1 Release 3.22.0 (?? Oct 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 * A new configure option --with-gdbscripts-dir lets you install
14 the gdb valgrind python monitor scripts in a specific location.
15 For example a distro could use it to install the scripts in a
16 safe load location --with-gdbscripts-dir=%{_datadir}/gdb/auto-load
17 It is also possible to configure --without-gdb-scripts-dir so no
18 .debug_gdb_scripts section is added to the vgpreload library and
19 no valgrind-monitor python scripts are installed at all.
21 * ================== PLATFORM CHANGES =================
23 * support has been added for FreeBSD 14 and FreeBSD 15.
26 * ==================== TOOL CHANGES ===================
29 * ==================== FIXED BUGS ====================
31 The following bugs have been fixed or resolved. Note that "n-i-bz"
32 stands for "not in bugzilla" -- that is, a bug that was reported to us
33 but never got a bugzilla entry. We encourage you to file bugs in
34 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
35 than mailing the developers (or mailing lists) directly -- bugs that
36 are not entered into bugzilla tend to get forgotten about or ignored.
38 426751 Valgrind reports «still reachable» memory using musl (alpine running inside docker)
39 460192 Add epoll_pwait2
40 469049 link failure on ppc64 (big endian) valgrind 3.20
41 469146 massif --ignore-fn does not ignore inlined functions
42 469768 Make it possible to install gdb scripts in a different location
43 470121 Can't run callgrind_control with valgrind 3.21.0 because of perl errors
44 470132 s390x: Assertion failure on VGM instruction
45 470520 Multiple realloc zero errors crash in MC_(eq_Error)
46 470713 Failure on the Yosys project: valgrind: m_libcfile.c:1802
47 (Bool vgPlain_realpath(const HChar *, HChar *)):
48 Assertion 'resolved' failed
49 470830 Don't print actions vgdb me ... continue for vgdb --multi mode
50 470978 s390x: Valgrind cannot start qemu-kvm when "sysctl vm.allocate_pgste=0"
51 471311 gdb --multi mode stdout redirecting to stderr
52 471807 Add support for lazy reading and downloading of DWARF debuginfo
53 472219 Syscall param ppoll(ufds.events) points to uninitialised byte(s)
54 472963 Broken regular expression in configure.ac
55 473604 Fix bug472219.c compile failure with Clang 16
56 473677 make check compile failure with Clang 16 based on GCC 13.x
57 473745 must-be-redirected function - strlen
58 473870 FreeBSD 14 applications fail early at startup
59 473944 Handle mold linker split RW PT_LOAD segments correctly
60 n-i-bz Allow arguments with spaces in .valgrindrc files
62 To see details of a given bug, visit
63 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
64 where XXXXXX is the bug number as listed above.
66 Release 3.21.0 (28 Apr 2023)
67 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
69 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
70 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
71 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
72 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
73 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
74 AMD64/macOS 10.13 and nanoMIPS/Linux.
76 * ==================== CORE CHANGES ===================
78 * When GDB is used to debug a program running under valgrind using
79 the valgrind gdbserver, GDB will automatically load some
80 python code provided in valgrind defining GDB front end commands
81 corresponding to the valgrind monitor commands.
82 These GDB front end commands accept the same format as
83 the monitor commands directly sent to the Valgrind gdbserver.
84 These GDB front end commands provide a better integration
85 in the GDB command line interface, so as to use for example
86 GDB auto-completion, command specific help, searching for
87 a command or command help matching a regexp, ...
88 For relevant monitor commands, GDB will evaluate arguments
89 to make the use of monitor commands easier.
90 For example, instead of having to print the address of a variable
91 to pass it to a subsequent monitor command, the GDB front end
92 command will evaluate the address argument. It is for example
94 (gdb) memcheck who_points_at &some_struct sizeof(some_struct)
97 $2 = (some_struct_type *) 0x1130a0 <some_struct>
98 (gdb) p sizeof(some_struct)
100 (gdb) monitor who_point_at 0x1130a0 40
102 * The vgdb utility now supports extended-remote protocol when
103 invoked with --multi. In this mode the GDB run command is
104 supported. Which means you don't need to run gdb and valgrind
105 from different terminals. So for example to start your program
106 in gdb and run it under valgrind you can do:
108 (gdb) set remote exec-file prog
110 (gdb) target extended-remote | vgdb --multi
113 * The behaviour of realloc with a size of zero can now
114 be changed for tools that intercept malloc. Those
115 tools are memcheck, helgrind, drd, massif and dhat.
116 Realloc implementations generally do one of two things
117 - free the memory like free() and return NULL
118 (GNU libc and ptmalloc).
119 - either free the memory and then allocate a
120 minimum sized block or just return the
121 original pointer. Return NULL if the
122 allocation of the minimum sized block fails
123 (jemalloc, musl, snmalloc, Solaris, macOS).
124 When Valgrind is configured and built it will
125 try to match the OS and libc behaviour. However
126 if you are using a non-default library to replace
127 malloc and family (e.g., musl on a glibc Linux or
128 tcmalloc on FreeBSD) then you can use a command line
129 option to change the behaviour of Valgrind:
130 --realloc-zero-bytes-frees=yes|no [yes on Linux glibc, no otherwise]
132 * ================== PLATFORM CHANGES =================
134 * Make the address space limit on FreeBSD amd64 128Gbytes
135 (the same as Linux and Solaris, it was 32Gbytes)
137 * ==================== TOOL CHANGES ===================
140 - When doing a delta leak_search, it is now possible to only
141 output the new loss records compared to the previous leak search.
142 This is available in the memcheck monitor command 'leak_search'
143 by specifying the "new" keyword or in your program by using
144 the client request VALGRIND_DO_NEW_LEAK_CHECK.
145 Whenever a "delta" leak search is done (i.e. when specifying
146 "new" or "increased" or "changed" in the monitor command),
147 the new loss records have a "new" marker.
148 - Valgrind now contains python code that defines GDB memcheck
149 front end monitor commands. See CORE CHANGES.
150 - Performs checks for the use of realloc with a size of zero.
151 This is non-portable and a source of errors. If memcheck
152 detects such a usage it will generate an error
153 realloc() with size 0
154 followed by the usual callstacks.
155 A switch has been added to allow this to be turned off:
156 --show-realloc-size-zero=yes|no [yes]
159 - The option ---history-backtrace-size=<number> allows to configure
160 the number of entries to record in the stack traces of "old"
161 accesses. Previously, this number was hardcoded to 8.
162 - Valgrind now contains python code that defines GDB helgrind
163 front end monitor commands. See CORE CHANGES.
166 - `--cache-sim=no` is now the default. The cache simulation is old and
167 unlikely to match any real modern machine. This means only the `Ir`
168 event are gathered by default, but that is by far the most useful
170 - `cg_annotate`, `cg_diff`, and `cg_merge` have been rewritten in
171 Python. As a result, they all have more flexible command line
172 argument handling, e.g. supporting `--show-percs` and
173 `--no-show-percs` forms as well as the existing `--show-percs=yes`
174 and `--show-percs=no`.
175 - `cg_annotate` has some functional changes.
176 - It's much faster, e.g. 3-4x on common cases.
177 - It now supports diffing (with `--diff`, `--mod-filename`, and
178 `--mod-funcname`) and merging (by passing multiple data files).
179 - It now provides more information at the file and function level.
180 There are now "File:function" and "Function:file" sections. These
181 are very useful for programs that use inlining a lot.
182 - Support for user-annotated files and the `-I`/`--include` option
183 has been removed, because it was of little use and blocked other
185 - The `--auto` option is renamed `--annotate`, though the old
186 `--auto=yes`/`--auto=no` forms are still supported.
187 - `cg_diff` and `cg_merge` are now deprecated, because `cg_annotate`
188 now does a better job of diffing and merging.
189 - The Cachegrind output file format has changed very slightly, but in
190 ways nobody is likely to notice.
193 - Valgrind now contains python code that defines GDB callgrind
194 front end monitor commands. See CORE CHANGES.
197 - Valgrind now contains python code that defines GDB massif
198 front end monitor commands. See CORE CHANGES.
201 - A new kind of user request has been added which allows you to
202 override the 1024 byte limit on access count histograms for blocks
203 of memory. The client request is DHAT_HISTOGRAM_MEMORY.
205 * ==================== FIXED BUGS ====================
207 The following bugs have been fixed or resolved. Note that "n-i-bz"
208 stands for "not in bugzilla" -- that is, a bug that was reported to us
209 but never got a bugzilla entry. We encourage you to file bugs in
210 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
211 than mailing the developers (or mailing lists) directly -- bugs that
212 are not entered into bugzilla tend to get forgotten about or ignored.
214 170510 Don't warn about ioctl of size 0 without direction hint
215 241072 List tools in --help output
216 327548 false positive while destroying mutex
217 382034 Testcases build fixes for musl
218 351857 confusing error message about valid command line option
219 374596 inconsistent RDTSCP support on x86_64
220 392331 Spurious lock not held error from inside pthread_cond_timedwait
221 397083 Likely false positive "uninitialised value(s)" for __wmemchr_avx2 and __wmemcmp_avx2_movbe
222 400793 pthread_rwlock_timedwrlock false positive
223 419054 Unhandled syscall getcpu on arm32
224 433873 openat2 syscall unimplemented on Linux
225 434057 Add stdio mode to valgrind's gdbserver
226 435441 valgrind fails to interpose malloc on musl 1.2.2 due to weak symbol name and no libc soname
227 436413 Warn about realloc of size zero
228 439685 compiler warning in callgrind/main.c
229 444110 priv/guest_ppc_toIR.c:36198:31: warning: duplicated 'if' condition.
230 444487 hginfo test detects an extra lock inside data symbol "_rtld_local"
231 444488 Use glibc.pthread.stack_cache_size tunable
232 444568 drd/tests/pth_barrier_thr_cr fails on Fedora 38
233 445743 "The impossible happened: mutex is locked simultaneously by two threads"
234 while using mutexes with priority inheritance and signals
235 449309 Missing loopback device ioctl(s)
236 459476 vgdb: allow address reuse to avoid "address already in use" errorsuse" errors
237 460356 s390: Sqrt32Fx4 -- cannot reduce tree
238 462830 WARNING: unhandled amd64-freebsd syscall: 474
239 463027 broken check for MPX instruction support in assembler
240 464103 Enhancement: add a client request to DHAT to mark memory to be histogrammed
241 464476 Firefox fails to start under Valgrind
242 464609 Valgrind memcheck should support Linux pidfd_open
243 464680 Show issues caused by memory policies like selinux deny_execmem
244 464859 Build failures with GCC-13 (drd tsan_unittest)
245 464969 D language demangling
246 465435 m_libcfile.c:66 (vgPlain_safe_fd): Assertion 'newfd >= VG_(fd_hard_limit)' failed.
247 466104 aligned_alloc problems, part 1
248 467036 Add time cost statistics for Regtest
249 467482 Build failure on aarch64 Alpine
250 467714 fdleak_* and rlimit tests fail when parent process has more than
251 64 descriptors opened
252 467839 Gdbserver: Improve compatibility of library directory name
253 468401 [PATCH] Add a style file for clang-format
254 468556 Build failure for vgdb
255 468606 build: remove "Valgrind relies on GCC" check/output
256 469097 ppc64(be) doesn't support SCV syscall instruction
257 n-i-bz FreeBSD rfork syscall fail with EINVAL or ENOSYS rather than VG_(unimplemented)
259 To see details of a given bug, visit
260 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
261 where XXXXXX is the bug number as listed above.
263 * ==================== KNOWN ISSUES ===================
265 * configure --enable-lto=yes is know to not work in all setups.
266 See bug 469049. Workaround: Build without LTO.
268 (3.21.0.RC1: 14 Apr 2023)
269 (3.21.0.RC2: 21 Apr 2023)
271 Release 3.20.0 (24 Oct 2022)
272 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
274 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
275 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
276 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
277 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
278 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
279 AMD64/macOS 10.13 and nanoMIPS/Linux.
281 * ==================== CORE CHANGES ===================
283 * The option "--vgdb-stop-at=event1,event2,..." accepts the new value abexit.
284 This indicates to invoke gdbserver when your program exits abnormally
285 (i.e. with a non zero exit code).
286 * Fix Rust v0 name demangling.
287 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
288 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
289 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with
291 * The option --enable-debuginfod=<no|yes> [default: yes] has been added on
293 * More DWARF5 support as generated by clang14.
295 * ==================== FIXED BUGS ====================
297 The following bugs have been fixed or resolved. Note that "n-i-bz"
298 stands for "not in bugzilla" -- that is, a bug that was reported to us
299 but never got a bugzilla entry. We encourage you to file bugs in
300 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
301 than mailing the developers (or mailing lists) directly -- bugs that
302 are not entered into bugzilla tend to get forgotten about or ignored.
304 131186 writev reports error in (vector[...])
305 434764 iconv_open causes ld.so v2.28+ to use optimised strncmp
306 446754 Improve error codes from alloc functions under memcheck
307 452274 memcheck crashes with Assertion 'sci->status.what == SsIdle' failed
308 452779 Valgrind fails to build on FreeBSD 13.0 with llvm-devel (15.0.0)
309 453055 shared_timed_mutex drd test fails with "Lock shared failed" message
310 453602 Missing command line option to enable/disable debuginfod
311 452802 Handle lld 9+ split RW PT_LOAD segments correctly
312 454040 s390x: False-positive memcheck:cond in memmem on arch13 systems
313 456171 [PATCH] FreeBSD: Don't record address errors when accessing the 'kern.ps_strings' sysctl struct
314 n-i-bz Implement vgdb invoker on FreeBSD
315 458845 PowerPC: The L field for the dcbf and sync instruction should be
317 458915 Remove register cache to fix 458915 gdbserver causes wrong syscall return
318 459031 Documentation on --error-exitcode incomplete
319 459477 XERROR messages lacks ending '\n' in vgdb
320 462007 Implicit int in none/tests/faultstatus.c
322 To see details of a given bug, visit
323 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
324 where XXXXXX is the bug number as listed above.
326 (3.20.0.RC1: 20 Oct 2022)
329 Release 3.19.0 (11 Apr 2022)
330 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
332 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
333 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
334 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
335 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
336 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
337 AMD64/macOS 10.13 and nanoMIPS/Linux.
339 * ==================== CORE CHANGES ===================
341 * Fix Rust v0 name demangling.
342 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
343 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
344 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with HardenedBSD
346 * ================== PLATFORM CHANGES =================
349 - ignore the "v8.x" architecture levels, only look at actual CPU features
350 present. Fixes mismatch detected between RDMA and atomics features
351 preventing startup on some QEMU configurations.
352 - Implement LD{,A}XP and ST{,L}XP
353 - Fix incorrect code emitted for doubleword CAS.
356 - Fix sys_ipc semtimedop syscall
357 - Fix VFLRX and WFLRX instructions
358 - Fix EXRL instruction with negative offset
361 - Reimplement the vbpermq instruction support to generate less Iops and
362 avoid overflowing internal buffers.
363 - Fix checking for scv support to avoid "Facility 'SCV' unavailable (12),
364 exception" messages in dmsg.
365 - Fix setting condition code for Vector Compare quad word instructions.
366 - Fix fix lxsibzx, lxsihzx and lxsihzx instructions so they only load
367 their respective sized data.
368 - Fix the prefixed stq instruction in PC relative mode.
370 * ==================== TOOL CHANGES ===================
373 - Speed up --track-origins=yes for large (in the range of hundreds to
374 thousands of megabytes) mmap/munmaps.
376 - Several fixes for new versions of libstd++ using new posix try_lock
379 * ==================== FIXED BUGS ====================
381 The following bugs have been fixed or resolved. Note that "n-i-bz"
382 stands for "not in bugzilla" -- that is, a bug that was reported to us
383 but never got a bugzilla entry. We encourage you to file bugs in
384 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
385 than mailing the developers (or mailing lists) directly -- bugs that
386 are not entered into bugzilla tend to get forgotten about or ignored.
388 403802 leak_cpp_interior fails with some reachable blocks different than expected
389 435732 memcheck/tests/leak_cpp_interior fails with gcc11
390 444242 s390x: Valgrind crashes on EXRL with negative offset
391 444399 arm64: unhandled instruction 0xC87F2D89 (LD{,A}XP and ST{,L}XP).
393 444481 gdb_server test failures on s390x
394 444495 dhat/tests/copy fails on s390x
395 444552 memcheck/tests/sem fails on s390x with glibc 2.34
396 444571 PPC, fix the lxsibzx and lxsihzx so they only load their respective
398 444836 PPC, pstq instruction for R=1 is not storing to the correct address.
399 444925 fexecve syscall wrapper not properly implemented
400 445032 valgrind/memcheck crash with SIGSEGV when SIGVTALRM timer used and
402 445211 Fix out of tree builds
403 445300 [PATCH] Fix building tests with Musl
404 445011 SIGCHLD is sent when valgrind uses debuginfod-find
405 445354 arm64 backend: incorrect code emitted for doubleword CAS
406 445415 arm64 front end: alignment checks missing for atomic instructions
407 445504 Using C++ condition_variable results in bogus "mutex is locked simultaneously by two threads" warning
408 445607 Unhandled amd64-freebsd syscall: 247
409 445668 Inline stack frame generation is broken for Rust binaries
410 445916 Demangle Rust v0 symbols with .llvm suffix
411 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false positives
412 446138 DRD/Helgrind with std::timed_mutex::try_lock_until false positives
413 446281 Add a DRD suppression for fwrite
414 446103 Memcheck: `--track-origins=yes` causes extreme slowdowns for large mmap/munmap
415 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false
416 446251 TARGET_SIGNAL_THR added to enum target_signal
417 446823 FreeBSD - missing syscalls when using libzm4
418 447991 s390x: Valgrind indicates illegal instruction on wflrx
419 447995 Valgrind segfault on power10 due to hwcap checking code
420 449483 Powerpc: vcmpgtsq., vcmpgtuq,, vcmpequq. instructions not setting the
421 condition code correctly.
422 449672 ppc64 --track-origins=yes failures because of bad cmov addHRegUse
423 449838 sigsegv liburing the 'impossible' happened for io_uring_setup
424 450025 Powerc: ACC file not implemented as a logical overlay of the VSR
426 450437 Warn for execve syscall with argv or argv[0] being NULL
427 450536 Powerpc: valgrind throws 'facility scv unavailable exception'
428 451626 Syscall param bpf(attr->raw_tracepoint.name) points to unaddressable byte(s)
429 451827 [ppc64le] VEX temporary storage exhausted with several vbpermq instructions
430 451843 valgrind fails to start on a FreeBSD system which enforces W^X
432 To see details of a given bug, visit
433 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
434 where XXXXXX is the bug number as listed above.
436 (3.19.0.RC1: 02 Apr 2022)
437 (3.19.0.RC2: 08 Apr 2022)
440 Release 3.18.0 (15 Oct 2021)
441 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
443 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
444 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
445 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
446 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
447 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
448 AMD64/macOS 10.13 and nanoMIPS/Linux.
450 * ==================== CORE CHANGES ===================
452 * The libiberty demangler has been updated, which brings support for
453 Rust v0 name demangling. [Update: alas, due to a bug, this support
454 isn't working in 3.18.0.]
456 * __libc_freeres isn't called anymore after the program recieves a
457 fatal signal. Causing some internal glibc resources to hang around,
458 but preventing any crashes after the program has ended.
460 * The DWARF reader is now very much faster at startup when just
461 --read-inline-info=yes (the default in most cases) is given.
463 * glibc 2.34, which moved various functions from libpthread.so into
464 libc.so, is now supported.
466 * ================== PLATFORM CHANGES =================
470 - v8.2 scalar and vector FABD, FACGE, FACGT and FADD.
471 - v8.2 FP compare & conditional compare instructions.
472 - Zero variants of v8.2 FP compare instructions.
476 - Support the miscellaneous-instruction-extensions facility 3 and
477 the vector-enhancements facility 2. This enables programs
478 compiled with "-march=arch13" or "-march=z15" to be executed
483 - ISA 3.1 support is now complete
484 - ISA 3.0 support for the darn instruction added.
485 - ISA 3.0 support for the vector system call instruction scv added.
486 - ISA 3.0 support for the copy, paste and cpabort instructions added.
488 * Support for X86/FreeBSD and AMD64/FreeBSD has been added.
490 * ==================== OTHER CHANGES ====================
492 * Memcheck on amd64: minor fixes to remove some false positive
495 * ==================== FIXED BUGS ====================
497 The following bugs have been fixed or resolved. Note that "n-i-bz"
498 stands for "not in bugzilla" -- that is, a bug that was reported to us
499 but never got a bugzilla entry. We encourage you to file bugs in
500 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
501 than mailing the developers (or mailing lists) directly -- bugs that
502 are not entered into bugzilla tend to get forgotten about or ignored.
