned
4497 and reimplemented, to address usability and scalability concerns:
4499 - The new algorithm has a lower false-error rate: it is much less
4500 likely to report races that do not really exist.
4502 - Helgrind will display full call stacks for both accesses involved
4503 in a race. This makes it easier to identify the root causes of
4506 - Limitations on the size of program that can run have been removed.
4508 - Performance has been modestly improved, although that is very
4511 - Direct support for Qt4 threading has been added.
4513 - pthread_barriers are now directly supported.
4515 - Helgrind works well on all supported Linux targets.
4517 * The DRD thread debugging tool has seen major improvements:
4519 - Greatly improved performance and significantly reduced memory
4522 - Support for several major threading libraries (Boost.Thread, Qt4,
4523 glib, OpenMP) has been added.
4525 - Support for atomic instructions, POSIX semaphores, barriers and
4526 reader-writer locks has been added.
4528 - Works now on PowerPC CPUs too.
4530 - Added support for printing thread stack usage at thread exit time.
4532 - Added support for debugging lock contention.
4534 - Added a manual for Drd.
4536 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4537 checks for misuses of pointers. In that sense it is a bit like
4538 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4539 detect overruns of stack and global arrays, it can detect
4540 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4541 detect accesses heap blocks that have been freed a very long time
4542 ago (millions of blocks in the past).
4544 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4545 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4546 of the main Valgrind documentation. As this is an experimental
4547 tool, we would be particularly interested in hearing about your
4548 experiences with it.
4550 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4551 longer built by default, although the code remains in the repository
4552 and the tarball. This is due to three factors: a perceived lack of
4553 users, a lack of maintenance, and concerns that it may not be
4554 possible to achieve reliable operation using the existing design.
4556 * As usual, support for the latest Linux distros and toolchain
4557 components has been added. It should work well on Fedora Core 10,
4558 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4559 state) is supported, as is glibc-2.9. The C++ demangler has been
4560 updated so as to work well with C++ compiled by even the most recent
4563 * You can now use frame-level wildcards in suppressions. This was a
4564 frequently-requested enhancement. A line "..." in a suppression now
4565 matches zero or more frames. This makes it easier to write
4566 suppressions which are precise yet insensitive to changes in
4569 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4571 * Very basic support for IBM Power6 has been added (64-bit processes only).
4573 * Valgrind is now cross-compilable. For example, it is possible to
4574 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4575 on a ppc32/64-linux target.
4577 * You can set the main thread's stack size at startup using the
4578 new --main-stacksize= flag (subject of course to ulimit settings).
4579 This is useful for running apps that need a lot of stack space.
4581 * The limitation that you can't use --trace-children=yes together
4582 with --db-attach=yes has been removed.
4584 * The following bugs have been fixed. Note that "n-i-bz" stands for
4585 "not in bugzilla" -- that is, a bug that was reported to us but
4586 never got a bugzilla entry. We encourage you to file bugs in
4587 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4588 mailing the developers (or mailing lists) directly.
4590 n-i-bz Make return types for some client requests 64-bit clean
4591 n-i-bz glibc 2.9 support
4592 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4593 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4594 n-i-bz Building in an env without gdb gives bogus gdb attach
4595 92456 Tracing the origin of uninitialised memory
4596 106497 Valgrind does not demangle some C++ template symbols
4598 151612 Suppression with "..." (frame-level wildcards in .supp files)
4599 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4600 159285 unhandled syscall:25 (stime, on x86-linux)
4601 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4602 160954 ppc build of valgrind crashes with illegal instruction (isel)
4603 160956 mallinfo implementation, w/ patch
4604 162092 Valgrind fails to start gnome-system-monitor
4605 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4606 163794 assertion failure with "--track-origins=yes"
4607 163933 sigcontext.err and .trapno must be set together
4608 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4609 164476 Missing kernel module loading system calls
4610 164669 SVN regression: mmap() drops posix file locks
4611 166581 Callgrind output corruption when program forks
4612 167288 Patch file for missing system calls on Cell BE
4613 168943 unsupported scas instruction pentium
4614 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4615 172417 x86->IR: 0x82 ...
4616 172563 amd64->IR: 0xD9 0xF5 - fprem1
4617 173099 .lds linker script generation error
4618 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4619 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4621 174908 --log-file value not expanded correctly for core file
4622 175044 Add lookup_dcookie for amd64
4623 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4625 Developer-visible changes:
4627 * Valgrind's debug-info reading machinery has been majorly overhauled.
4628 It can now correctly establish the addresses for ELF data symbols,
4629 which is something that has never worked properly before now.
4631 Also, Valgrind can now read DWARF3 type and location information for
4632 stack and global variables. This makes it possible to use the
4633 framework to build tools that rely on knowing the type and locations
4634 of stack and global variables, for example exp-Ptrcheck.
4636 Reading of such information is disabled by default, because most
4637 tools don't need it, and because it is expensive in space and time.
4638 However, you can force Valgrind to read it, using the
4639 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4640 make use of such information, if present, to provide source-level
4641 descriptions of data addresses in the error messages they create.
4643 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4644 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).