504 208531 [PATCH]: FreeBSD support for valgrind
505 368960 WARNING: unhandled amd64-linux syscall: 163 (acct)
506 407589 [Linux] Add support for C11 aligned_alloc() and GNU reallocarray()
507 423963 Error in child thread when CLONE_PIDFD is used
508 426148 crash with "impossible happened" when running BPF CO-RE programs
509 429375 PPC ISA 3.1 support is missing, part 9
510 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
511 431306 Update demangler to support Rust v0 name mangling
512 432387 s390x: z15 instructions support
513 433437 FreeBSD support, part 1
514 433438 FreeBSD support, part 2
515 433439 FreeBSD support, part 3
516 433469 FreeBSD support, part 4
517 433473 FreeBSD support, part 5
518 433477 FreeBSD support, part 6
519 433479 FreeBSD support, part 7
520 433504 FreeBSD support, part 8
521 433506 FreeBSD support, part 9
522 433507 FreeBSD support, part 10
523 433508 FreeBSD support, part 11
524 433510 FreeBSD support, part 12
525 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
526 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
527 434296 s390x: False-positive memcheck diagnostics from vector string
529 434840 PPC64 darn instruction not supported
530 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
531 435908 valgrind tries to fetch from deubginfod for files which already
532 have debug information
533 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2
534 439046 valgrind is unusably large when linked with lld
535 439090 Implement close_range(2)
536 439326 Valgrind 3.17.0 won't compile with Intel 2021 oneAPI compilers
537 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
538 440670 unhandled ppc64le-linux syscall: 252 statfs64 and 253 fstatfs64
539 440906 Fix impossible constraint issue in P10 testcase.
540 441512 Remove a unneeded / unnecessary prefix check.
541 441534 Update the expected output for test_isa_3_1_VRT.
542 442061 very slow execution under Fedora 34 (readdwarf3)
543 443031 Gcc -many change requires explicit .machine directives
544 443033 Add support for the ISA 3.0 mcrxrx instruction
545 443034 Sraw, srawi, srad, sradi, mfs
546 443178 Powerpc, test jm-mfspr expected output needs to be updated.
547 443179 Need new test for the lxvx and stxvx instructions on ISA 2.07 and
549 443180 The subnormal test and the ISA 3.0 test generate compiler warnings
550 443314 In the latest GIT version, Valgrind with "--trace-flags" crashes
552 443605 Don't call final_tidyup (__libc_freeres) on FatalSignal
554 To see details of a given bug, visit
555 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
556 where XXXXXX is the bug number as listed below.
558 (3.18.0.RC1: 12 Oct 2021)
559 (3.18.0: 15 Oct 2021)
563 Release 3.17.0 (19 Mar 2021)
564 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
566 3.17.0 fixes a number of bugs and adds some functional changes: support for
567 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
568 some new instructions for Arm64, S390 and POWER. There are also some tool
571 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
572 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
573 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
574 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
575 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
577 * ==================== CORE CHANGES ===================
579 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
582 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
583 debugging information. When a debuginfo file cannot be found locally,
584 Valgrind is able to query debuginfod servers for the file using its
585 build-id. See the user manual for more information about debuginfod support.
587 * ================== PLATFORM CHANGES =================
591 - Inaccuracies resulting from double-rounding in the simulation of
592 floating-point multiply-add/subtract instructions have been fixed. These
593 should now behave exactly as the hardware does.
595 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
596 ongoing. Support for the half-word variants of at least the following
597 instructions has been added:
599 FABS <Vd>.<T>, <Vn>.<T>
601 FNEG <Vd>.<T>, <Vn>.<T>
603 FSQRT <Vd>.<T>, <Vn>.<T>
608 - Implement the new instructions/features that were added to z/Architecture
609 with the vector-enhancements facility 1. Also cover the instructions from
610 the vector-packed-decimal facility that are defined outside the chapter
611 "Vector Decimal Instructions", but not the ones from that chapter itself.
613 For a detailed list of newly supported instructions see the updates to
614 `docs/internals/s390-opcodes.csv'.
616 Since the miscellaneous instruction extensions facility 2 was already
617 added in Valgrind 3.16.0, this completes the support necessary to run
618 general programs built with `--march=z14' under Valgrind. The
619 vector-packed-decimal facility is currently not exploited by the standard
620 toolchain and libraries.
624 - Various bug fixes. Fix for the sync field to limit setting just two of
625 the two bits in the L-field. Fix the write size for the stxsibx and
626 stxsihx instructions. Fix the modsw and modsd instructions.
628 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
629 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
630 conversion instructions are still missing.
632 * ==================== TOOL CHANGES ====================
634 * General tool changes
636 - All the tools and their vgpreload libraries are now installed under
637 libexec because they cannot be executed directly and should be run through
638 the valgrind executable. This should be an internal, not user visible,
639 change, but might impact valgrind packagers.
641 - The --track-fds option now respects -q, --quiet and won't output anything
642 if no file descriptors are leaked. It also won't report the standard stdin
643 (0), stdout (1) or stderr (2) descriptors as being leaked with
644 --trace-fds=yes anymore. To track whether the standard file descriptors
645 are still open at the end of the program run use --trace-fds=all.
649 - DHAT has been extended, with two new modes of operation. The new
650 --mode=copy flag triggers copy profiling, which records calls to memcpy,
651 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
652 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
653 the new dhat/dhat.h file. This is useful for learning more about hot code
654 paths. See the user manual for more information about the new modes.
656 - Because of these changes, DHAT's file format has changed. DHAT output
657 files produced with earlier versions of DHAT will not work with this
658 version of DHAT's viewer, and DHAT output files produced with this version
659 of DHAT will not work with earlier versions of DHAT's viewer.
661 * ==================== FIXED BUGS ====================
663 The following bugs have been fixed or resolved. Note that "n-i-bz"
664 stands for "not in bugzilla" -- that is, a bug that was reported to us
665 but never got a bugzilla entry. We encourage you to file bugs in
666 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
667 than mailing the developers (or mailing lists) directly -- bugs that
668 are not entered into bugzilla tend to get forgotten about or ignored.
670 To see details of a given bug, visit
671 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
672 where XXXXXX is the bug number as listed below.
674 140178 open("/proc/self/exe", ...); doesn't quite work
675 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
676 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
677 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
678 345077 linux syscall execveat support (linux 3.19)
679 361770 Missing F_ADD_SEALS
680 369029 handle linux syscalls sched_getattr and sched_setattr
681 384729 __libc_freeres inhibits cross-platform valgrind
682 388787 Support for C++17 new/delete
683 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
684 396656 Warnings while reading debug info
685 397605 ioctl FICLONE mishandled
686 401416 Compile failure with openmpi 4.0
687 408663 Suppression file for musl libc
688 404076 s390x: z14 vector instructions not implemented
689 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
690 (actually affected all x86 and nanomips regardless of host bitness)
691 413547 regression test does not check for Arm 64 features.
692 414268 Enable AArch64 feature detection and decoding for v8.x instructions
693 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
694 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
695 422261 platform selection fails for unqualified client name
696 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
697 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
698 423195 PPC ISA 3.1 support is missing, part 1
699 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
700 424012 crash with readv/writev having invalid but not NULL arg2 iovec
701 424298 amd64: Implement RDSEED
702 425232 PPC ISA 3.1 support is missing, part 2
703 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
704 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
705 426123 PPC ISA 3.1 support is missing, part 3
706 426144 Fix "condition variable has not been initialized" on Fedora 33.
707 427400 PPC ISA 3.1 support is missing, part 4
708 427401 PPC ISA 3.1 support is missing, part 5
709 427404 PPC ISA 3.1 support is missing, part 6
710 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
711 427787 Support new faccessat2 linux syscall (439)
712 427969 debuginfo section duplicates a section in the main ELF file
713 428035 drd: Unbreak the musl build
714 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
715 428716 cppcheck detects potential leak in VEX/useful/smchash.c
716 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
717 429352 PPC ISA 3.1 support is missing, part 7
718 429354 PPC ISA 3.1 support is missing, part 8
719 429692 unhandled ppc64le-linux syscall: 147 (getsid)
720 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
722 429952 Errors when building regtest with clang
723 430354 ppc stxsibx and stxsihx instructions write too much data
724 430429 valgrind.h doesn't compile on s390x with clang
725 430485 expr_is_guardable doesn't handle Iex_Qop
726 431556 Complete arm64 FADDP v8.2 instruction support
727 432102 Add support for DWARF5 as produced by GCC11
728 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
729 432381 drd: Process STACK_REGISTER client requests
730 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
731 432672 vg_regtest: test-specific environment variables not reset between tests
732 432809 VEX should support REX.W + POPF
733 432861 PPC modsw and modsd give incorrect results for 1 mod 12
734 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
735 432215 Add debuginfod functionality
736 433323 Use pkglibexecdir as vglibdir
737 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
738 433629 valgrind/README has type "abd" instead of "and"
739 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
740 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
741 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
742 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
743 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
745 (3.17.0.RC1: 13 Mar 2021)
746 (3.17.0.RC2: 17 Mar 2021)
747 (3.17.0: 19 Mar 2021)
751 Release 3.16.1 (22 June 2020)
752 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
754 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
755 fixes character encoding problems in the documentation HTML.
757 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
758 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
760 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
764 Release 3.16.0 (27 May 2020)
765 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
767 3.16.0 is a feature release with many improvements and the usual collection of
770 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
771 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
772 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
773 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
774 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
776 * ==================== CORE CHANGES ===================
778 * It is now possible to dynamically change the value of many command line
779 options while your program (or its children) are running under Valgrind.
781 To see the list of dynamically changeable options, run
782 "valgrind --help-dyn-options".
784 You can change the options from the shell by using vgdb to launch
785 the monitor command "v.clo <clo option>...".
786 The same monitor command can be used from a gdb connected
787 to the valgrind gdbserver.
788 Your program can also change the dynamically changeable options using
789 the client request VALGRIND_CLO_CHANGE(option).
791 * ================== PLATFORM CHANGES =================
793 * MIPS: preliminary support for nanoMIPS instruction set has been added.
795 * ==================== TOOL CHANGES ====================
799 - The implicit memcpy done by each call to realloc now counts towards the
800 read and write counts of resized heap blocks, making those counts higher
805 - cg_annotate's --auto and --show-percs options now default to 'yes', because
806 they are usually wanted.
810 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
811 because they are usually wanted.
813 - The command option --collect-systime has been enhanced to specify
814 the unit used to record the elapsed time spent during system calls.
815 The command option now accepts the values no|yes|msec|usec|nsec,
816 where yes is a synonym of msec. When giving the value nsec, the
817 system cpu time of system calls is also recorded.
821 - Several memcheck options are now dynamically changeable.
822 Use valgrind --help-dyn-options to list them.
824 - The release 3.15 introduced a backward incompatible change for
825 some suppression entries related to preadv and pwritev syscalls.
826 When reading a suppression entry using the unsupported 3.14 format,
827 valgrind will now produce a warning to say the suppression entry will not
828 work, and suggest the needed change.
830 - Significantly fewer false positive errors on optimised code generated by
831 Clang and GCC. In particular, Memcheck now deals better with the
832 situation where the compiler will transform C-level "A && B" into "B && A"
833 under certain circumstances (in which the transformation is valid).
834 Handling of integer equality/non-equality checks on partially defined
835 values is also improved on some architectures.
839 - The exprimental Stack and Global Array Checking tool has been removed.
840 It only ever worked on x86 and amd64, and even on those it had a
841 high false positive rate and was slow. An alternative for detecting
842 stack and global array overruns is using the AddressSanitizer (ASAN)
843 facility of the GCC and Clang compilers, which require you to rebuild
844 your code with -fsanitize=address.
846 * ==================== OTHER CHANGES ====================
848 * New and modified GDB server monitor features:
850 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
852 - The gdbserver monitor commands that require an address and an optional
853 length argument now accepts the alternate 'C like' syntax "address[length]".
854 For example, the memcheck command "monitor who_points_at 0x12345678 120"
855 can now also be given as "monitor who_points_at 0x12345678[120]".
857 * ==================== FIXED BUGS ====================
859 The following bugs have been fixed or resolved. Note that "n-i-bz"
860 stands for "not in bugzilla" -- that is, a bug that was reported to us
861 but never got a bugzilla entry. We encourage you to file bugs in
862 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
863 than mailing the developers (or mailing lists) directly -- bugs that
864 are not entered into bugzilla tend to get forgotten about or ignored.
866 To see details of a given bug, visit
867 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
868 where XXXXXX is the bug number as listed below.
870 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
871 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
872 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
873 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
874 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
875 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
876 400872 Add nanoMIPS support to Valgrind
877 403212 drd/tests/trylock hangs on FreeBSD
878 404406 s390x: z14 miscellaneous instructions not implemented
879 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
880 406561 mcinfcallWSRU gdbserver_test fails on ppc64
881 406824 Unsupported baseline
882 407218 Add support for the copy_file_range syscall
883 407307 Intercept stpcpy also in ld.so for arm64
884 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
886 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
887 408009 Expose rdrand and f16c even on avx if host cpu supports them
888 408091 Missing pkey syscalls
889 408414 Add support for missing for preadv2 and pwritev2 syscalls
890 409141 Valgrind hangs when SIGKILLed
891 409206 Support for Linux PPS and PTP ioctls
892 409367 exit_group() after signal to thread waiting in futex() causes hangs
893 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
894 409780 References to non-existent configure.in
895 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
896 410599 Non-deterministic behaviour of pth_self_kill_15_other test
897 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
898 411134 Allow the user to change a set of command line options during execution
899 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
900 412344 Problem setting mips flags with specific paths
901 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
902 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
903 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
904 413603 callgrind_annotate/cg_annotate truncate function names at '#'
905 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
906 415136 ARMv8.1 Compare-and-Swap instructions are not supported
907 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
908 416239 valgrind crashes when handling clock_adjtime
909 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
910 416286 DRD reports "conflicting load" error on std::mutex::lock()
911 416301 s390x: "compare and signal" not supported
912 416387 finit_module and bpf syscalls are unhandled on arm64
913 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
914 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
915 416753 new 32bit time syscalls for 2038+
916 417075 pwritev(vector[...]) suppression ignored
917 417075 is not fixed, but incompatible supp entries are detected
918 and a warning is produced for these.
919 417187 [MIPS] Conditional branch problem since 'grail' changes
920 417238 Test memcheck/tests/vbit-test fails on mips64 BE
921 417266 Make memcheck/tests/linux/sigqueue usable with musl
922 417281 s390x: /bin/true segfaults with "grail" enabled
923 417427 commit to fix vki_siginfo_t definition created numerous regression
925 417452 s390_insn_store_emit: dst->tag for HRcVec128
926 417578 Add suppressions for glibc DTV leaks
927 417906 clone with CLONE_VFORK and no CLONE_VM fails
928 418004 Grail code additions break ppc64.
929 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
930 418997 s390x: Support Iex_ITE for float and vector types
931 419503 s390x: Avoid modifying registers returned from isel functions
932 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
933 421570 std_mutex fails on Arm v8.1 h/w
934 434035 vgdb might crash if valgrind is killed
935 n-i-bz Fix minor one time leaks in dhat.
936 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
937 n-i-bz Add support for the Linux io_uring system calls
938 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
939 n-i-bz Fix non-glibc build of test suite with s390x_features
940 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
941 423195 PPC ISA 3.1 support is missing, part 1
943 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
944 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
945 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
949 Release 3.15.0 (12 April 2019)
950 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
952 3.15.0 is a feature release with many improvements and the usual collection of
955 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
956 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
957 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
958 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
959 support for X86/macOS 10.13 and AMD64/macOS 10.13.
961 * ==================== CORE CHANGES ===================
963 * The XTree Massif output format now makes use of the information obtained
964 when specifying --read-inline-info=yes.
966 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
968 * ==================== TOOL CHANGES ====================
972 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
973 it has been promoted from an experimental tool to a regular tool. Run it
974 with --tool=dhat instead of --tool=exp-dhat.
976 - DHAT now prints only minimal data when the program ends, instead writing
977 the bulk of the profiling data to a file. As a result, the --show-top-n
978 and --sort-by options have been removed.
980 - Profile results can be viewed with the new viewer, dh_view.html. When
981 a run ends, a short message is printed, explaining how to view the result.
983 - See the documentation for more details.
987 - cg_annotate has a new option, --show-percs, which prints percentages next
992 - callgrind_annotate has a new option, --show-percs, which prints percentages
993 next to all event counts.
995 - callgrind_annotate now inserts commas in call counts, and
996 sort the caller/callee lists in the call tree.
1000 - The default value for --read-inline-info is now "yes" on
1001 Linux/Android/Solaris. It is still "no" on other OS.
1005 - The option --xtree-leak=yes (to output leak result in xtree format)
1006 automatically activates the option --show-leak-kinds=all, as xtree
1007 visualisation tools such as kcachegrind can in any case select what kind
1008 of leak to visualise.
1010 - There has been further work to avoid false positives. In particular,
1011 integer equality on partially defined inputs (C == and !=) is now handled
1014 * ==================== OTHER CHANGES ====================
1016 * The new option --show-error-list=no|yes displays, at the end of the run, the
1017 list of detected errors and the used suppressions. Prior to this change,
1018 showing this information could only be done by specifying "-v -v", but that
1019 also produced a lot of other possibly-non-useful messages. The option -s is
1020 equivalent to --show-error-list=yes.
1022 * ==================== FIXED BUGS ====================
1024 The following bugs have been fixed or resolved. Note that "n-i-bz"
1025 stands for "not in bugzilla" -- that is, a bug that was reported to us
1026 but never got a bugzilla entry. We encourage you to file bugs in
1027 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1028 than mailing the developers (or mailing lists) directly -- bugs that
1029 are not entered into bugzilla tend to get forgotten about or ignored.
1031 To see details of a given bug, visit
1032 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1033 where XXXXXX is the bug number as listed below.
1035 385411 s390x: z13 vector floating-point instructions not implemented
1036 397187 z13 vector register support for vgdb gdbserver
1037 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
1038 398870 Please add support for instruction vcvtps2ph
1039 399287 amd64 front end: Illegal Instruction vcmptrueps
1040 399301 Use inlined frames in Massif XTree output.
1041 399322 Improve callgrind_annotate output
1042 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
1043 400164 helgrind test encounters mips x-compiler warnings and assembler error
1044 400490 s390x: VRs allocated as if separate from FPRs
1045 400491 s390x: Operand of LOCH treated as unsigned integer
1046 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
1047 options, which specify a mips64 processor
1048 401112 LLVM 5.0 generates comparison against partially initialized data
1049 401277 More bugs in z13 support
1050 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
1051 401578 drd: crashes sometimes on fork()
1052 401627 memcheck errors with glibc avx2 optimized wcsncmp
1053 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
1054 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
1055 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
1057 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
1058 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
1059 402123 invalid assembler opcodes for mips32r2
1060 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
1061 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
1062 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
1063 402351 mips64 libvexmultiarch_test fails on s390x
1064 402369 Overhaul DHAT
1065 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
1066 402480 Do not use %rsp in clobber list
1067 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
1068 402515 Implement new option --show-error-list=no|yes / -s
1069 402519 POWER 3.0 addex instruction incorrectly implemented
1070 402781 Redo the cache used to process indirect branch targets
1071 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
1072 403552 s390x: wrong facility bit checked for vector facility
1073 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
1074 404638 Add VG_(replaceIndexXA)
1075 404843 s390x: backtrace sometimes ends prematurely
1076 404888 autotools cleanup series
1077 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
1078 405182 Valgrind fails to build with Clang
1079 405205 filter_libc: remove the line holding the futex syscall error entirely
1080 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
1081 the upper and lower 32-bits of the 64-bit result
1082 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
1083 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
1084 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
1086 405403 s390x disassembler cannot be used on x86
1087 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
1088 405458 MIPS mkFormVEC arguments swapped?
1089 405716 drd: Fix an integer overflow in the stack margin calculation
1090 405722 Support arm64 core dump
1091 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
1092 of the 64-bit destination field.
1093 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
1094 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
1095 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
1097 406256 PPC64, vector floating point instructions don't handle subnormal
1098 according to VSCR[NJ] bit setting.
1099 406352 cachegrind/callgrind fails ann tests because of missing a.c
1100 406354 dhat is broken on x86 (32bit)
1101 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
1102 406357 gdbserver_tests fails because of gdb output change
1103 406360 memcheck/tests/libstdc++.supp needs more supression variants
1104 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
1105 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
1106 407340 PPC64, does not support the vlogefp, vexptefp instructions.
1107 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
1108 n-i-bz Fix callgrind_annotate non deterministic order for equal total
1109 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
1110 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
1111 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
1113 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
1114 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
1115 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
1119 Release 3.14.0 (9 October 2018)
1120 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1122 3.14.0 is a feature release with many improvements and the usual collection of
1125 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1126 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1127 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1128 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
1129 support for X86/macOS 10.13, AMD64/macOS 10.13.
1131 * ==================== CORE CHANGES ===================
1133 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
1134 debug info for unloaded code. This allows saved stack traces (e.g. for
1135 memory leaks) to include file/line info for code that has been dlclose'd (or
1136 similar). See the user manual for more information and known limitations.
1138 * Ability to specify suppressions based on source file name and line number.
1140 * Majorly overhauled register allocator. No end-user changes, but the JIT
1141 generates code a bit more quickly now.
1143 * ================== PLATFORM CHANGES =================
1145 * Preliminary support for macOS 10.13 has been added.
1147 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
1149 * mips: support for MIPS SIMD architecture (MSA) has been added.
1151 * mips: support for MIPS N32 ABI has been added.
1153 * s390: partial support for vector instructions (integer and string) has been
1156 * ==================== TOOL CHANGES ====================
1158 * Helgrind: Addition of a flag
1159 --delta-stacktrace=no|yes [yes on linux amd64/x86]
1160 which specifies how full history stack traces should be computed.
1161 Setting this to =yes can speed up Helgrind by 25% when using
1162 --history-level=full.
1164 * Memcheck: reduced false positive rate for optimised code created by Clang 6
1165 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
1166 blocks more carefully to determine where it can avoid expensive definedness
1167 checks without loss of precision. This is controlled by the flag
1168 --expensive-definedness-checks=no|auto|yes [auto].
1170 * ==================== OTHER CHANGES ====================
1172 * Valgrind is now buildable with link-time optimisation (LTO). A new
1173 configure option --enable-lto=yes allows building Valgrind with LTO. If the
1174 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
1175 Note that if you are doing Valgrind development, --enable-lto=yes massively
1176 slows down the build process.
1178 * ==================== FIXED BUGS ====================
1180 The following bugs have been fixed or resolved. Note that "n-i-bz"
1181 stands for "not in bugzilla" -- that is, a bug that was reported to us
1182 but never got a bugzilla entry. We encourage you to file bugs in
1183 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1184 than mailing the developers (or mailing lists) directly -- bugs that
1185 are not entered into bugzilla tend to get forgotten about or ignored.
1187 To see details of a given bug, visit
1188 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1189 where XXXXXX is the bug number as listed below.
1191 79362 Debug info is lost for .so files when they are dlclose'd
1192 208052 strlcpy error when n = 0
1193 255603 exp-sgcheck Assertion '!already_present' failed
1194 338252 building valgrind with -flto (link time optimisation) fails
1195 345763 MIPS N32 ABI support
1196 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
1197 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
1198 372347 Replacement problem of the additional c++14/c++17 new/delete operators
1199 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1200 376257 helgrind history full speed up using a cached stack
1201 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
1203 379748 Fix missing pselect syscall (OS X 10.11)
1204 379754 Fix missing syscall ulock_wait (OS X 10.12)
1205 380397 s390x: __GI_strcspn() replacemenet needed
1206 381162 possible array overrun in VEX register allocator
1207 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
1208 381274 powerpc too chatty even with --sigill-diagnostics=no
1209 381289 epoll_pwait can have a NULL sigmask
1210 381553 VEX register allocator v3
1211 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
1212 381769 Use ucontext_t instead of struct ucontext
1213 381805 arm32 needs ld.so index hardwire for new glibc security fixes
1214 382256 gz compiler flag test doesn't work for gold
1215 382407 vg_perf needs "--terse" command line option
1216 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
1217 382563 MIPS MSA ASE support
1218 382998 xml-socket doesn't work
1219 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
1220 383723 Fix missing kevent_qos syscall (macOS 10.11)
1221 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
1222 384096 Mention AddrCheck at Memcheck's command line option [..]
1223 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
1224 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
1225 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
1226 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
1227 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
1228 384337 performance improvements to VEX register allocator v2 and v3
1229 384526 reduce number of spill insns generated by VEX register allocator v3
1230 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
1231 384631 Sanitise client args as printed with -v
1232 384633 Add a simple progress-reporting facility
1233 384987 VEX regalloc: allocate caller-save registers for short lived vregs
1234 385055 PPC VEX temporary storage exhausted
1235 385182 PPC64 is missing support for the DSCR
1236 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
1237 385207 PPC64, generate_store_FPRF() generates too many Iops
1238 385208 PPC64, xxperm instruction exhausts temporary memory
1239 385210 PPC64, vpermr instruction could exhaust temporary memory
1240 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
1241 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
1242 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
1243 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
1244 385408 s390x: z13 vector "support" instructions not implemented
1245 385409 s390x: z13 vector integer instructions not implemented
1246 385410 s390x: z13 vector string instructions not implemented
1247 385412 s390x: new non-vector z13 instructions not implemented
1248 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
1249 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
1250 385939 Optionally exit on the first error
1251 386318 valgrind.org/info/tools.html is missing SGCheck
1252 386425 running valgrind + wine on armv7l gives illegal opcode
1253 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
1254 387410 MIPSr6 support
1255 387664 Memcheck: make expensive-definedness-checks be the default
1256 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
1257 387766 asm shifts cause false positive "Conditional jump or move depends
1258 on uninitialised value"
1259 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
1260 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
1261 388786 Support bpf syscall in amd64 Linux
1262 388862 Add replacements for wmemchr and wcsnlen on Linux
1263 389065 valgrind meets gcc flag -Wlogical-op
1264 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
1265 390471 suppression by specification of source-file line number
1266 390723 make xtree dump files world wide readable, similar to log files
1267 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
1268 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
1269 392118 unhandled amd64-linux syscall: 332 (statx)
1270 392449 callgrind not clearing the number of calls properly
1271 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
1272 lxssp, stxssp and stxvl instructions.
1273 393023 callgrind_control risks using the wrong vgdb
1274 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
1275 393099 posix_memalign() invalid write if alignment == 0
1276 393146 failing assert "is_DebugInfo_active(di)"
1277 395709 PPC64 is missing support for the xvnegsp instruction
1278 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
1280 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
1281 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
1282 396839 s390x: Trap instructions not implemented
1283 396887 arch_prctl should return EINVAL on unknown option
1284 == 397286 crash before launching binary (Unsupported arch_prctl option)
1285 == 397393 valgrind: the 'impossible' happened: (Archlinux)
1286 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
1287 396906 compile tests failure on mips32-linux: broken inline asm in tests on
1289 397012 glibc ld.so uses arch_prctl on i386
1290 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
1291 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
1292 397424 glibc 2.27 and gdb_server tests
1293 398028 Assertion `cfsi_fits` failing in simple C program
1294 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
1296 n-i-bz Fix missing workq_ops operations (macOS)
1297 n-i-bz fix bug in strspn replacement
1298 n-i-bz Add support for the Linux BLKFLSBUF ioctl
1299 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
1300 n-i-bz Fix possible stack trashing by semctl syscall wrapping
1301 n-i-bz Add support for the Linux membarrier() system call
1302 n-i-bz x86 front end: recognise and handle UD2 correctly
1303 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
1304 correctly aligned before entering the handler.
1306 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
1307 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
1308 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
1312 Release 3.13.0 (15 June 2017)
1313 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1315 3.13.0 is a feature release with many improvements and the usual collection of
1318 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1319 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1320 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1321 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
1323 * ==================== CORE CHANGES ===================
1325 * The translation cache size has been increased to keep up with the demands of
1326 large applications. The maximum number of sectors has increased from 24 to
1327 48. The default number of sectors has increased from 16 to 32 on all
1328 targets except Android, where the increase is from 6 to 12.
1330 * The amount of memory that Valgrind can use has been increased from 64GB to
1331 128GB. In particular this means your application can allocate up to about
1332 60GB when running on Memcheck.
1334 * Valgrind's default load address has been changed from 0x3800'0000 to
1335 0x5800'0000, so as to make it possible to load larger executables. This
1336 should make it possible to load executables of size at least 1200MB.
1338 * A massive spaceleak caused by reading compressed debuginfo files has been
1339 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
1342 * The C++ demangler has been updated.
1344 * Support for demangling Rust symbols has been added.
1346 * A new representation of stack traces, the "XTree", has been added. An XTree
1347 is a tree of stacktraces with data associated with the stacktraces. This is
1348 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
1349 consumption of your program. Reporting is controlled by the new options
1350 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
1352 A report can also be produced on demand using the gdbserver monitor command
1353 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
1354 format' and 'massif format. The existing visualisers for these formats (e.g.
1355 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
1356 analyse these reports.
1358 Memcheck can also produce XTree leak reports using the Callgrind file
1359 format. For more details, see the user manual.
1361 * ================== PLATFORM CHANGES =================
1363 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
1365 * amd64: fixes for JIT failure problems on long AVX2 code blocks
1367 * amd64 and x86: support for CET prefixes has been added
1369 * arm32: a few missing ARMv8 instructions have been implemented
1371 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
1372 Store-Conditional instructions has been added. This is to deal with
1373 processor implementations that implement the LL/SC specifications strictly
1374 and as a result cause Valgrind to hang in certain situations. The
1375 alternative implementation is automatically enabled at startup, as required.
1376 You can use the option --sim-hints=fallback-llsc to force-enable it if you
1379 * Support for OSX 10.12 has been improved.
1381 * On Linux, clone handling has been improved to honour CLONE_VFORK that
1382 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
1383 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
1384 CLONE_VM exact semantics will (still) not work.
1386 * The TileGX/Linux port has been removed because it appears to be both unused
1389 * ==================== TOOL CHANGES ====================
1393 - Memcheck should give fewer false positives when running optimised
1394 Clang/LLVM generated code.
1396 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1398 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
1399 to produce the end of execution leak report in a xtree callgrind format
1402 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
1403 the leak report in an xtree file.
1407 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1409 - For some workloads (typically, for big applications), Massif memory
1410 consumption and CPU consumption has decreased significantly.
1414 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1416 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
1417 for Ada gnat compiled applications.
1419 * ==================== OTHER CHANGES ====================
1421 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
1422 append the inner guest stacktrace to the inner host stacktrace. This helps
1423 to investigate the errors reported by the outer, when they are caused by the
1424 inner guest program (such as an inner regtest). See README_DEVELOPERS for
1427 * To allow fast detection of callgrind files by desktop environments and file
1428 managers, the format was extended to have an optional first line that
1429 uniquely identifies the format ("# callgrind format"). Callgrind creates
1430 this line now, as does the new xtree functionality.
1432 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
1433 have a new %n format letter that is replaced by a sequence number.
1435 * "--version -v" now shows the SVN revision numbers from which Valgrind was
1438 * ==================== FIXED BUGS ====================
1440 The following bugs have been fixed or resolved. Note that "n-i-bz"
1441 stands for "not in bugzilla" -- that is, a bug that was reported to us
1442 but never got a bugzilla entry. We encourage you to file bugs in
1443 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1444 than mailing the developers (or mailing lists) directly -- bugs that
1445 are not entered into bugzilla tend to get forgotten about or ignored.
1447 To see details of a given bug, visit
1448 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1449 where XXXXXX is the bug number as listed below.
1451 162848 --log-file output isn't split when a program forks
1452 340777 Illegal instruction on mips (ar71xx)
1453 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
1454 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
1455 to a different stack.
1456 344139 x86 stack-seg overrides, needed by the Wine people
1457 344524 store conditional of guest applications always fail - observed on
1459 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
1460 352395 Please provide SVN revision info in --version -v
1461 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
1462 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
1463 INVALID_POSIX_THREADID' failed
1464 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
1465 barrier implementation
1466 358697 valgrind.h: Some code remains even when defining NVALGRIND
1467 359202 Add musl libc configure/compile
1468 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
1469 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
1470 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
1471 362223 assertion failed when .valgrindrc is a directory instead of a file
1472 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
1473 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
1474 368507 can't malloc chunks larger than about 34GB
1475 368529 Android arm target link error, missing atexit and pthread_atfork
1476 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
1477 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
1478 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
1479 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
1480 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
1481 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
1482 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
1483 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
1484 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
1485 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
1486 370028 Reduce the number of compiler warnings on MIPS platforms
1487 370635 arm64 missing syscall getcpu
1488 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
1489 371227 Clean AArch64 syscall table
1490 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
1491 371471 Valgrind complains about non legit memory leaks on placement new (C++)
1492 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
1493 371503 disInstr(arm64): unhandled instruction 0xF89F0000
1494 371869 support '%' in symbol Z-encoding
1495 371916 execution tree xtree concept
1496 372120 c++ demangler demangles symbols which are not c++
1497 372185 Support of valgrind on ARMv8 with 32 bit executable
1498 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1499 372195 Power PC, xxsel instruction is not always recognized.
1500 372504 Hanging on exit_group
1501 372600 process loops forever when fatal signals are arriving quickly
1502 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1503 373046 Stacks registered by core are never deregistered
1504 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1505 373086 Implement additional Xen hypercalls
1506 373192 Calling posix_spawn in glibc 2.24 completely broken
1507 373488 Support for fanotify API on ARM64 architecture
1508 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1509 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1510 373938 const IRExpr arguments for matchIRExpr()
1511 374719 some spelling fixes
1512 374963 increase valgrind's load address to prevent mmap failure
1513 375514 valgrind_get_tls_addr() does not work in case of static TLS
1514 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1515 for ML_(find_rx_mapping)()
1516 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1517 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1518 == 377159 "vex: the `impossible' happened" still present
1519 == 375150 Assertion 'tres.status == VexTransOK' failed
1520 == 378068 valgrind crashes on AVX2 function in FFmpeg
1521 376142 Segfaults on MIPS Cavium Octeon boards
1522 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1523 376455 Solaris: unhandled syscall lgrpsys(180)
1524 376518 Solaris: unhandled fast trap getlgrp(6)
1525 376611 ppc64 and arm64 don't know about prlimit64 syscall
1526 376729 PPC64, remove R2 from the clobber list
1528 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1529 to be wrongly marked as addressable
1530 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1531 PIE enabled by default
1532 377376 memcheck/tests/linux/getregset fails with glibc2.24
1533 377427 PPC64, lxv instruction failing on odd destination register
1534 377478 PPC64: ISA 3.0 setup fixes
1535 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1536 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1537 and 2 args for FUTEX_TRYLOCK_PI
1538 377717 Fix massive space leak when reading compressed debuginfo sections
1539 377891 Update Xen 4.6 domctl wrappers
1540 377930 fcntl syscall wrapper is missing flock structure check
1541 378524 libvexmultiarch_test regression on s390x and ppc64
1542 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1543 378673 Update libiberty demangler
1544 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1545 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1546 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1547 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1548 (task_register_dyld_image_infos)
1549 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1550 (task_register_dyld_shared_cache_image_info)
1551 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1552 379473 MIPS: add support for rdhwr cycle counter register
1553 379504 remove TileGX/Linux port
1554 379525 Support more x86 nop opcodes
1555 379838 disAMode(x86): not an addr!
1556 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1557 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1558 379895 clock_gettime does not execute POST syscall wrapper
1559 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1560 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1561 380200 xtree generated callgrind files refer to files without directory name
1562 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1563 380397 s390x: __GI_strcspn() replacement needed
1564 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1566 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1567 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1568 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1572 Release 3.12.0 (20 October 2016)
1573 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1575 3.12.0 is a feature release with many improvements and the usual
1576 collection of bug fixes.
1578 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1579 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1580 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1581 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1582 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1583 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1585 * ================== PLATFORM CHANGES =================
1587 * POWER: Support for ISA 3.0 has been added
1589 * mips: support for O32 FPXX ABI has been added.
1590 * mips: improved recognition of different processors
1591 * mips: determination of page size now done at run time
1593 * amd64: Partial support for AMD FMA4 instructions.
1595 * arm, arm64: Support for v8 crypto and CRC instructions.
1597 * Improvements and robustification of the Solaris port.
1599 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1601 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1602 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1603 where possible. Valgrind's support for 32-bit x86 has stagnated in
1604 recent years and has fallen far behind that for 64-bit x86
1605 instructions. By contrast 64-bit x86 is well supported, up to and
1608 * ==================== TOOL CHANGES ====================
1612 - Added meta mempool support for describing a custom allocator which:
1613 - Auto-frees all chunks assuming that destroying a pool destroys all
1615 - Uses itself to allocate other memory blocks
1617 - New flag --ignore-range-below-sp to ignore memory accesses below
1618 the stack pointer, if you really have to. The related flag
1619 --workaround-gcc296-bugs=yes is now deprecated. Use
1620 --ignore-range-below-sp=1024-1 as a replacement.
1624 - Improved thread startup time significantly on non-Linux platforms.
1628 - Added collection of the metric "tot-blocks-allocd"
1630 * ==================== OTHER CHANGES ====================
1632 * Replacement/wrapping of malloc/new related functions is now done not just
1633 for system libraries by default, but for any globally defined malloc/new
1634 related function (both in shared libraries and statically linked alternative
1635 malloc implementations). The dynamic (runtime) linker is excluded, though.
1636 To only intercept malloc/new related functions in
1637 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1638 "nouserintercepts" can be any non-existing library name).
1639 This new functionality is not implemented for MacOS X.
1641 * The maximum number of callers in a suppression entry is now equal to
1642 the maximum size for --num-callers (500).
1643 Note that --gen-suppressions=yes|all similarly generates suppressions
1644 containing up to --num-callers frames.
1646 * New and modified GDB server monitor features:
1648 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1649 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1652 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1653 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1656 * Valgrind is able to read compressed debuginfo sections in two formats:
1657 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1658 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1660 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1661 for the most common use case (x86_64-linux, Memcheck) has been
1664 * Improved performance for programs that do a lot of discarding of
1665 instruction address ranges of 8KB or less.
1667 * The C++ symbol demangler has been updated.
1669 * More robustness against invalid syscall parameters on Linux.
1671 * ==================== FIXED BUGS ====================
1673 The following bugs have been fixed or resolved. Note that "n-i-bz"
1674 stands for "not in bugzilla" -- that is, a bug that was reported to us
1675 but never got a bugzilla entry. We encourage you to file bugs in
1676 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1677 than mailing the developers (or mailing lists) directly -- bugs that
1678 are not entered into bugzilla tend to get forgotten about or ignored.
1680 To see details of a given bug, visit
1681 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1682 where XXXXXX is the bug number as listed below.
1684 191069 Exiting due to signal not reported in XML output
1685 199468 Suppressions: stack size limited to 25
1686 while --num-callers allows more frames
1687 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1688 278744 cvtps2pd with redundant RexW
1689 303877 valgrind doesn't support compressed debuginfo sections.
1690 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1691 348345 Assertion fails for negative lineno
1692 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1693 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1694 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1695 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1696 352197 mips: mmap2() not wrapped correctly for page size > 4096
1697 353083 arm64 doesn't implement various xattr system calls
1698 353084 arm64 doesn't support sigpending system call
1699 353137 www: update info for Supported Platforms
1700 353138 www: update "The Valgrind Developers" page
1701 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1704 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1705 353398 WARNING: unhandled amd64-solaris syscall: 207
1706 353660 XML in auxwhat tag not escaping reserved symbols properly
1707 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1708 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1709 353802 ELF debug info reader confused with multiple .rodata sections
1710 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1711 353917 unhandled amd64-solaris syscall fchdir(120)
1712 353920 unhandled amd64-solaris syscall: 170
1713 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1714 354392 unhandled amd64-solaris syscall: 171
1715 354797 Vbit test does not include Iops for Power 8 instruction support
1716 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1720 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1721 355188 valgrind should intercept all malloc related global functions
1722 355454 do not intercept malloc related symbols from the runtime linker
1723 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1724 356044 Dwarf line info reader misinterprets is_stmt register
1725 356112 mips: replace addi with addiu
1726 356393 valgrind (vex) crashes because isZeroU happened
1729 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1730 356678 arm64-linux: unhandled syscall 232 (mincore)
1731 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1732 356823 Unsupported ARM instruction: stlex
1733 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1734 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1735 357673 crash if I try to run valgrind with a binary link with libcurl
1736 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1737 357871 pthread_spin_destroy not properly wrapped
1738 357887 Calls to VG_(fclose) do not close the file descriptor
1739 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1740 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1741 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1742 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1743 359181 Buffer Overflow during Demangling
1744 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1745 359289 s390x: popcnt (B9E1) not implemented
1746 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1747 359503 Add missing syscalls for aarch64 (arm64)
1748 359645 "You need libc6-dbg" help message could be more helpful
1749 359703 s390: wire up separate socketcalls system calls
1750 359724 getsockname might crash - deref_UInt should call safe_to_deref
1751 359733 amd64 implement ld.so strchr/index override like x86
1752 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1753 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1755 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1756 359871 Incorrect mask handling in ppoll
1757 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1758 360008 Contents of Power vr registers contents is not printed correctly when
1759 the --vgdb-shadow-registers=yes option is used
1760 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1761 non-zero shadow bits
1762 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1763 360425 arm64 unsupported instruction ldpsw
1765 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1766 360571 Error about the Android Runtime reading below the stack pointer on ARM
1767 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1768 360749 kludge for multiple .rodata sections on Solaris no longer needed
1769 360752 raise the number of reserved fds in m_main.c from 10 to 12
1770 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1771 361226 s390x: risbgn (EC59) not implemented
1772 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1773 361354 ppc64[le]: wire up separate socketcalls system calls
1774 361615 Inconsistent termination for multithreaded process terminated by signal
1775 361926 Unhandled Solaris syscall: sysfs(84)
1776 362009 V dumps core on unimplemented functionality before threads are created
1777 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1778 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1779 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1780 362953 Request for an update to the Valgrind Developers page
1781 363680 add renameat2() support
1782 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1783 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1784 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1785 364058 clarify in manual limitations of array overruns detections
1786 364413 pselect sycallwrapper mishandles NULL sigmask
1787 364728 Power PC, missing support for several HW registers in
1788 get_otrack_shadow_offset_wrk()
1789 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1790 365273 Invalid write to stack location reported after signal handler runs
1791 365912 ppc64BE segfault during jm-insns test (RELRO)
1792 366079 FPXX Support for MIPS32 Valgrind
1793 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1794 366344 Multiple unhandled instruction for Aarch64
1795 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1796 367995 Integration of memcheck with custom memory allocator
1797 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1798 368412 False positive result for altivec capability check
1799 368416 Add tc06_two_races_xml.exp output for ppc64
1800 368419 Perf Events ioctls not implemented
1801 368461 mmapunmap test fails on ppc64
1802 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1803 369000 AMD64 fma4 instructions unsupported.
1804 369169 ppc64 fails jm_int_isa_2_07 test
1805 369175 jm_vec_isa_2_07 test crashes on ppc64
1806 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1807 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1808 369359 msghdr_foreachfield can crash when handling bad iovec
1809 369360 Bad sigprocmask old or new sets can crash valgrind
1810 369361 vmsplice syscall wrapper crashes on bad iovec
1811 369362 Bad sigaction arguments crash valgrind
1812 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1813 369402 Bad set/get_thread_area pointer crashes valgrind
1814 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1815 369446 valgrind crashes on unknown fcntl command
1816 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1817 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1818 370265 ISA 3.0 HW cap stuff needs updating
1819 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1820 372195 Power PC, xxsel instruction is not always recognized
1822 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1823 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1824 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1825 n-i-bz Fix ppoll_alarm exclusion on OS X
1826 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1827 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1828 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1829 n-i-bz amd64: memcheck false positive with shr %edx
1830 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1831 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1832 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1833 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1835 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1836 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1837 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1841 Release 3.11.0 (22 September 2015)
1842 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1844 3.11.0 is a feature release with many improvements and the usual
1845 collection of bug fixes.
1847 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1848 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1849 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1850 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1851 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1852 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1854 * ================== PLATFORM CHANGES =================
1856 * Support for Solaris/x86 and Solaris/amd64 has been added.
1858 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1860 * Preliminary support for the Tilera TileGX architecture has been added.
1862 * s390x: It is now required for the host to have the "long displacement"
1863 facility. The oldest supported machine model is z990.
1865 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1866 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1868 * The JIT's register allocator is significantly faster, making the JIT
1869 as a whole somewhat faster, so JIT-intensive activities, for example
1870 program startup, are modestly faster, around 5%.
1872 * There have been changes to the default settings of several command
1873 line flags, as detailed below.
1875 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1876 capable hosts, the simulated CPUID will now indicate AVX2 support.
1878 * ==================== TOOL CHANGES ====================
1882 - The default value for --leak-check-heuristics has been changed from
1883 "none" to "all". This helps to reduce the number of possibly
1884 lost blocks, in particular for C++ applications.
1886 - The default value for --keep-stacktraces has been changed from
1887 "malloc-then-free" to "malloc-and-free". This has a small cost in
1888 memory (one word per malloc-ed block) but allows Memcheck to show the
1889 3 stacktraces of a dangling reference: where the block was allocated,
1890 where it was freed, and where it is acccessed after being freed.
1892 - The default value for --partial-loads-ok has been changed from "no" to
1893 "yes", so as to avoid false positive errors resulting from some kinds
1894 of vectorised loops.
1896 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1897 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1898 than get_vbits when you need to associate byte data value with
1899 their corresponding validity bits.
1901 - The 'block_list' monitor command has been enhanced:
1902 o it can print a range of loss records
1903 o it now accepts an optional argument 'limited <max_blocks>'
1904 to control the number of blocks printed.
1905 o if a block has been found using a heuristic, then
1906 'block_list' now shows the heuristic after the block size.
1907 o the loss records/blocks to print can be limited to the blocks
1908 found via specified heuristics.
1910 - The C helper functions used to instrument loads on
1911 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1912 replaced by handwritten assembly sequences. This gives speedups
1913 in the region of 0% to 7% for those targets only.
1915 - A new command line option, --expensive-definedness-checks=yes|no,
1916 has been added. This is useful for avoiding occasional invalid
1917 uninitialised-value errors in optimised code. Watch out for
1918 runtime degradation, as this can be up to 25%. As always, though,
1919 the slowdown is highly application specific. The default setting
1924 - A new monitor command 'all_snapshots <filename>' dumps all
1925 snapshots taken so far.
1929 - Significant memory reduction and moderate speedups for
1930 --history-level=full for applications accessing a lot of memory
1931 with many different stacktraces.
1933 - The default value for --conflict-cache-size=N has been doubled to
1934 2000000. Users that were not using the default value should
1935 preferably also double the value they give.
1937 The default was changed due to the changes in the "full history"
1938 implementation. Doubling the value gives on average a slightly more
1939 complete history and uses similar memory (or significantly less memory
1940 in the worst case) than the previous implementation.
1942 - The Helgrind monitor command 'info locks' now accepts an optional
1943 argument 'lock_addr', which shows information about the lock at the
1946 - When using --history-level=full, the new Helgrind monitor command
1947 'accesshistory <addr> [<len>]' will show the recorded accesses for
1948 <len> (or 1) bytes at <addr>.
1950 * ==================== OTHER CHANGES ====================
1952 * The default value for the --smc-check option has been changed from
1953 "stack" to "all-non-file" on targets that provide automatic D-I
1954 cache coherence (x86, amd64 and s390x). The result is to provide,
1955 by default, transparent support for JIT generated and self-modifying
1956 code on all targets.
1958 * Mac OS X only: the default value for the --dsymutil option has been
1959 changed from "no" to "yes", since any serious usage on Mac OS X
1960 always required it to be "yes".
1962 * The command line options --db-attach and --db-command have been removed.
1963 They were deprecated in 3.10.0.
1965 * When a process dies due to a signal, Valgrind now shows the signal
1966 and the stacktrace at default verbosity (i.e. verbosity 1).
1968 * The address description logic used by Memcheck and Helgrind now
1969 describes addresses in anonymous segments, file mmap-ed segments,
1970 shared memory segments and the brk data segment.
1972 * The new option --error-markers=<begin>,<end> can be used to mark the
1973 begin/end of errors in textual output mode, to facilitate
1974 searching/extracting errors in output files that mix valgrind errors
1975 with program output.
1977 * The new option --max-threads=<number> can be used to change the number
1978 of threads valgrind can handle. The default is 500 threads which
1979 should be more than enough for most applications.
1981 * The new option --valgrind-stacksize=<number> can be used to change the
1982 size of the private thread stacks used by Valgrind. This is useful
1983 for reducing memory use or increasing the stack size if Valgrind
1984 segfaults due to stack overflow.
1986 * The new option --avg-transtab-entry-size=<number> can be used to specify
1987 the expected instrumented block size, either to reduce memory use or
1988 to avoid excessive retranslation.
1990 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1992 * New and modified GDB server monitor features:
1994 - When a signal is reported in GDB, you can now use the GDB convenience
1995 variable $_siginfo to examine detailed signal information.
1997 - Valgrind's gdbserver now allows the user to change the signal
1998 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1999 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
2000 continue without passing the signal to the process.
2002 - With GDB >= 7.10, the command 'target remote'
2003 will automatically load the executable file of the process running
2004 under Valgrind. This means you do not need to specify the executable
2005 file yourself, GDB will discover it itself. See GDB documentation about
2006 'qXfer:exec-file:read' packet for more info.
2008 * ==================== FIXED BUGS ====================
2010 The following bugs have been fixed or resolved. Note that "n-i-bz"
2011 stands for "not in bugzilla" -- that is, a bug that was reported to us
2012 but never got a bugzilla entry. We encourage you to file bugs in
2013 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2014 than mailing the developers (or mailing lists) directly -- bugs that
2015 are not entered into bugzilla tend to get forgotten about or ignored.
2017 To see details of a given bug, visit
2018 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2019 where XXXXXX is the bug number as listed below.
2021 116002 VG_(printf): Problems with justification of strings and integers
2022 155125 avoid cutting away file:lineno after long function name
2023 197259 Unsupported arch_prtctl PR_SET_GS option
2024 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
2025 201216 Fix Valgrind does not support pthread_sigmask() on OS X
2026 201435 Fix Darwin: -v does not show kernel version
2027 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
2028 211256 Fixed an outdated comment regarding the default platform.
2029 211529 Incomplete call stacks for code compiled by newer versions of MSVC
2030 211926 Avoid compilation warnings in valgrind.h with -pedantic
2031 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
2033 226609 Crediting upstream authors in man page
2034 231257 Valgrind omits path when executing script from shebang line
2035 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
2036 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
2037 269360 s390x: Fix addressing mode selection for compare-and-swap
2038 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
2040 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
2041 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
2042 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
2043 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
2044 330147 libmpiwrap PMPI_Get_count returns undefined value
2045 333051 mmap of huge pages fails due to incorrect alignment
2047 334802 valgrind does not always explain why a given option is bad
2048 335618 mov.w rN, pc/sp (ARM32)
2049 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
2054 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
2055 338602 AVX2 bit in CPUID missing
2056 338606 Strange message for scripts with invalid interpreter
2057 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
2058 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
2059 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
2061 339156 gdbsrv not called for fatal signal
2062 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
2063 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
2064 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
2065 339442 Fix testsuite build failure on OS X 10.9
2066 339542 Enable compilation with Intel's ICC compiler
2067 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
2068 339688 Mac-specific ASM does not support .version directive (cpuid,
2069 tronical and pushfpopf tests)
2070 339745 Valgrind crash when check Marmalade app (partial fix)
2071 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
2072 339778 Linux/TileGx platform support for Valgrind
2073 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
2074 339789 Fix none/tests/execve test on Mac OS X 10.9
2075 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
2076 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
2077 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
2078 340392 Allow user to select more accurate definedness checking in memcheck
2079 to avoid invalid complaints on optimised code
2080 340430 Fix some grammatical weirdness in the manual.
2081 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
2082 341419 Signal handler ucontext_t not filled out correctly on OS X
2083 341539 VG_(describe_addr) should not describe address as belonging to client
2084 segment if it is past the heap end
2085 341613 Enable building of manythreads and thread-exits tests on Mac OS X
2086 341615 Fix none/tests/darwin/access_extended test on Mac OS X
2087 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
2088 341789 aarch64: shmat fails with valgrind on ARMv8
2089 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
2090 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
2091 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
2092 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
2093 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
2094 342221 socket connect false positive uninit memory for unknown af family
2095 342353 Allow dumping full massif output while valgrind is still running
2096 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
2100 342603 Add I2C_SMBUS ioctl support
2101 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
2102 342683 Mark memory past the initial brk limit as unaddressable
2103 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
2104 342795 Internal glibc __GI_mempcpy call should be intercepted
2105 342841 s390x: Support instructions fiebr(a) and fidbr(a)
2106 343012 Unhandled syscall 319 (memfd_create)
2107 343069 Patch updating v4l2 API support
2108 343173 helgrind crash during stack unwind
2109 343219 fix GET_STARTREGS for arm
2110 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
2111 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
2112 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
2113 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
2114 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
2115 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
2116 343597 ppc64le: incorrect use of offseof macro
2117 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
2118 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
2119 343732 Unhandled syscall 144 (setgid) on aarch64
2120 343733 Unhandled syscall 187 (msgctl and related) on aarch64
2121 343802 s390x: False positive "conditional jump or move depends on [..]
2122 343902 --vgdb=yes doesn't break when --xml=yes is used
2123 343967 Don't warn about setuid/setgid/setcap executable for directories
2124 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
2125 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
2126 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
2127 344054 www - update info for Solaris/illumos
2128 344416 'make regtest' does not work cleanly on OS X
2129 344235 Remove duplicate include of pub_core_aspacemgr.h
2130 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
2131 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
2132 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
2133 344314 callgrind_annotate ... warnings about commands containing newlines
2134 344318 socketcall should wrap recvmmsg and sendmmsg
2135 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
2136 344416 Fix 'make regtest' does not work cleanly on OS X
2137 344499 Fix compilation for Linux kernel >= 4.0.0
2138 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
2139 unix:349 (__pthread_fchdir)
2140 344559 Garbage collection of unused segment names in address space manager
2141 344560 Fix stack traces missing penultimate frame on OS X
2142 344621 Fix memcheck/tests/err_disable4 test on OS X
2143 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
2144 344702 Fix missing libobjc suppressions on OS X 10.10
2146 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
2147 344939 Fix memcheck/tests/xml1 on OS X 10.10
2148 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
2149 345079 Fix build problems in VEX/useful/test_main.c
2150 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
2151 345177 arm64: prfm (reg) not implemented
2152 345215 Performance improvements for the register allocator
2153 345248 add support for Solaris OS in valgrind
2154 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
2155 345394 Fix memcheck/tests/strchr on OS X
2156 345637 Fix memcheck/tests/sendmsg on OS X
2157 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
2158 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
2159 345887 Fix an assertion in the address space manager
2160 345928 amd64: callstack only contains current function for small stacks
2161 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
2162 345987 MIPS64: Implement cavium LHX instruction
2163 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
2164 346185 Fix typo saving altivec register v24
2165 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
2166 and LibVEX_GuestPPC64_get_CR()
2167 346270 Regression tests none/tests/jm_vec/isa_2_07 and
2168 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
2169 346307 fuse filesystem syscall deadlocks
2170 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
2171 346411 MIPS: SysRes::_valEx handling is incorrect
2172 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
2173 346474 PPC64 Power 8, spr TEXASRU register not supported
2174 346487 Compiler generates "note" about a future ABI change for PPC64
2175 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
2176 and causing spurious "invalid read of size 8" warnings
2177 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
2178 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
2179 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
2180 347322 Power PC regression test cleanup
2181 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
2183 347389 unhandled syscall: 373 (Linux ARM syncfs)
2184 347686 Patch set to cleanup PPC64 regtests
2185 347978 Remove bash dependencies where not needed
2186 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
2187 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
2189 348102 Patch updating v4l2 API support
2190 348247 amd64 front end: jno jumps wrongly when overflow is not set
2191 348269 Improve mmap MAP_HUGETLB support.
2192 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
2193 348345 Assertion fails for negative lineno
2194 348377 Unsupported ARM instruction: yield
2195 348565 Fix detection of command line option availability for clang
2196 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
2197 348728 Fix broken check for VIDIOC_G_ENC_INDEX
2198 348748 Fix redundant condition
2199 348890 Fix clang warning about unsupported --param inline-unit-growth=900
2200 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
2201 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
2202 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
2203 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
2204 349626 Implemented additional Xen hypercalls
2205 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
2206 349790 Clean up of the hardware capability checking utilities.
2207 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
2208 349874 Fix typos in source code
2209 349879 memcheck: add handwritten assembly for helperc_LOADV*
2210 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
2211 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
2212 350202 Add limited param to 'monitor block_list'
2213 350290 s390x: Support instructions fixbr(a)
2214 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
2215 350809 Fix none/tests/async-sigs for Solaris
2216 350811 Remove reference to --db-attach which has been removed.
2217 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
2218 350854 hard-to-understand code in VG_(load_ELF)()
2219 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2220 351386 Solaris: Cannot run ld.so.1 under Valgrind
2221 351474 Fix VG_(iseqsigset) as obvious
2222 351531 Typo in /include/vki/vki-xen-physdev.h header guard
2223 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
2224 351858 ldsoexec support on Solaris
2225 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
2226 352130 helgrind reports false races for printfs using mempcpy on FILE* state
2227 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
2228 352320 arm64 crash on none/tests/nestedfs
2229 352765 Vbit test fails on Power 6
2230 352768 The mbar instruction is missing from the Power PC support
2231 352769 Power PC program priority register (PPR) is not supported
2232 n-i-bz Provide implementations of certain compiler builtins to support
2233 compilers that may not provide those
2234 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
2235 n-i-bz Fix compilation on distros with glibc < 2.5
2236 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
2237 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
2238 n-i-bz mremap did not work properly on shared memory
2239 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
2240 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
2242 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
2243 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
2244 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
2248 Release 3.10.1 (25 November 2014)
2249 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2250 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
2251 and backports fixes for all reported missing AArch64 ARMv8 instructions
2252 and syscalls from the trunk. If you package or deliver 3.10.0 for others
2253 to use, you might want to consider upgrading to 3.10.1 instead.
2255 The following bugs have been fixed or resolved. Note that "n-i-bz"
2256 stands for "not in bugzilla" -- that is, a bug that was reported to us
2257 but never got a bugzilla entry. We encourage you to file bugs in
2258 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2259 than mailing the developers (or mailing lists) directly -- bugs that
2260 are not entered into bugzilla tend to get forgotten about or ignored.
2262 To see details of a given bug, visit
2263 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2264 where XXXXXX is the bug number as listed below.
2266 335440 arm64: ld1 (single structure) is not implemented
2267 335713 arm64: unhanded instruction: prfm (immediate)
2268 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
2269 339182 ppc64: AvSplat ought to load destination vector register with [..]
2270 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
2271 339433 ppc64 lxvw4x instruction uses four 32-byte loads
2272 339645 Use correct tag names in sys_getdents/64 wrappers
2273 339706 Fix false positive for ioctl(TIOCSIG) on linux
2274 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
2275 339853 arm64 times syscall unknown
2276 339855 arm64 unhandled getsid/setsid syscalls
2277 339858 arm64 dmb sy not implemented
2278 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
2279 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
2280 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
2282 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
2283 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
2284 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
2285 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
2286 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
2287 340509 arm64: unhandled instruction fcvtas
2288 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
2289 340632 arm64: unhandled instruction fcvtas
2290 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
2291 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
2292 340788 warning: unhandled syscall: 318 (getrandom)
2293 340807 disInstr(arm): unhandled instruction: 0xEE989B20
2294 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
2295 340922 arm64: unhandled getgroups/setgroups syscalls
2296 350251 Fix typo in VEX utility program (test_main.c).
2297 350407 arm64: unhandled instruction ucvtf (vector, integer)
2298 350809 none/tests/async-sigs breaks when run under cron on Solaris
2299 350811 update README.solaris after r15445
2300 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
2301 350854 strange code in VG_(load_ELF)()
2302 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2303 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
2304 n-i-bz Add missing ]] to terminate CDATA.
2305 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
2306 n-i-bz Enable sys_fadvise64_64 on arm32.
2307 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
2308 n-i-bz Add test cases for all known arm64 load/store instructions.
2309 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
2310 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
2311 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
2312 n-i-bz arm64: Implement "fcvtpu w, s".
2313 n-i-bz arm64: implement ADDP and various others
2314 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
2315 n-i-bz arm64: enable FCVT{A,N}S X,S.
2317 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
2321 Release 3.10.0 (10 September 2014)
2322 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2324 3.10.0 is a feature release with many improvements and the usual
2325 collection of bug fixes.
2327 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
2328 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
2329 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
2330 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
2331 significantly improved relative to the 3.9.0 release.
2333 * ================== PLATFORM CHANGES =================
2335 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
2336 is mostly complete, and is usable, but some SIMD instructions are as
2339 * Support for little-endian variant of the 64-bit POWER architecture.
2341 * Support for Android on MIPS32.
2343 * Support for 64bit FPU on MIPS32 platforms.
2345 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
2347 * Configuration for and running on Android targets has changed.
2348 See README.android in the source tree for details.
2350 * ================== DEPRECATED FEATURES =================
2352 * --db-attach is now deprecated and will be removed in the next
2353 valgrind feature release. The built-in GDB server capabilities are
2354 superior and should be used instead. Learn more here:
2355 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
2357 * ==================== TOOL CHANGES ====================
2361 - Client code can now selectively disable and re-enable reporting of
2362 invalid address errors in specific ranges using the new client
2363 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
2364 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
2366 - Leak checker: there is a new leak check heuristic called
2367 "length64". This is used to detect interior pointers pointing 8
2368 bytes inside a block, on the assumption that the first 8 bytes
2369 holds the value "block size - 8". This is used by
2370 sqlite3MemMalloc, for example.
2372 - Checking of system call parameters: if a syscall parameter
2373 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
2374 several fields not initialised, an error is now reported for each
2375 field. Previously, an error was reported only for the first
2376 uninitialised field.
2378 - Mismatched alloc/free checking: a new flag
2379 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
2380 such checks if necessary.
2384 - Improvements to error messages:
2386 o Race condition error message involving heap allocated blocks also
2387 show the thread number that allocated the raced-on block.
2389 o All locks referenced by an error message are now announced.
2390 Previously, some error messages only showed the lock addresses.
2392 o The message indicating where a lock was first observed now also
2393 describes the address/location of the lock.
2395 - Helgrind now understands the Ada task termination rules and
2396 creates a happens-before relationship between a terminated task
2397 and its master. This avoids some false positives and avoids a big
2398 memory leak when a lot of Ada tasks are created and terminated.
2399 The interceptions are only activated with forthcoming releases of
2400 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
2402 - A new GDB server monitor command "info locks" giving the list of
2403 locks, their location, and their status.
2407 - callgrind_control now supports the --vgdb-prefix argument,
2408 which is needed if valgrind was started with this same argument.
2410 * ==================== OTHER CHANGES ====================
2412 * Unwinding through inlined function calls. Stack unwinding can now
2413 make use of Dwarf3 inlined-unwind information if it is available.
2414 The practical effect is that inlined calls become visible in stack
2415 traces. The suppression matching machinery has been adjusted
2416 accordingly. This is controlled by the new option
2417 --read-inline-info=yes|no. Currently this is enabled by default
2418 only on Linux and Android targets and only for the tools Memcheck,
2421 * Valgrind can now read EXIDX unwind information on 32-bit ARM
2422 targets. If an object contains both CFI and EXIDX unwind
2423 information, Valgrind will prefer the CFI over the EXIDX. This
2424 facilitates unwinding through system libraries on arm-android
2427 * Address description logic has been improved and is now common
2428 between Memcheck and Helgrind, resulting in better address
2429 descriptions for some kinds of error messages.
2431 * Error messages about dubious arguments (eg, to malloc or calloc) are
2432 output like other errors. This means that they can be suppressed
2433 and they have a stack trace.
2435 * The C++ demangler has been updated for better C++11 support.
2437 * New and modified GDB server monitor features:
2439 - Thread local variables/storage (__thread) can now be displayed.
2441 - The GDB server monitor command "v.info location <address>"
2442 displays information about an address. The information produced
2443 depends on the tool and on the options given to valgrind.
2444 Possibly, the following are described: global variables, local
2445 (stack) variables, allocated or freed blocks, ...
2447 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
2448 ask the GDB server to stop at the start of program execution, at
2449 the end of the program execution and on Valgrind internal errors.
2451 - A new monitor command "v.info stats" shows various Valgrind core
2452 and tool statistics.
2454 - A new monitor command "v.set hostvisibility" allows the GDB server
2455 to provide access to Valgrind internal host status/memory.
2457 * A new option "--aspace-minaddr=<address>" can in some situations
2458 allow the use of more memory by decreasing the address above which
2459 Valgrind maps memory. It can also be used to solve address
2460 conflicts with system libraries by increasing the default value.
2461 See user manual for details.
2463 * The amount of memory used by Valgrind to store debug info (unwind
2464 info, line number information and symbol data) has been
2465 significantly reduced, even though Valgrind now reads more
2466 information in order to support unwinding of inlined function calls.
2468 * Dwarf3 handling with --read-var-info=yes has been improved:
2470 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
2472 - Code compiled with
2473 -ffunction-sections -fdata-sections -Wl,--gc-sections
2474 no longer causes assertion failures.
2476 * Improved checking for the --sim-hints= and --kernel-variant=
2477 options. Unknown strings are now detected and reported to the user
2480 * The semantics of stack start/end boundaries in the valgrind.h
2481 VALGRIND_STACK_REGISTER client request has been clarified and
2482 documented. The convention is that start and end are respectively
2483 the lowest and highest addressable bytes of the stack.
2485 * ==================== FIXED BUGS ====================
2487 The following bugs have been fixed or resolved. Note that "n-i-bz"
2488 stands for "not in bugzilla" -- that is, a bug that was reported to us
2489 but never got a bugzilla entry. We encourage you to file bugs in
2490 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2491 than mailing the developers (or mailing lists) directly -- bugs that
2492 are not entered into bugzilla tend to get forgotten about or ignored.
2494 To see details of a given bug, visit
2495 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2496 where XXXXXX is the bug number as listed below.
2498 175819 Support for ipv6 socket reporting with --track-fds
2499 232510 make distcheck fails
2500 249435 Analyzing wine programs with callgrind triggers a crash
2501 278972 support for inlined function calls in stacktraces and suppression
2503 291310 FXSAVE instruction marks memory as undefined on amd64
2504 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2505 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2506 315199 vgcore file for threaded app does not show which thread crashed
2507 315952 tun/tap ioctls are not supported
2508 323178 Unhandled instruction: PLDW register (ARM)
2509 323179 Unhandled instruction: PLDW immediate (ARM)
2510 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2511 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2512 325124 [MIPSEL] Compilation error
2513 325477 Phase 4 support for IBM Power ISA 2.07
2514 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2515 325628 Phase 5 support for IBM Power ISA 2.07
2516 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2517 325751 Missing the two privileged Power PC Transactional Memory Instructions
2518 325816 Phase 6 support for IBM Power ISA 2.07
2519 325856 Make SGCheck fail gracefully on unsupported platforms
2520 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2521 326436 DRD: False positive in libstdc++ std::list::push_back
2522 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2523 326462 Refactor vgdb to isolate invoker stuff into separate module
2524 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2525 326623 DRD: false positive conflict report in a field assignment
2526 326724 Valgrind does not compile on OSX 1.9 Mavericks
2527 326816 Intercept for __strncpy_sse2_unaligned missing?
2528 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2529 326983 Clear direction flag after tests on amd64.
2530 327212 Do not prepend the current directory to absolute path names.
2531 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2532 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2533 327284 s390x: Fix translation of the risbg instruction
2534 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2535 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2536 327916 DW_TAG_typedef may have no name
2537 327943 s390x: add a redirection for the 'index' function
2538 328100 XABORT not implemented
2539 328205 Implement additional Xen hypercalls
2540 328454 add support Backtraces with ARM unwind tables (EXIDX)
2541 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2542 328711 valgrind.1 manpage "memcheck options" section is badly generated
2543 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2544 329612 Incorrect handling of AT_BASE for image execution
2545 329694 clang warns about using uninitialized variable
2546 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2547 330228 mmap must align to VKI_SHMLBA on mips32
2548 330257 LLVM does not support `-mno-dynamic-no-pic` option
2549 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2550 330459 --track-fds=yes doesn't track eventfds
2551 330469 Add clock_adjtime syscall support
2552 330594 Missing sysalls on PowerPC / uClibc
2553 330622 Add test to regression suite for POWER instruction: dcbzl
2554 330939 Support for AMD's syscall instruction on x86
2556 330941 Typo in PRE(poll) syscall wrapper
2557 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2558 331254 Fix expected output for memcheck/tests/dw4
2559 331255 Fix race condition in test none/tests/coolo_sigaction
2560 331257 Fix type of jump buffer in test none/tests/faultstatus
2561 331305 configure uses bash specific syntax
2562 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2563 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2564 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2565 331829 Unexpected ioctl opcode sign extension
2566 331830 ppc64: WARNING: unhandled syscall: 96/97
2567 331839 drd/tests/sem_open specifies invalid semaphore name
2568 331847 outcome of drd/tests/thread_name is nondeterministic
2569 332037 Valgrind cannot handle Thumb "add pc, reg"
2570 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2571 consistency checks enabled
2572 332263 intercepts for pthread_rwlock_timedrdlock and
2573 pthread_rwlock_timedwrlock are incorrect
2574 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2576 332276 Implement additional Xen hypercalls
2577 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2578 332765 Fix ms_print to create temporary files in a proper directory
2579 333072 drd: Add semaphore annotations
2580 333145 Tests for missaligned PC+#imm access for arm
2581 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2582 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2583 333248 WARNING: unhandled syscall: unix:443
2584 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2585 333501 cachegrind: assertion: Cache set count is not a power of two.
2588 333666 Recognize MPX instructions and bnd prefix.
2589 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2590 333817 Valgrind reports the memory areas written to by the SG_IO
2592 334049 lzcnt fails silently (x86_32)
2593 334384 Valgrind does not have support Little Endian support for
2595 334585 recvmmsg unhandled (+patch) (arm)
2596 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2597 334727 Build fails with -Werror=format-security
2598 334788 clarify doc about --log-file initial program directory
2599 334834 PPC64 Little Endian support, patch 2
2600 334836 PPC64 Little Endian support, patch 3 testcase fixes
2601 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2602 335034 Unhandled ioctl: HCIGETDEVLIST
2603 335155 vgdb, fix error print statement.
2604 335262 arm64: movi 8bit version is not supported
2605 335263 arm64: dmb instruction is not implemented
2606 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2607 335496 arm64: sbc/abc instructions are not implemented
2608 335554 arm64: unhandled instruction: abs
2609 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2610 335735 arm64: unhandled instruction: cnt
2611 335736 arm64: unhandled instruction: uaddlv
2612 335848 arm64: unhandled instruction: {s,u}cvtf
2613 335902 arm64: unhandled instruction: sli
2614 335903 arm64: unhandled instruction: umull (vector)
2615 336055 arm64: unhandled instruction: mov (element)
2616 336062 arm64: unhandled instruction: shrn{,2}
2617 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2618 336189 arm64: unhandled Instruction: mvn
2619 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2620 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2621 336772 Make moans about unknown ioctls more informative
2622 336957 Add a section about the Solaris/illumos port on the webpage
2623 337094 ifunc wrapper is broken on ppc64
2624 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2625 337528 leak check heuristic for block prefixed by length as 64bit number
2626 337740 Implement additional Xen hypercalls
2627 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2628 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2629 337871 deprecate --db-attach
2630 338023 Add support for all V4L2/media ioctls
2631 338024 inlined functions are not shown if DW_AT_ranges is used
2632 338106 Add support for 'kcmp' syscall
2633 338115 DRD: computed conflict set differs from actual after fork
2634 338160 implement display of thread local storage in gdbsrv
2635 338205 configure.ac and check for -Wno-tautological-compare
2636 338300 coredumps are missing one byte of every segment
2637 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2638 338499 --sim-hints parsing broken due to wrong order in tokens
2639 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2640 338681 Unable to unwind through clone thread created on i386-linux
2641 338698 race condition between gdbsrv and vgdb on startup
2642 338703 helgrind on arm-linux gets false positives in dynamic loader
2643 338791 alt dwz files can be relative of debug/main file
2644 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2645 338932 build V-trunk with gcc-trunk
2646 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2647 345079 Fix build problems in VEX/useful/test_main.c
2648 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2649 n-i-bz s390x: Fix memory corruption for multithreaded applications
2650 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2651 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2652 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2653 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2654 n-i-bz Add command-line option --default-suppressions
2655 n-i-bz Add support for BLKDISCARDZEROES ioctl
2656 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2657 n-i-bz Add support for sys_pivot_root and sys_unshare
2659 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2660 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2661 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2665 Release 3.9.0 (31 October 2013)
2666 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2667 3.9.0 is a feature release with many improvements and the usual
2668 collection of bug fixes.
2670 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2671 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2672 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2673 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2675 * ================== PLATFORM CHANGES =================
2677 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2678 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2680 * Support for MIPS DSP ASE on MIPS32 platforms.
2682 * Support for s390x Decimal Floating Point instructions on hosts that
2683 have the DFP facility installed.
2685 * Support for POWER8 (Power ISA 2.07) instructions
2687 * Support for Intel AVX2 instructions. This is available only on 64
2690 * Initial support for Intel Transactional Synchronization Extensions,
2693 * Initial support for Hardware Transactional Memory on POWER.
2695 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2696 run large GUI apps tolerably well.
2698 * ==================== TOOL CHANGES ====================
2702 - Improvements in handling of vectorised code, leading to
2703 significantly fewer false error reports. You need to use the flag
2704 --partial-loads-ok=yes to get the benefits of these changes.
2706 - Better control over the leak checker. It is now possible to
2707 specify which leak kinds (definite/indirect/possible/reachable)
2708 should be displayed, which should be regarded as errors, and which
2709 should be suppressed by a given leak suppression. This is done
2710 using the options --show-leak-kinds=kind1,kind2,..,
2711 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2712 "match-leak-kinds:" line in suppression entries, respectively.
2714 Note that generated leak suppressions contain this new line and
2715 are therefore more specific than in previous releases. To get the
2716 same behaviour as previous releases, remove the "match-leak-kinds:"
2717 line from generated suppressions before using them.
2719 - Reduced "possible leak" reports from the leak checker by the use
2720 of better heuristics. The available heuristics provide detection
2721 of valid interior pointers to std::stdstring, to new[] allocated
2722 arrays with elements having destructors and to interior pointers
2723 pointing to an inner part of a C++ object using multiple
2724 inheritance. They can be selected individually using the
2725 option --leak-check-heuristics=heur1,heur2,...
2727 - Better control of stacktrace acquisition for heap-allocated
2728 blocks. Using the --keep-stacktraces option, it is possible to
2729 control independently whether a stack trace is acquired for each
2730 allocation and deallocation. This can be used to create better
2731 "use after free" errors or to decrease Valgrind's resource
2732 consumption by recording less information.
2734 - Better reporting of leak suppression usage. The list of used
2735 suppressions (shown when the -v option is given) now shows, for
2736 each leak suppressions, how many blocks and bytes it suppressed
2737 during the last leak search.
2741 - False errors resulting from the use of statically initialised
2742 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2745 - False errors resulting from the use of pthread_cond_waits that
2746 timeout, have been removed.
2748 * ==================== OTHER CHANGES ====================
2750 * Some attempt to tune Valgrind's space requirements to the expected
2751 capabilities of the target:
2753 - The default size of the translation cache has been reduced from 8
2754 sectors to 6 on Android platforms, since each sector occupies
2755 about 40MB when using Memcheck.
2757 - The default size of the translation cache has been increased to 16
2758 sectors on all other platforms, reflecting the fact that large
2759 applications require instrumentation and storage of huge amounts
2760 of code. For similar reasons, the number of memory mapped
2761 segments that can be tracked has been increased by a factor of 6.
2763 - In all cases, the maximum number of sectors in the translation
2764 cache can be controlled by the new flag --num-transtab-sectors.
2766 * Changes in how debug info (line numbers, etc) is read:
2768 - Valgrind no longer temporarily mmaps the entire object to read
2769 from it. Instead, reading is done through a small fixed sized
2770 buffer. This avoids virtual memory usage spikes when Valgrind
2771 reads debuginfo from large shared objects.
2773 - A new experimental remote debug info server. Valgrind can read
2774 debug info from a different machine (typically, a build host)
2775 where debuginfo objects are stored. This can save a lot of time
2776 and hassle when running Valgrind on resource-constrained targets
2777 (phones, tablets) when the full debuginfo objects are stored
2778 somewhere else. This is enabled by the --debuginfo-server=
2781 - Consistency checking between main and debug objects can be
2782 disabled using the --allow-mismatched-debuginfo option.
2784 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2785 scanning can recover stack traces in some cases when the normal
2786 unwind mechanisms fail. Stack scanning is best described as "a
2787 nasty, dangerous and misleading hack" and so is disabled by default.
2788 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2791 * Detection and merging of recursive stack frame cycles. When your
2792 program has recursive algorithms, this limits the memory used by
2793 Valgrind for recorded stack traces and avoids recording
2794 uninteresting repeated calls. This is controlled by the command
2795 line option --merge-recursive-frame and by the monitor command
2796 "v.set merge-recursive-frames".
2798 * File name and line numbers for used suppressions. The list of used
2799 suppressions (shown when the -v option is given) now shows, for each
2800 used suppression, the file name and line number where the suppression
2803 * New and modified GDB server monitor features:
2805 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2806 that can be used to execute gdbserver monitor commands from the
2809 - A new monitor command, "v.info open_fds", that gives the list of
2810 open file descriptors and additional details.
2812 - An optional message in the "v.info n_errs_found" monitor command,
2813 for example "v.info n_errs_found test 1234 finished", allowing a
2814 comment string to be added to the process output, perhaps for the
2815 purpose of separating errors of different tests or test phases.
2817 - A new monitor command "v.info execontext" that shows information
2818 about the stack traces recorded by Valgrind.
2820 - A new monitor command "v.do expensive_sanity_check_general" to run
2821 some internal consistency checks.
2823 * New flag --sigill-diagnostics to control whether a diagnostic
2824 message is printed when the JIT encounters an instruction it can't
2825 translate. The actual behavior -- delivery of SIGILL to the
2826 application -- is unchanged.
2828 * The maximum amount of memory that Valgrind can use on 64 bit targets
2829 has been increased from 32GB to 64GB. This should make it possible
2830 to run applications on Memcheck that natively require up to about 35GB.
2832 * ==================== FIXED BUGS ====================
2834 The following bugs have been fixed or resolved. Note that "n-i-bz"
2835 stands for "not in bugzilla" -- that is, a bug that was reported to us
2836 but never got a bugzilla entry. We encourage you to file bugs in
2837 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2838 than mailing the developers (or mailing lists) directly -- bugs that
2839 are not entered into bugzilla tend to get forgotten about or ignored.
2841 To see details of a given bug, visit
2842 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2843 where XXXXXX is the bug number as listed below.
2845 123837 system call: 4th argument is optional, depending on cmd
2846 135425 memcheck should tell you where Freed blocks were Mallocd
2847 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2848 207815 Adds some of the drm ioctls to syswrap-linux.c
2849 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2850 252955 Impossible to compile with ccache
2851 253519 Memcheck reports auxv pointer accesses as invalid reads.
2852 263034 Crash when loading some PPC64 binaries
2853 269599 Increase deepest backtrace
2854 274695 s390x: Support "compare to/from logical" instructions (z196)
2855 275800 s390x: Autodetect cache info (part 2)
2856 280271 Valgrind reports possible memory leaks on still-reachable std::string
2857 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2858 289578 Backtraces with ARM unwind tables (stack scan flags)
2859 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2860 304832 ppc32: build failure
2861 305431 Use find_buildid shdr fallback for separate .debug files
2862 305728 Add support for AVX2 instructions
2863 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2864 306035 s390x: Fix IR generation for LAAG and friends
2865 306054 s390x: Condition code computation for convert-to-int/logical
2866 306098 s390x: alternate opcode form for convert to/from fixed
2867 306587 Fix cache line detection from auxiliary vector for PPC.
2868 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2869 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2870 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2871 307101 sys_capget second argument can be NULL
2872 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2873 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2874 307113 s390x: DFP support
2875 307141 valgrind does't work in mips-linux system
2876 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2877 307285 x86_amd64 feature test for avx in test suite is wrong
2878 307290 memcheck overlap testcase needs memcpy version filter
2879 307463 Please add "&limit=0" to the "all open bugs" link
2880 307465 --show-possibly-lost=no should reduce the error count / exit code
2881 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2882 307729 pkgconfig support broken valgrind.pc
2883 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2884 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2885 308089 Unhandled syscall on ppc64: prctl
2886 308135 PPC32 MPC8xx has 16 bytes cache size
2887 308321 testsuite memcheck filter interferes with gdb_filter
2889 308341 vgdb should report process exit (or fatal signal)
2890 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2891 308495 Remove build dependency on installed Xen headers
2892 308573 Internal error on 64-bit instruction executed in 32-bit mode
2894 308627 pmovmskb validity bit propagation is imprecise
2895 308644 vgdb command for having the info for the track-fds option
2896 308711 give more info about aspacemgr and arenas in out_of_memory
2897 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2898 308718 ARM implement SMLALBB family of instructions
2899 308886 Missing support for PTRACE_SET/GETREGSET
2900 308930 syscall name_to_handle_at (303 on amd64) not handled
2901 309229 V-bit tester does not report number of tests generated
2902 309323 print unrecognized instuction on MIPS
2903 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2904 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2905 309430 Self hosting ppc64 encounters a vassert error on operand type
2906 309600 valgrind is a bit confused about 0-sized sections
2907 309823 Generate errors for still reachable blocks
2908 309921 PCMPISTRI validity bit propagation is imprecise
2909 309922 none/tests/ppc64/test_dfp5 sometimes fails
2910 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2911 310424 --read-var-info does not properly describe static variables
2912 310792 search additional path for debug symbols
2913 310931 s390x: Message-security assist (MSA) instruction extension [..]
2914 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2915 311318 ARM: "128-bit constant is not implemented" error message
2916 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2917 311690 V crashes because it redirects branches inside of a redirected function
2918 311880 x86_64: make regtest hangs at shell_valid1
2919 311922 WARNING: unhandled syscall: 170
2921 312171 ppc: insn selection for DFP
2922 312571 Rounding mode call wrong for the DFP Iops [..]
2923 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2924 312913 Dangling pointers error should also report the alloc stack trace
2925 312980 Building on Mountain Lion generates some compiler warnings
2926 313267 Adding MIPS64/Linux port to Valgrind
2929 313811 Buffer overflow in assert_fail
2930 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2931 314269 ppc: dead code in insn selection
2932 314718 ARM: implement integer divide instruction (sdiv and udiv)
2933 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2934 315441 sendmsg syscall should ignore unset msghdr msg_flags
2935 315534 msgrcv inside a thread causes valgrind to hang (block)
2936 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2937 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2938 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2939 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2940 316144 valgrind.1 manpage contains unknown ??? strings [..]
2941 316145 callgrind command line options in manpage reference (unknown) [..]
2942 316145 callgrind command line options in manpage reference [..]
2943 316181 drd: Fixed a 4x slowdown for certain applications
2944 316503 Valgrind does not support SSE4 "movntdqa" instruction
2945 316535 Use of |signed int| instead of |size_t| in valgrind messages
2946 316696 fluidanimate program of parsec 2.1 stuck
2947 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2948 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2949 317186 "Impossible happens" when occurs VCVT instruction on ARM
2950 317318 Support for Threading Building Blocks "scalable_malloc"
2951 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2952 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2953 317463 bmi testcase IR SANITY CHECK FAILURE
2954 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2955 318050 libmpiwrap fails to compile with out-of-source build
2956 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2957 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2958 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2959 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2960 318932 Add missing PPC64 and PPC32 system call support
2961 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2962 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2963 319494 VEX Makefile-gcc standalone build update after r2702
2964 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2965 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2966 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2967 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2968 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2969 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2970 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2971 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2972 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2973 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2974 320895 add fanotify support (patch included)
2975 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2976 321065 Valgrind updates for Xen 4.3
2977 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2978 321363 Unhandled instruction: SSAX (ARM + Thumb)
2979 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2980 321466 Unhandled instruction: SHASX (ARM + Thumb)
2981 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2982 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2983 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2984 321620 Unhandled instruction: UASX (ARM + Thumb)
2985 321621 Unhandled instruction: USAX (ARM + Thumb)
2986 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2987 321693 Unhandled instruction: LDRSBT (Thumb)
2988 321694 Unhandled instruction: UQASX (ARM + Thumb)
2989 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2990 321697 Unhandled instruction: UHASX (ARM + Thumb)
2991 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2992 321704 Unhandled instruction: REVSH (ARM + Thumb)
2993 321730 Add cg_diff and cg_merge man pages
2994 321738 Add vgdb and valgrind-listener man pages
2996 321891 Unhandled instruction: LDRHT (Thumb)
2997 321960 pthread_create() then alloca() causing invalid stack write errors
2998 321969 ppc32 and ppc64 don't support [lf]setxattr
2999 322254 Show threadname together with tid if set by application
3000 322294 Add initial support for IBM Power ISA 2.07
3001 322368 Assertion failure in wqthread_hijack under OS X 10.8
3002 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
3003 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
3004 322851 0bXXX binary literal syntax is not standard
3005 323035 Unhandled instruction: LDRSHT(Thumb)
3006 323036 Unhandled instruction: SMMLS (ARM and Thumb)
3007 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
3008 323175 Unhandled instruction: SMLALD (ARM + Thumb)
3009 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
3010 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
3011 323437 Phase 2 support for IBM Power ISA 2.07
3012 323713 Support mmxext (integer sse) subset on i386 (athlon)
3013 323803 Transactional memory instructions are not supported for Power
3014 323893 SSE3 not available on amd cpus in valgrind
3015 323905 Probable false positive from Valgrind/drd on close()
3016 323912 valgrind.h header isn't compatible for mingw64
3017 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
3018 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
3019 324181 mmap does not handle MAP_32BIT
3020 324227 memcheck false positive leak when a thread calls exit+block [..]
3021 324421 Support for fanotify API on ARM architecture
3022 324514 gdbserver monitor cmd output behaviour consistency [..]
3023 324518 ppc64: Emulation of dcbt instructions does not handle [..]
3024 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
3025 324582 When access is made to freed memory, report both allocation [..]
3026 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
3027 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
3028 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
3029 324834 Unhandled instructions in Microsoft C run-time for x86_64
3030 324894 Phase 3 support for IBM Power ISA 2.07
3031 326091 drd: Avoid false race reports from optimized strlen() impls
3032 326113 valgrind libvex hwcaps error on AMD64
3033 n-i-bz Some wrong command line options could be ignored
3034 n-i-bz patch to allow fair-sched on android
3035 n-i-bz report error for vgdb snapshot requested before execution
3036 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
3038 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
3042 Release 3.8.1 (19 September 2012)
3043 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3044 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
3045 that occur moderately frequently in real use cases, adds support for
3046 some missing instructions on ARM, and fixes a deadlock condition on
3047 MacOSX. If you package or deliver 3.8.0 for others to use, you might
3048 want to consider upgrading to 3.8.1 instead.
3050 The following bugs have been fixed or resolved. Note that "n-i-bz"
3051 stands for "not in bugzilla" -- that is, a bug that was reported to us
3052 but never got a bugzilla entry. We encourage you to file bugs in
3053 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3054 than mailing the developers (or mailing lists) directly -- bugs that
3055 are not entered into bugzilla tend to get forgotten about or ignored.
3057 To see details of a given bug, visit
3058 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3059 where XXXXXX is the bug number as listed below.
3062 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
3063 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
3064 298281 wcslen causes false(?) uninitialised value warnings
3065 301281 valgrind hangs on OS X when the process calls system()
3066 304035 disInstr(arm): unhandled instruction 0xE1023053
3067 304867 implement MOVBE instruction in x86 mode
3068 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
3069 305042 amd64: implement 0F 7F encoding of movq between two registers
3070 305199 ARM: implement QDADD and QDSUB
3071 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
3072 305513 killed by fatal signal: SIGSEGV
3073 305690 DRD reporting invalid semaphore when sem_trywait fails
3074 305926 Invalid alignment checks for some AVX instructions
3075 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
3076 306310 3.8.0 release tarball missing some files
3077 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
3078 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
3079 n-i-bz shmat of a segment > 4Gb does not work
3080 n-i-bz simulate_control_c script wrong USR1 signal number on mips
3081 n-i-bz vgdb ptrace calls wrong on mips [...]
3082 n-i-bz Fixes for more MPI false positives
3083 n-i-bz exp-sgcheck's memcpy causes programs to segfault
3084 n-i-bz OSX build w/ clang: asserts at startup
3085 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
3086 n-i-bz fix a couple of union tag-vs-field mixups
3087 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
3089 The following bugs were fixed in 3.8.0 but not listed in this NEWS
3092 254088 Valgrind should know about UD2 instruction
3095 304754 NEWS blows TeX's little mind
3097 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
3101 Release 3.8.0 (10 August 2012)
3102 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3103 3.8.0 is a feature release with many improvements and the usual
3104 collection of bug fixes.
3106 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3107 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
3108 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
3109 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
3110 There is initial support for MacOSX 10.8, but it is not usable for
3111 serious work at present.
3113 * ================== PLATFORM CHANGES =================
3115 * Support for MIPS32 platforms running Linux. Valgrind has been
3116 tested on MIPS32 and MIPS32r2 platforms running different Debian
3117 Squeeze and MeeGo distributions. Both little-endian and big-endian
3118 cores are supported. The tools Memcheck, Massif and Lackey have
3119 been tested and are known to work. See README.mips for more details.
3121 * Preliminary support for Android running on x86.
3123 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
3125 * Support for Intel AVX instructions and for AES instructions. This
3126 support is available only for 64 bit code.
3128 * Support for POWER Decimal Floating Point instructions.
3130 * ==================== TOOL CHANGES ====================
3132 * Non-libc malloc implementations are now supported. This is useful
3133 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
3134 Using the new option --soname-synonyms, such tools can be informed
3135 that the malloc implementation is either linked statically into the
3136 executable, or is present in some other shared library different
3137 from libc.so. This makes it possible to process statically linked
3138 programs, and programs using other malloc libraries, for example
3139 TCMalloc or JEMalloc.
3141 * For tools that provide their own replacement for malloc et al, the
3142 option --redzone-size=<number> allows users to specify the size of
3143 the padding blocks (redzones) added before and after each client
3144 allocated block. Smaller redzones decrease the memory needed by
3145 Valgrind. Bigger redzones increase the chance to detect blocks
3146 overrun or underrun. Prior to this change, the redzone size was
3147 hardwired to 16 bytes in Memcheck.
3151 - The leak_check GDB server monitor command now can
3152 control the maximum nr of loss records to output.
3154 - Reduction of memory use for applications allocating
3155 many blocks and/or having many partially defined bytes.
3157 - Addition of GDB server monitor command 'block_list' that lists
3158 the addresses/sizes of the blocks of a leak search loss record.
3160 - Addition of GDB server monitor command 'who_points_at' that lists
3161 the locations pointing at a block.
3163 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
3164 detect an invalid access of these redzones, by marking them
3165 noaccess. Similarly, if a redzone size is given for a memory
3166 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
3167 This still allows to find some bugs if the user has forgotten to
3168 mark the pool superblock noaccess.
3170 - Performance of memory leak check has been improved, especially in
3171 cases where there are many leaked blocks and/or many suppression
3172 rules used to suppress leak reports.
3174 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
3175 more precise analysis, which is important for LLVM/Clang
3176 generated code. This is at the cost of somewhat reduced
3177 performance. Note there is no change to analysis precision or
3178 costs on Linux targets.
3182 - Added even more facilities that can help finding the cause of a data
3183 race, namely the command-line option --ptrace-addr and the macro
3184 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
3186 - Fixed a subtle bug that could cause false positive data race reports.
3188 * ==================== OTHER CHANGES ====================
3190 * The C++ demangler has been updated so as to work well with C++
3191 compiled by up to at least g++ 4.6.
3193 * Tool developers can make replacement/wrapping more flexible thanks
3194 to the new option --soname-synonyms. This was reported above, but
3195 in fact is very general and applies to all function
3196 replacement/wrapping, not just to malloc-family functions.
3198 * Round-robin scheduling of threads can be selected, using the new
3199 option --fair-sched= yes. Prior to this change, the pipe-based
3200 thread serialisation mechanism (which is still the default) could
3201 give very unfair scheduling. --fair-sched=yes improves
3202 responsiveness of interactive multithreaded applications, and
3203 improves repeatability of results from the thread checkers Helgrind
3206 * For tool developers: support to run Valgrind on Valgrind has been
3207 improved. We can now routinely Valgrind on Helgrind or Memcheck.
3209 * gdbserver now shows the float shadow registers as integer
3210 rather than float values, as the shadow values are mostly
3211 used as bit patterns.
3213 * Increased limit for the --num-callers command line flag to 500.
3215 * Performance improvements for error matching when there are many
3216 suppression records in use.
3218 * Improved support for DWARF4 debugging information (bug 284184).
3220 * Initial support for DWZ compressed Dwarf debug info.
3222 * Improved control over the IR optimiser's handling of the tradeoff
3223 between performance and precision of exceptions. Specifically,
3224 --vex-iropt-precise-memory-exns has been removed and replaced by
3225 --vex-iropt-register-updates, with extended functionality. This
3226 allows the Valgrind gdbserver to always show up to date register
3229 * Modest performance gains through the use of translation chaining for
3232 * ==================== FIXED BUGS ====================
3234 The following bugs have been fixed or resolved. Note that "n-i-bz"
3235 stands for "not in bugzilla" -- that is, a bug that was reported to us
3236 but never got a bugzilla entry. We encourage you to file bugs in
3237 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3238 than mailing the developers (or mailing lists) directly -- bugs that
3239 are not entered into bugzilla tend to get forgotten about or ignored.
3241 To see details of a given bug, visit
3242 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3243 where XXXXXX is the bug number as listed below.
3245 197914 Building valgrind from svn now requires automake-1.10
3246 203877 increase to 16Mb maximum allowed alignment for memalign et al
3247 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
3248 247386 make perf does not run all performance tests
3249 270006 Valgrind scheduler unfair
3250 270777 Adding MIPS/Linux port to Valgrind
3251 270796 s390x: Removed broken support for the TS insn
3252 271438 Fix configure for proper SSE4.2 detection
3253 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
3254 273475 Add support for AVX instructions
3255 274078 improved configure logic for mpicc
3256 276993 fix mremap 'no thrash checks'
3257 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
3258 281482 memcheck incorrect byte allocation count in realloc() for silly argument
3259 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
3260 283413 Fix wrong sanity check
3261 283671 Robustize alignment computation in LibVEX_Alloc
3262 283961 Adding support for some HCI IOCTLs
3263 284124 parse_type_DIE: confused by: DWARF 4
3264 284864 == 273475 (Add support for AVX instructions)
3265 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
3266 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
3267 285725 == 273475 (Add support for AVX instructions)
3268 286261 add wrapper for linux I2C_RDWR ioctl
3269 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
3270 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
3271 286384 configure fails "checking for a supported version of gcc"
3272 286497 == 273475 (Add support for AVX instructions)
3273 286596 == 273475 (Add support for AVX instructions)
3274 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
3275 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
3276 287260 Incorrect conditional jump or move depends on uninitialised value(s)
3277 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
3278 287307 == 273475 (Add support for AVX instructions)
3279 287858 VG_(strerror): unknown error
3280 288298 (MacOSX) unhandled syscall shm_unlink
3281 288995 == 273475 (Add support for AVX instructions)
3282 289470 Loading of large Mach-O thin binaries fails.
3283 289656 == 273475 (Add support for AVX instructions)
3284 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
3285 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
3286 289839 s390x: Provide support for unicode conversion instructions
3287 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
3288 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
3289 290655 Add support for AESKEYGENASSIST instruction
3290 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
3291 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
3292 291253 ES register not initialised in valgrind simulation
3293 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
3294 291865 s390x: Support the "Compare Double and Swap" family of instructions
3295 292300 == 273475 (Add support for AVX instructions)
3296 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
3297 292493 == 273475 (Add support for AVX instructions)
3298 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
3299 292627 Missing support for some SCSI ioctls
3300 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
3301 292841 == 273475 (Add support for AVX instructions)
3302 292993 implement the getcpu syscall on amd64-linux
3303 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
3304 293088 Add some VEX sanity checks for ppc64 unhandled instructions
3305 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
3306 293754 PCMPxSTRx not implemented for 16-bit characters
3307 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
3308 293808 CLFLUSH not supported by latest VEX for amd64
3309 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
3310 294048 MPSADBW instruction not implemented
3311 294055 regtest none/tests/shell fails when locale is not set to C
3312 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
3313 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
3314 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
3315 294260 disInstr_AMD64: disInstr miscalculated next %rip
3316 294523 --partial-loads-ok=yes causes false negatives
3317 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
3318 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
3319 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
3320 295089 can not annotate source for both helgrind and drd
3321 295221 POWER Processor decimal floating point instruction support missing
3322 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
3323 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
3324 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
3325 295617 ARM - Add some missing syscalls
3326 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
3327 296229 Linux user input device ioctls missing wrappers
3328 296318 ELF Debug info improvements (more than one rx/rw mapping)
3329 296422 Add translation chaining support
3330 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
3331 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
3332 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
3333 297078 gdbserver signal handling problems [..]
3334 297147 drd false positives on newly allocated memory
3335 297329 disallow decoding of IBM Power DFP insns on some machines
3336 297497 POWER Processor decimal floating point instruction support missing
3337 297701 Another alias for strncasecmp_l in libc-2.13.so
3338 297911 'invalid write' not reported when using APIs for custom mem allocators.
3339 297976 s390x: revisit EX implementation
3340 297991 Valgrind interferes with mmap()+ftell()
3341 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
3342 297993 Fix compilation of valgrind with gcc -g3.
3343 298080 POWER Processor DFP support missing, part 3
3344 298227 == 273475 (Add support for AVX instructions)
3345 298335 == 273475 (Add support for AVX instructions)
3346 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
3347 298394 s390x: Don't bail out on an unknown machine model. [..]
3348 298421 accept4() syscall (366) support is missing for ARM
3349 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
3350 298732 valgrind installation problem in ubuntu with kernel version 3.x
3351 298862 POWER Processor DFP instruction support missing, part 4
3352 298864 DWARF reader mis-parses DW_FORM_ref_addr
3353 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
3354 299053 Support DWARF4 DW_AT_high_pc constant form
3355 299104 == 273475 (Add support for AVX instructions)
3356 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
3357 299629 dup3() syscall (358) support is missing for ARM
3358 299694 POWER Processor DFP instruction support missing, part 5
3359 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
3360 299803 == 273475 (Add support for AVX instructions)
3361 299804 == 273475 (Add support for AVX instructions)
3362 299805 == 273475 (Add support for AVX instructions)
3363 300140 ARM - Missing (T1) SMMUL
3364 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
3365 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
3366 300414 FCOM and FCOMP unimplemented for amd64 guest
3367 301204 infinite loop in canonicaliseSymtab with ifunc symbol
3368 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
3369 301265 add x86 support to Android build
3370 301984 configure script doesn't detect certain versions of clang
3371 302205 Fix compiler warnings for POWER VEX code and POWER test cases
3372 302287 Unhandled movbe instruction on Atom processors
3373 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
3374 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
3375 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
3376 302656 == 273475 (Add support for AVX instructions)
3377 302709 valgrind for ARM needs extra tls support for android emulator [..]
3378 302827 add wrapper for CDROM_GET_CAPABILITY
3379 302901 Valgrind crashes with dwz optimized debuginfo
3380 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
3381 303116 Add support for the POWER instruction popcntb
3382 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
3383 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
3384 303466 == 273475 (Add support for AVX instructions)
3385 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
3386 303963 strstr() function produces wrong results under valgrind callgrind
3387 304054 CALL_FN_xx macros need to enforce stack alignment
3388 304561 tee system call not supported
3389 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
3390 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
3391 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
3392 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
3393 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
3394 n-i-bz s390x: Shadow registers can now be examined using vgdb
3396 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
3397 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
3401 Release 3.7.0 (5 November 2011)
3402 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3403 3.7.0 is a feature release with many significant improvements and the
3404 usual collection of bug fixes.
3406 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3407 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
3408 Support for recent distros and toolchain components (glibc 2.14, gcc
3409 4.6, MacOSX 10.7) has been added.
3411 * ================== PLATFORM CHANGES =================
3413 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
3414 analyse 64-bit programs running on z/Architecture. Most user space
3415 instructions up to and including z10 are supported. Valgrind has
3416 been tested extensively on z9, z10, and z196 machines running SLES
3417 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
3418 known to work well. Callgrind, Helgrind, and DRD work reasonably
3419 well on z9 and later models. See README.s390 for more details.
3421 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
3422 64-bit processes are supported. Some complex threaded applications
3423 (Firefox) are observed to hang when run as 32 bit applications,
3424 whereas 64-bit versions run OK. The cause is unknown. Memcheck
3425 will likely report some false errors. In general, expect some rough
3426 spots. This release also supports MacOSX 10.6, but drops support
3429 * Preliminary support for Android (on ARM). Valgrind can now run
3430 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
3431 README.android for more details, plus instructions on how to get
3434 * Support for the IBM Power ISA 2.06 (Power7 instructions)
3436 * General correctness and performance improvements for ARM/Linux, and,
3437 by extension, ARM/Android.
3439 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
3440 instruction set support is under development but is not available in
3443 * Support for AIX5 has been removed.
3445 * ==================== TOOL CHANGES ====================
3447 * Memcheck: some incremental changes:
3449 - reduction of memory use in some circumstances
3451 - improved handling of freed memory, which in some circumstances
3452 can cause detection of use-after-free that would previously have
3455 - fix of a longstanding bug that could cause false negatives (missed
3456 errors) in programs doing vector saturated narrowing instructions.
3458 * Helgrind: performance improvements and major memory use reductions,
3459 particularly for large, long running applications which perform many
3460 synchronisation (lock, unlock, etc) events. Plus many smaller
3463 - display of locksets for both threads involved in a race
3465 - general improvements in formatting/clarity of error messages
3467 - addition of facilities and documentation regarding annotation
3468 of thread safe reference counted C++ classes
3470 - new flag --check-stack-refs=no|yes [yes], to disable race checking
3471 on thread stacks (a performance hack)
3473 - new flag --free-is-write=no|yes [no], to enable detection of races
3474 where one thread accesses heap memory but another one frees it,
3475 without any coordinating synchronisation event
3477 * DRD: enabled XML output; added support for delayed thread deletion
3478 in order to detect races that occur close to the end of a thread
3479 (--join-list-vol); fixed a memory leak triggered by repeated client
3480 memory allocatation and deallocation; improved Darwin support.
3482 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
3484 * exp-sgcheck: this tool has been reduced in scope so as to improve
3485 performance and remove checking that Memcheck does better.
3486 Specifically, the ability to check for overruns for stack and global
3487 arrays is unchanged, but the ability to check for overruns of heap
3488 blocks has been removed. The tool has accordingly been renamed to
3489 exp-sgcheck ("Stack and Global Array Checking").
3491 * ==================== OTHER CHANGES ====================
3493 * GDB server: Valgrind now has an embedded GDB server. That means it
3494 is possible to control a Valgrind run from GDB, doing all the usual
3495 things that GDB can do (single stepping, breakpoints, examining
3496 data, etc). Tool-specific functionality is also available. For
3497 example, it is possible to query the definedness state of variables
3498 or memory from within GDB when running Memcheck; arbitrarily large
3499 memory watchpoints are supported, etc. To use the GDB server, start
3500 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3503 * Improved support for unfriendly self-modifying code: a new option
3504 --smc-check=all-non-file is available. This adds the relevant
3505 consistency checks only to code that originates in non-file-backed
3506 mappings. In effect this confines the consistency checking only to
3507 code that is or might be JIT generated, and avoids checks on code
3508 that must have been compiled ahead of time. This significantly
3509 improves performance on applications that generate code at run time.
3511 * It is now possible to build a working Valgrind using Clang-2.9 on
3514 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3515 These enable and disable error reporting on a per-thread, and
3516 nestable, basis. This is useful for hiding errors in particularly
3517 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3518 now uses this facility.
3520 * Added the --mod-funcname option to cg_diff.
3522 * ==================== FIXED BUGS ====================
3524 The following bugs have been fixed or resolved. Note that "n-i-bz"
3525 stands for "not in bugzilla" -- that is, a bug that was reported to us
3526 but never got a bugzilla entry. We encourage you to file bugs in
3527 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3528 mailing the developers (or mailing lists) directly -- bugs that are
3529 not entered into bugzilla tend to get forgotten about or ignored.
3531 To see details of a given bug, visit
3532 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3533 where XXXXXX is the bug number as listed below.
3535 79311 malloc silly arg warning does not give stack trace
3536 210935 port valgrind.h (not valgrind) to win32 to support client requests
3537 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3538 243404 Port to zSeries
3539 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3540 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3541 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3542 253206 Some fixes for the faultstatus testcase
3543 255223 capget testcase fails when running as root
3544 256703 xlc_dbl_u32.c testcase broken
3545 256726 Helgrind tests have broken inline asm
3546 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3547 264800 testcase compile failure on zseries
3548 265762 make public VEX headers compilable by G++ 3.x
3549 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3550 266753 configure script does not give the user the option to not use QtCore
3551 266931 gen_insn_test.pl is broken
3552 266961 ld-linux.so.2 i?86-linux strlen issues
3553 266990 setns instruction causes false positive
3554 267020 Make directory for temporary files configurable at run-time.
3555 267342 == 267997 (segmentation fault on Mac OS 10.6)
3556 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3557 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3558 267488 regtest: darwin support for 64-bit build
3559 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3560 267630 Add support for IBM Power ISA 2.06 -- stage 1
3561 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3562 267819 Add client request for informing the core about reallocation
3563 267925 laog data structure quadratic for a single sequence of lock
3564 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3565 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3566 268513 missed optimizations in fold_Expr
3567 268619 s390x: fpr - gpr transfer facility
3568 268620 s390x: reconsider "long displacement" requirement
3569 268621 s390x: improve IR generation for XC
3570 268715 s390x: FLOGR is not universally available
3571 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3572 268930 s390x: MHY is not universally available
3573 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3574 269079 Support ptrace system call on ARM
3575 269144 missing "Bad option" error message
3576 269209 conditional load and store facility (z196)
3577 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3578 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3579 269736 s390x: minor code generation tweaks
3580 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3581 269863 s390x: remove unused function parameters
3582 269864 s390x: tweak s390_emit_load_cc
3583 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3584 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3585 270115 s390x: rewrite some testcases
3586 270309 == 267997 (valgrind crash on startup)
3587 270320 add support for Linux FIOQSIZE ioctl() call
3588 270326 segfault while trying to sanitize the environment passed to execle
3589 270794 IBM POWER7 support patch causes regression in none/tests
3590 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3591 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3592 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3593 270959 s390x: invalid use of R0 as base register
3594 271042 VSX configure check fails when it should not
3595 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3596 271259 s390x: fix code confusion
3597 271337 == 267997 (Valgrind segfaults on MacOS X)
3598 271385 s390x: Implement Ist_MBE
3599 271501 s390x: misc cleanups
3600 271504 s390x: promote likely and unlikely
3601 271579 ppc: using wrong enum type
3602 271615 unhandled instruction "popcnt" (arch=amd10h)
3603 271730 Fix bug when checking ioctls: duplicate check
3604 271776 s390x: provide STFLE instruction support
3605 271779 s390x: provide clock instructions like STCK
3606 271799 Darwin: ioctls without an arg report a memory error
3607 271820 arm: fix type confusion
3608 271917 pthread_cond_timedwait failure leads to not-locked false positive
3609 272067 s390x: fix DISP20 macro
3610 272615 A typo in debug output in mc_leakcheck.c
3611 272661 callgrind_annotate chokes when run from paths containing regex chars
3612 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3613 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3614 272967 make documentation build-system more robust
3615 272986 Fix gcc-4.6 warnings with valgrind.h
3616 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3617 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3618 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3619 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3620 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3621 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3622 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3623 273778 exp-ptrcheck: unhandled sysno == 259
3624 274089 exp-ptrcheck: unhandled sysno == 208
3625 274378 s390x: Various dispatcher tweaks
3626 274447 WARNING: unhandled syscall: 340
3627 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3628 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3629 274926 valgrind does not build against linux-3
3630 275148 configure FAIL with glibc-2.14
3631 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3632 275168 Make Valgrind work for MacOSX 10.7 Lion
3633 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3634 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3635 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3636 275308 Fix implementation for ppc64 fres instruc
3637 275339 s390x: fix testcase compile warnings
3638 275517 s390x: Provide support for CKSM instruction
3639 275710 s390x: get rid of redundant address mode calculation
3640 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3641 275852 == 250101 (valgrind uses all swap space and is killed)
3642 276784 Add support for IBM Power ISA 2.06 -- stage 3
3643 276987 gdbsrv: fix tests following recent commits
3644 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3645 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3646 277471 Unhandled syscall: 340
3647 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3648 277653 ARM: support Thumb2 PLD instruction
3649 277663 ARM: NEON float VMUL by scalar incorrect
3650 277689 ARM: tests for VSTn with register post-index are broken
3651 277694 ARM: BLX LR instruction broken in ARM mode
3652 277780 ARM: VMOV.F32 (immediate) instruction is broken
3653 278057 fuse filesystem syscall deadlocks
3654 278078 Unimplemented syscall 280 on ppc32
3655 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3656 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3657 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3658 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3659 279027 Support for MVCL and CLCL instruction
3660 279027 s390x: Provide support for CLCL and MVCL instructions
3661 279062 Remove a redundant check in the insn selector for ppc.
3662 279071 JDK creates PTEST with redundant REX.W prefix
3663 279212 gdbsrv: add monitor cmd v.info scheduler.
3664 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3665 279698 memcheck discards valid-bits for packuswb
3666 279795 memcheck reports uninitialised values for mincore on amd64
3667 279994 Add support for IBM Power ISA 2.06 -- stage 3
3668 280083 mempolicy syscall check errors
3669 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3670 280710 s390x: config files for nightly builds
3671 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3672 280965 Valgrind breaks fcntl locks when program does mmap
3673 281138 WARNING: unhandled syscall: 340
3674 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3675 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3676 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3677 281468 s390x: handle do_clone and gcc clones in call traces
3678 281488 ARM: VFP register corruption
3679 281828 == 275284 (false memmove warning: "Source and destination overlap")
3680 281883 s390x: Fix system call wrapper for "clone".
3681 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3682 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3683 282238 SLES10: make check fails
3684 282979 strcasestr needs replacement with recent(>=2.12) glibc
3685 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3686 283243 Regression in ppc64 memcheck tests
3687 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3688 283427 re-connect epoll_pwait syscall on ARM linux
3689 283600 gdbsrv: android: port vgdb.c
3690 283709 none/tests/faultstatus needs to account for page size
3691 284305 filter_gdb needs enhancement to work on ppc64
3692 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3693 284472 Thumb2 ROR.W encoding T2 not implemented
3694 284621 XML-escape process command line in XML output
3695 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3696 that have non-power-of-2 sizes (also AMDs)
3697 n-i-bz don't be spooked by libraries mashed by elfhack
3698 n-i-bz don't be spooked by libxul.so linked with gold
3699 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3701 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3702 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3703 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3707 Release 3.6.1 (16 February 2011)
3708 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3709 3.6.1 is a bug fix release. It adds support for some SSE4
3710 instructions that were omitted in 3.6.0 due to lack of time. Initial
3711 support for glibc-2.13 has been added. A number of bugs causing
3712 crashing or assertion failures have been fixed.
3714 The following bugs have been fixed or resolved. Note that "n-i-bz"
3715 stands for "not in bugzilla" -- that is, a bug that was reported to us
3716 but never got a bugzilla entry. We encourage you to file bugs in
3717 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3718 mailing the developers (or mailing lists) directly -- bugs that are
3719 not entered into bugzilla tend to get forgotten about or ignored.
3721 To see details of a given bug, visit
3722 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3723 where XXXXXX is the bug number as listed below.
3725 188572 Valgrind on Mac should suppress setenv() mem leak
3726 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3727 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3728 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3729 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3730 254420 memory pool tracking broken
3731 254957 Test code failing to compile due to changes in memcheck.h
3732 255009 helgrind/drd: crash on chmod with invalid parameter
3733 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3734 255355 helgrind/drd: crash on threaded programs doing fork
3736 255418 (SSE4.x) rint call compiled with ICC
3737 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3738 255888 closing valgrindoutput tag outputted to log-stream on error
3739 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3740 255966 Slowness when using mempool annotations
3741 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3742 256600 super-optimized strcasecmp() false positive
3743 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3744 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3745 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3746 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3747 257276 Missing case in memcheck --track-origins=yes
3748 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3749 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3750 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3751 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3752 263099 callgrind_annotate counts Ir improperly [...]
3753 263877 undefined coprocessor instruction on ARMv7
3754 265964 configure FAIL with glibc-2.13
3755 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3756 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3757 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3758 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3759 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3761 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3765 Release 3.6.0 (21 October 2010)
3766 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3767 3.6.0 is a feature release with many significant improvements and the
3768 usual collection of bug fixes.
3770 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3771 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3772 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3774 -------------------------
3776 Here are some highlights. Details are shown further down:
3778 * Support for ARM/Linux.
3780 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3782 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3784 * Support for the SSE4.2 instruction set.
3786 * Enhancements to the Callgrind profiler, including the ability to
3787 handle CPUs with three levels of cache.
3789 * A new experimental heap profiler, DHAT.
3791 * A huge number of bug fixes and small enhancements.
3793 -------------------------
3795 Here are details of the above changes, together with descriptions of
3796 many other changes, and a list of fixed bugs.
3798 * ================== PLATFORM CHANGES =================
3800 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3801 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3802 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3804 This requires a CPU capable of running the ARMv7-A instruction set
3805 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3806 of the user space instruction set, including ARM and Thumb integer
3807 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3808 Cachegrind and Massif tools work properly; other tools work to
3811 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3812 with support for recent releases of the underlying toolchain
3813 components, notably gcc-4.5 and glibc-2.12.
3815 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3816 support also works much better on OS X 10.5, and is as solid as
3819 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3820 64-bit mode. In 32-bit mode, support is only available up to and
3821 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3822 supported in 64-bit mode, and 32-bit mode does in fact support the
3823 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3824 10.6 on 32-bit targets.
3826 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3827 and including version 2.05 is supported.
3829 * ==================== TOOL CHANGES ====================
3831 * Cachegrind has a new processing script, cg_diff, which finds the
3832 difference between two profiles. It's very useful for evaluating
3833 the performance effects of a change in a program.
3835 Related to this change, the meaning of cg_annotate's (rarely-used)
3836 --threshold option has changed; this is unlikely to affect many
3837 people, if you do use it please see the user manual for details.
3839 * Callgrind now can do branch prediction simulation, similar to
3840 Cachegrind. In addition, it optionally can count the number of
3841 executed global bus events. Both can be used for a better
3842 approximation of a "Cycle Estimation" as derived event (you need to
3843 update the event formula in KCachegrind yourself).
3845 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3846 rather than the L2 cache. This is to accommodate machines with
3847 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3848 cache configuration of such a machine it will run the simulation as
3849 if the L2 cache isn't present. This means the results are less
3850 likely to match the true result for the machine, but
3851 Cachegrind/Callgrind's results are already only approximate, and
3852 should not be considered authoritative. The results are still
3853 useful for giving a general idea about a program's locality.
3855 * Massif has a new option, --pages-as-heap, which is disabled by
3856 default. When enabled, instead of tracking allocations at the level
3857 of heap blocks (as allocated with malloc/new/new[]), it instead
3858 tracks memory allocations at the level of memory pages (as mapped by
3859 mmap, brk, etc). Each mapped page is treated as its own block.
3860 Interpreting the page-level output is harder than the heap-level
3861 output, but this option is useful if you want to account for every
3862 byte of memory used by a program.
3864 * DRD has two new command-line options: --free-is-write and
3865 --trace-alloc. The former allows to detect reading from already freed
3866 memory, and the latter allows tracing of all memory allocations and
3869 * DRD has several new annotations. Custom barrier implementations can
3870 now be annotated, as well as benign races on static variables.
3872 * DRD's happens before / happens after annotations have been made more
3873 powerful, so that they can now also be used to annotate e.g. a smart
3874 pointer implementation.
3876 * Helgrind's annotation set has also been drastically improved, so as
3877 to provide to users a general set of annotations to describe locks,
3878 semaphores, barriers and condition variables. Annotations to
3879 describe thread-safe reference counted heap objects have also been
3882 * Memcheck has a new command-line option, --show-possibly-lost, which
3883 is enabled by default. When disabled, the leak detector will not
3884 show possibly-lost blocks.
3886 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3887 has been added. DHAT keeps track of allocated heap blocks, and also
3888 inspects every memory reference to see which block (if any) is being
3889 accessed. This gives a lot of insight into block lifetimes,
3890 utilisation, turnover, liveness, and the location of hot and cold
3891 fields. You can use DHAT to do hot-field profiling.
3893 * ==================== OTHER CHANGES ====================
3895 * Improved support for unfriendly self-modifying code: the extra
3896 overhead incurred by --smc-check=all has been reduced by
3897 approximately a factor of 5 as compared with 3.5.0.
3899 * Ability to show directory names for source files in error messages.
3900 This is combined with a flexible mechanism for specifying which
3901 parts of the paths should be shown. This is enabled by the new flag
3904 * A new flag, --require-text-symbol, which will stop the run if a
3905 specified symbol is not found it a given shared object when it is
3906 loaded into the process. This makes advanced working with function
3907 intercepting and wrapping safer and more reliable.
3909 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3910 and control of Valgrind is now available for the tools Memcheck and
3911 Helgrind. XML output from Valgrind is available for Memcheck,
3912 Helgrind and exp-Ptrcheck.
3914 * More reliable stack unwinding on amd64-linux, particularly in the
3915 presence of function wrappers, and with gcc-4.5 compiled code.
3917 * Modest scalability (performance improvements) for massive
3918 long-running applications, particularly for those with huge amounts
3921 * Support for analyzing programs running under Wine with has been
3922 improved. The header files <valgrind/valgrind.h>,
3923 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3924 Windows-programs compiled with MinGW or one of the Microsoft Visual
3927 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3928 The 32-bit simulator was not affected. This did not occur often,
3929 but when it did would usually crash the program under test.
3932 * A large number of bugs were fixed. These are shown below.
3934 * A number of bugs were investigated, and were candidates for fixing,
3935 but are not fixed in 3.6.0, due to lack of developer time. They may
3936 get fixed in later releases. They are:
3938 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3939 212419 false positive "lock order violated" (A+B vs A)
3940 213685 Undefined value propagates past dependency breaking instruction
3941 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3942 237920 valgrind segfault on fork failure
3943 242137 support for code compiled by LLVM-2.8
3944 242423 Another unknown Intel cache config value
3945 243232 Inconsistent Lock Orderings report with trylock
3946 243483 ppc: callgrind triggers VEX assertion failure
3947 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3948 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3950 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3951 249435 Analyzing wine programs with callgrind triggers a crash
3952 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3953 250065 Handling large allocations
3954 250101 huge "free" memory usage due to m_mallocfree.c
3955 "superblocks fragmentation"
3956 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3957 252091 Callgrind on ARM does not detect function returns correctly
3958 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3959 254420 memory pool tracking broken
3960 n-i-bz support for adding symbols for JIT generated code
3963 The following bugs have been fixed or resolved. Note that "n-i-bz"
3964 stands for "not in bugzilla" -- that is, a bug that was reported to us
3965 but never got a bugzilla entry. We encourage you to file bugs in
3966 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3967 mailing the developers (or mailing lists) directly -- bugs that are
3968 not entered into bugzilla tend to get forgotten about or ignored.
3970 To see details of a given bug, visit
3971 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3972 where XXXXXX is the bug number as listed below.
3974 135264 dcbzl instruction missing
3976 153699 Valgrind should report unaligned reads with movdqa
3978 190429 Valgrind reports lost of errors in ld.so
3979 with x86_64 2.9.90 glibc
3980 197266 valgrind appears to choke on the xmms instruction
3982 197988 Crash when demangling very large symbol names
3983 202315 unhandled syscall: 332 (inotify_init1)
3984 203256 Add page-level profiling to Massif
3985 205093 dsymutil=yes needs quotes, locking (partial fix)
3986 205241 Snow Leopard 10.6 support (partial fix)
3987 206600 Leak checker fails to upgrade indirect blocks when their
3988 parent becomes reachable
3989 210935 port valgrind.h (not valgrind) to win32 so apps run under
3990 wine can make client requests
3991 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3992 within Linux ip-stack checksum functions
3993 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3995 213685 Undefined value propagates past dependency breaking instruction
3997 215914 Valgrind inserts bogus empty environment variable
3999 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
4000 222545 shmat fails under valgind on some arm targets
4001 222560 ARM NEON support
4004 232509 Docs build fails with formatting inside <title></title> elements
4006 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
4007 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
4008 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
4009 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
4010 237485 symlink (syscall 57) is not supported on Mac OS
4011 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
4013 238208 is_just_below_ESP doesn't take into account red-zone
4014 238345 valgrind passes wrong $0 when executing a shell script
4015 238679 mq_timedreceive syscall doesn't flag the reception buffer
4017 238696 fcntl command F_DUPFD_CLOEXEC not supported
4018 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4019 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4020 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
4022 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
4028 242606 unhandled syscall: setegid (in Ptrcheck)
4029 242814 Helgrind "Impossible has happened" during
4030 QApplication::initInstance();
4031 243064 Valgrind attempting to read debug information from iso
4032 243270 Make stack unwinding in Valgrind wrappers more reliable
4033 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
4034 sysno = 277 (mq_open)
4035 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
4036 244493 ARM VFP d16-d31 registers support
4037 244670 add support for audit_session_self syscall on Mac OS 10.6
4038 244921 The xml report of helgrind tool is not well format
4039 244923 In the xml report file, the <preamble> not escape the
4040 xml char, eg '<','&','>'
4041 245535 print full path names in plain text reports
4042 245925 x86-64 red zone handling problem
4043 246258 Valgrind not catching integer underruns + new [] s
4044 246311 reg/reg cmpxchg doesn't work on amd64
4045 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
4046 246888 Improve Makefile.vex.am
4047 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
4048 to [f]chmod_extended
4049 247526 IBM POWER6 (ISA 2.05) support is incomplete
4050 247561 Some leak testcases fails due to reachable addresses in
4052 247875 sizeofIRType to handle Ity_I128
4053 247894 [PATCH] unhandled syscall sys_readahead
4054 247980 Doesn't honor CFLAGS passed to configure
4055 248373 darwin10.supp is empty in the trunk
4056 248822 Linux FIBMAP ioctl has int parameter instead of long
4057 248893 [PATCH] make readdwarf.c big endianess safe to enable
4058 unwinding on big endian systems
4059 249224 Syscall 336 not supported (SYS_proc_info)
4061 249775 Incorrect scheme for detecting NEON capabilities of host CPU
4062 249943 jni JVM init fails when using valgrind
4063 249991 Valgrind incorrectly declares AESKEYGENASSIST support
4065 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
4066 250799 frexp$fenv_access_off function generates SIGILL
4067 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
4068 251251 support pclmulqdq insn
4069 251362 valgrind: ARM: attach to debugger either fails or provokes
4071 251674 Unhandled syscall 294
4074 254257 Add support for debugfiles found by build-id
4075 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
4076 254646 Wrapped functions cause stack misalignment on OS X
4077 (and possibly Linux)
4078 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
4080 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
4084 Release 3.5.0 (19 August 2009)
4085 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4086 3.5.0 is a feature release with many significant improvements and the
4087 usual collection of bug fixes. The main improvement is that Valgrind
4088 now works on Mac OS X.
4090 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
4091 and X86/Darwin. Support for recent distros and toolchain components
4092 (glibc 2.10, gcc 4.5) has been added.
4094 -------------------------
4096 Here is a short summary of the changes. Details are shown further
4099 * Support for Mac OS X (10.5.x).
4101 * Improvements and simplifications to Memcheck's leak checker.
4103 * Clarification and simplifications in various aspects of Valgrind's
4106 * XML output for Helgrind and Ptrcheck.
4108 * Performance and stability improvements for Helgrind and DRD.
4110 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
4112 * A new experimental tool, BBV, useful for computer architecture
4115 * Improved Wine support, including ability to read Windows PDB
4118 -------------------------
4120 Here are details of the above changes, followed by descriptions of
4121 many other minor changes, and a list of fixed bugs.
4124 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
4125 called "Darwin" because that is the name of the OS core, which is the
4126 level that Valgrind works at.)
4130 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
4131 because it would require work and 10.4 is only becoming less common.
4133 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
4134 fairly well. For 10.5.x, 32-bit programs are the default even on
4135 64-bit machines, so it handles most current programs.
4137 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
4138 officially supported, but simple programs at least will probably work.
4139 However, start-up is slow.
4141 - PowerPC machines are not supported.
4143 Things that don't work:
4145 - The Ptrcheck tool.
4147 - Objective-C garbage collection.
4151 - If you have Rogue Amoeba's "Instant Hijack" program installed,
4152 Valgrind will fail with a SIGTRAP at start-up. See
4153 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
4158 - You will likely find --dsymutil=yes a useful option, as error
4159 messages may be imprecise without it.
4161 - Mac OS X support is new and therefore will be less robust than the
4162 Linux support. Please report any bugs you find.
4164 - Threaded programs may run more slowly than on Linux.
4166 Many thanks to Greg Parker for developing this port over several years.
4169 * Memcheck's leak checker has been improved.
4171 - The results for --leak-check=summary now match the summary results
4172 for --leak-check=full. Previously they could differ because
4173 --leak-check=summary counted "indirectly lost" blocks and
4174 "suppressed" blocks as "definitely lost".
4176 - Blocks that are only reachable via at least one interior-pointer,
4177 but are directly pointed to by a start-pointer, were previously
4178 marked as "still reachable". They are now correctly marked as
4181 - The default value for the --leak-resolution option has been
4182 changed from "low" to "high". In general, this means that more
4183 leak reports will be produced, but each leak report will describe
4184 fewer leaked blocks.
4186 - With --leak-check=full, "definitely lost" and "possibly lost"
4187 leaks are now considered as proper errors, ie. they are counted
4188 for the "ERROR SUMMARY" and affect the behaviour of
4189 --error-exitcode. These leaks are not counted as errors if
4190 --leak-check=summary is specified, however.
4192 - Documentation for the leak checker has been improved.
4195 * Various aspects of Valgrind's text output have changed.
4197 - Valgrind's start-up message has changed. It is shorter but also
4198 includes the command being run, which makes it easier to use
4199 --trace-children=yes. An example:
4201 - Valgrind's shut-down messages have also changed. This is most
4202 noticeable with Memcheck, where the leak summary now occurs before
4203 the error summary. This change was necessary to allow leaks to be
4204 counted as proper errors (see the description of the leak checker
4205 changes above for more details). This was also necessary to fix a
4206 longstanding bug in which uses of suppressions against leaks were
4207 not "counted", leading to difficulties in maintaining suppression
4208 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
4210 - Behavior of -v has changed. In previous versions, -v printed out
4211 a mixture of marginally-user-useful information, and tool/core
4212 statistics. The statistics printing has now been moved to its own
4213 flag, --stats=yes. This means -v is less verbose and more likely
4214 to convey useful end-user information.
4216 - The format of some (non-XML) stack trace entries has changed a
4217 little. Previously there were six possible forms:
4219 0x80483BF: really (a.c:20)
4220 0x80483BF: really (in /foo/a.out)
4222 0x80483BF: (within /foo/a.out)
4223 0x80483BF: ??? (a.c:20)
4226 The third and fourth of these forms have been made more consistent
4227 with the others. The six possible forms are now:
4229 0x80483BF: really (a.c:20)
4230 0x80483BF: really (in /foo/a.out)
4231 0x80483BF: really (in ???)
4232 0x80483BF: ??? (in /foo/a.out)
4233 0x80483BF: ??? (a.c:20)
4236 Stack traces produced when --xml=yes is specified are different
4240 * Helgrind and Ptrcheck now support XML output, so they can be used
4241 from GUI tools. Also, the XML output mechanism has been
4244 - The XML format has been overhauled and generalised, so it is more
4245 suitable for error reporting tools in general. The Memcheck
4246 specific aspects of it have been removed. The new format, which
4247 is an evolution of the old format, is described in
4248 docs/internals/xml-output-protocol4.txt.
4250 - Memcheck has been updated to use the new format.
4252 - Helgrind and Ptrcheck are now able to emit output in this format.
4254 - The XML output mechanism has been overhauled. XML is now output
4255 to its own file descriptor, which means that:
4257 * Valgrind can output text and XML independently.
4259 * The longstanding problem of XML output being corrupted by
4260 unexpected un-tagged text messages is solved.
4262 As before, the destination for text output is specified using
4263 --log-file=, --log-fd= or --log-socket=.
4265 As before, XML output for a tool is enabled using --xml=yes.
4267 Because there's a new XML output channel, the XML output
4268 destination is now specified by --xml-file=, --xml-fd= or
4271 Initial feedback has shown this causes some confusion. To
4272 clarify, the two envisaged usage scenarios are:
4274 (1) Normal text output. In this case, do not specify --xml=yes
4275 nor any of --xml-file=, --xml-fd= or --xml-socket=.
4277 (2) XML output. In this case, specify --xml=yes, and one of
4278 --xml-file=, --xml-fd= or --xml-socket= to select the XML
4279 destination, one of --log-file=, --log-fd= or --log-socket=
4280 to select the destination for any remaining text messages,
4281 and, importantly, -q.
4283 -q makes Valgrind completely silent on the text channel,
4284 except in the case of critical failures, such as Valgrind
4285 itself segfaulting, or failing to read debugging information.
4286 Hence, in this scenario, it suffices to check whether or not
4287 any output appeared on the text channel. If yes, then it is
4288 likely to be a critical error which should be brought to the
4289 attention of the user. If no (the text channel produced no
4290 output) then it can be assumed that the run was successful.
4292 This allows GUIs to make the critical distinction they need to
4293 make (did the run fail or not?) without having to search or
4294 filter the text output channel in any way.
4296 It is also recommended to use --child-silent-after-fork=yes in
4300 * Improvements and changes in Helgrind:
4302 - XML output, as described above
4304 - Checks for consistent association between pthread condition
4305 variables and their associated mutexes are now performed.
4307 - pthread_spinlock functions are supported.
4309 - Modest performance improvements.
4311 - Initial (skeletal) support for describing the behaviour of
4312 non-POSIX synchronisation objects through ThreadSanitizer
4313 compatible ANNOTATE_* macros.
4315 - More controllable tradeoffs between performance and the level of
4316 detail of "previous" accesses in a race. There are now three
4319 * --history-level=full. This is the default, and was also the
4320 default in 3.4.x. It shows both stacks involved in a race, but
4321 requires a lot of memory and can be very slow in programs that
4322 do many inter-thread synchronisation events.
4324 * --history-level=none. This only shows the later stack involved
4325 in a race. This can be much faster than --history-level=full,
4326 but makes it much more difficult to find the other access
4327 involved in the race.
4329 The new intermediate setting is
4331 * --history-level=approx
4333 For the earlier (other) access, two stacks are presented. The
4334 earlier access is guaranteed to be somewhere in between the two
4335 program points denoted by those stacks. This is not as useful
4336 as showing the exact stack for the previous access (as per
4337 --history-level=full), but it is better than nothing, and it's
4338 almost as fast as --history-level=none.
4341 * New features and improvements in DRD:
4343 - The error messages printed by DRD are now easier to interpret.
4344 Instead of using two different numbers to identify each thread
4345 (Valgrind thread ID and DRD thread ID), DRD does now identify
4346 threads via a single number (the DRD thread ID). Furthermore
4347 "first observed at" information is now printed for all error
4348 messages related to synchronization objects.
4350 - Added support for named semaphores (sem_open() and sem_close()).
4352 - Race conditions between pthread_barrier_wait() and
4353 pthread_barrier_destroy() calls are now reported.
4355 - Added support for custom allocators through the macros
4356 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
4357 in <valgrind/valgrind.h>). An alternative for these two macros is
4358 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
4361 - Added support for annotating non-POSIX synchronization objects
4362 through several new ANNOTATE_*() macros.
4364 - OpenMP: added support for the OpenMP runtime (libgomp) included
4365 with gcc versions 4.3.0 and 4.4.0.
4369 - Added two new command-line options (--first-race-only and
4370 --segment-merging-interval).
4373 * Genuinely atomic support for x86/amd64/ppc atomic instructions
4375 Valgrind will now preserve (memory-access) atomicity of LOCK-
4376 prefixed x86/amd64 instructions, and any others implying a global
4377 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
4379 This means that Valgrinded processes will "play nicely" in
4380 situations where communication with other processes, or the kernel,
4381 is done through shared memory and coordinated with such atomic
4382 instructions. Prior to this change, such arrangements usually
4383 resulted in hangs, races or other synchronisation failures, because
4384 Valgrind did not honour atomicity of such instructions.
4387 * A new experimental tool, BBV, has been added. BBV generates basic
4388 block vectors for use with the SimPoint analysis tool, which allows
4389 a program's overall behaviour to be approximated by running only a
4390 fraction of it. This is useful for computer architecture
4391 researchers. You can run BBV by specifying --tool=exp-bbv (the
4392 "exp-" prefix is short for "experimental"). BBV was written by
4396 * Modestly improved support for running Windows applications under
4397 Wine. In particular, initial support for reading Windows .PDB debug
4398 information has been added.
4401 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
4402 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
4406 * The Valgrind client requests VALGRIND_PRINTF and
4407 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
4408 the string was always printed immediately on its own line. Now, the
4409 string will be added to a buffer but not printed until a newline is
4410 encountered, or other Valgrind output is printed (note that for
4411 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
4412 "other Valgrind output"). This allows you to use multiple
4413 VALGRIND_PRINTF calls to build up a single output line, and also to
4414 print multiple output lines with a single request (by embedding
4415 multiple newlines in the string).
4418 * The graphs drawn by Massif's ms_print program have changed slightly:
4420 - The half-height chars '.' and ',' are no longer drawn, because
4421 they are confusing. The --y option can be used if the default
4422 y-resolution is not high enough.
4424 - Horizontal lines are now drawn after the top of a snapshot if
4425 there is a gap until the next snapshot. This makes it clear that
4426 the memory usage has not dropped to zero between snapshots.
4429 * Something that happened in 3.4.0, but wasn't clearly announced: the
4430 option --read-var-info=yes can be used by some tools (Memcheck,
4431 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
4432 variable type and location information. This makes those tools
4433 start up more slowly and increases memory consumption, but
4434 descriptions of data addresses in error messages become more
4438 * exp-Omega, an experimental instantaneous leak-detecting tool, was
4439 disabled in 3.4.0 due to a lack of interest and maintenance,
4440 although the source code was still in the distribution. The source
4441 code has now been removed from the distribution. For anyone
4442 interested, the removal occurred in SVN revision r10247.
4445 * Some changes have been made to the build system.
4447 - VEX/ is now integrated properly into the build system. This means
4448 that dependency tracking within VEX/ now works properly, "make
4449 install" will work without requiring "make" before it, and
4450 parallel builds (ie. 'make -j') now work (previously a
4451 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
4452 was effectively ignored).
4454 - The --with-vex configure option has been removed. It was of
4455 little use and removing it simplified the build system.
4457 - The location of some install files has changed. This should not
4458 affect most users. Those who might be affected:
4460 * For people who use Valgrind with MPI programs, the installed
4461 libmpiwrap.so library has moved from
4462 $(INSTALL)/<platform>/libmpiwrap.so to
4463 $(INSTALL)/libmpiwrap-<platform>.so.
4465 * For people who distribute standalone Valgrind tools, the
4466 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
4467 have moved to $(INSTALL)/libcoregrind-<platform>.a.
4469 These changes simplify the build system.
4471 - Previously, all the distributed suppression (*.supp) files were
4472 installed. Now, only default.supp is installed. This should not
4473 affect users as the other installed suppression files were not
4474 read; the fact that they were installed was a mistake.
4477 * KNOWN LIMITATIONS:
4479 - Memcheck is unusable with the Intel compiler suite version 11.1,
4480 when it generates code for SSE2-and-above capable targets. This
4481 is because of icc's use of highly optimised inlined strlen
4482 implementations. It causes Memcheck to report huge numbers of
4483 false errors even in simple programs. Helgrind and DRD may also
4486 Versions 11.0 and earlier may be OK, but this has not been
4490 The following bugs have been fixed or resolved. Note that "n-i-bz"
4491 stands for "not in bugzilla" -- that is, a bug that was reported to us
4492 but never got a bugzilla entry. We encourage you to file bugs in
4493 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4494 mailing the developers (or mailing lists) directly -- bugs that are
4495 not entered into bugzilla tend to get forgotten about or ignored.
4497 To see details of a given bug, visit
4498 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4499 where XXXXXX is the bug number as listed below.
4501 84303 How about a LockCheck tool?
4502 91633 dereference of null ptr in vgPlain_st_basetype
4503 97452 Valgrind doesn't report any pthreads problems
4504 100628 leak-check gets assertion failure when using
4505 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4506 108528 NPTL pthread cleanup handlers not called
4507 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4508 110128 mallinfo is not implemented...
4509 110770 VEX: Generated files not always updated when making valgrind
4510 111102 Memcheck: problems with large (memory footprint) applications
4511 115673 Vex's decoder should never assert
4512 117564 False positive: Syscall param clone(child_tidptr) contains
4513 uninitialised byte(s)
4514 119404 executing ssh from inside valgrind fails
4515 133679 Callgrind does not write path names to sources with dwarf debug
4517 135847 configure.in problem with non gnu compilers (and possible fix)
4518 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4519 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4520 136230 memcheck reports "possibly lost", should be "still reachable"
4521 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4522 137904 Valgrind reports a memory leak when using POSIX threads,
4524 139076 valgrind VT_GETSTATE error
4525 142228 complaint of elf_dynamic_do_rela in trivial usage
4526 145347 spurious warning with USBDEVFS_REAPURB
4527 148441 (wine) can't find memory leak in Wine, win32 binary
4529 148742 Leak-check fails assert on exit
4530 149878 add (proper) check for calloc integer overflow
4531 150606 Call graph is broken when using callgrind control
4532 152393 leak errors produce an exit code of 0. I need some way to
4533 cause leak errors to result in a nonzero exit code.
4534 157154 documentation (leak-resolution doc speaks about num-callers
4535 def=4) + what is a loss record
4536 159501 incorrect handling of ALSA ioctls
4537 162020 Valgrinding an empty/zero-byte file crashes valgrind
4538 162482 ppc: Valgrind crashes while reading stabs information
4539 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4540 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4541 163560 VEX/test_main.c is missing from valgrind-3.3.1
4542 164353 malloc_usable_size() doesn't return a usable size
4543 165468 Inconsistent formatting in memcheck manual -- please fix
4544 169505 main.c:286 (endOfInstr):
4545 Assertion 'ii->cost_offset == *cost_offset' failed
4546 177206 Generate default.supp during compile instead of configure
4547 177209 Configure valt_load_address based on arch+os
4548 177305 eventfd / syscall 323 patch lost
4549 179731 Tests fail to build because of inlining of non-local asm labels
4550 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4551 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4552 181594 Bogus warning for empty text segment
4553 181707 dwarf doesn't require enumerations to have name
4554 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4555 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4556 Assertion '!already_present' failed.
4557 185359 exp-ptrcheck: unhandled syscall getresuid()
4558 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4559 185816 Valgrind is unable to handle debug info for files with split
4560 debug info that are prelinked afterwards
4561 185980 [darwin] unhandled syscall: sem_open
4562 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4563 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4564 186790 Suppression pattern used for leaks are not reported
4565 186796 Symbols with length>200 in suppression files are ignored
4566 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4567 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4568 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4569 188046 bashisms in the configure script
4570 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4571 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4572 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4573 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4574 assertion fail "!lock->heldBy"
4575 188427 Add support for epoll_create1 (with patch)
4576 188530 Support for SIOCGSTAMPNS
4577 188560 Include valgrind.spec in the tarball
4578 188572 Valgrind on Mac should suppress setenv() mem leak
4579 189054 Valgrind fails to build because of duplicate non-local asm labels
4580 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4581 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4582 189763 drd assertion failure: s_threadinfo[tid].is_recording
4583 190219 unhandled syscall: 328 (x86-linux)
4584 190391 dup of 181394; see above
4585 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4586 190820 No debug information on powerpc-linux
4587 191095 PATCH: Improve usbdevfs ioctl handling
4588 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4590 191189 --xml=yes should obey --gen-suppressions=all
4591 191192 syslog() needs a suppression on macosx
4592 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4593 191761 getrlimit on MacOSX
4594 191992 multiple --fn-skip only works sometimes; dependent on order
4595 192634 V. reports "aspacem sync_check_mapping_callback:
4596 segment mismatch" on Darwin
4597 192954 __extension__ missing on 2 client requests
4598 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4599 194474 "INSTALL" file has different build instructions than "README"
4600 194671 Unhandled syscall (sem_wait?) from mac valgrind
4601 195069 memcheck: reports leak (memory still reachable) for
4603 195169 drd: (vgDrd_barrier_post_wait):
4604 Assertion 'r->sg[p->post_iteration]' failed.
4605 195268 valgrind --log-file doesn't accept ~/...
4606 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4607 195860 WARNING: unhandled syscall: unix:223
4608 196528 need a error suppression for pthread_rwlock_init under os x?
4609 197227 Support aio_* syscalls on Darwin
4610 197456 valgrind should reject --suppressions=(directory)
4611 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4612 197591 unhandled syscall 27 (mincore)
4613 197793 Merge DCAS branch to the trunk == 85756, 142103
4614 197794 Avoid duplicate filenames in Vex
4615 197898 make check fails on current SVN
4616 197901 make check fails also under exp-ptrcheck in current SVN
4617 197929 Make --leak-resolution=high the default
4618 197930 Reduce spacing between leak reports
4619 197933 Print command line of client at start-up, and shorten preamble
4620 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4621 198395 add BBV to the distribution as an experimental tool
4622 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4623 198649 callgrind_annotate doesn't cumulate counters
4624 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4625 199977 Valgrind complains about an unrecognized instruction in the
4626 atomic_incs test program
4627 200029 valgrind isn't able to read Fedora 12 debuginfo
4628 200760 darwin unhandled syscall: unix:284
4629 200827 DRD doesn't work on Mac OS X
4630 200990 VG_(read_millisecond_timer)() does not work correctly
4631 201016 Valgrind does not support pthread_kill() on Mac OS
4632 201169 Document --read-var-info
4633 201323 Pre-3.5.0 performance sanity checking
4634 201384 Review user manual for the 3.5.0 release
4635 201585 mfpvr not implemented on ppc
4636 201708 tests failing because x86 direction flag is left set
4637 201757 Valgrind doesn't handle any recent sys_futex additions
4638 204377 64-bit valgrind can not start a shell script
4639 (with #!/path/to/shell) if the shell is a 32-bit executable
4640 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4641 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4642 about memory allocation events with memory access tracing enabled
4643 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4645 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4649 Release 3.4.1 (28 February 2009)
4650 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4651 3.4.1 is a bug-fix release that fixes some regressions and assertion
4652 failures in debug info reading in 3.4.0, most notably incorrect stack
4653 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4654 other debug info problems are also fixed. A number of bugs in the
4655 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4657 In view of the fact that 3.4.0 contains user-visible regressions
4658 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4659 encouraged to ship 3.4.1 in preference to 3.4.0.
4661 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4662 bugzilla" -- that is, a bug that was reported to us but never got a
4663 bugzilla entry. We encourage you to file bugs in bugzilla
4664 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4665 developers (or mailing lists) directly -- bugs that are not entered
4666 into bugzilla tend to get forgotten about or ignored.
4668 n-i-bz Fix various bugs reading icc-11 generated debug info
4669 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4670 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4671 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4672 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4673 179618 exp-ptrcheck crashed / exit prematurely
4674 179624 helgrind: false positive races with pthread_create and
4675 recv/open/close/read
4676 134207 pkg-config output contains @VG_PLATFORM@
4677 176926 floating point exception at valgrind startup with PPC 440EPX
4678 181594 Bogus warning for empty text segment
4679 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4680 181707 Dwarf3 doesn't require enumerations to have name
4681 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4682 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4683 Assertion '!already_present' failed.
4684 185359 exp-ptrcheck unhandled syscall getresuid()
4686 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4687 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4691 Release 3.4.0 (2 January 2009)
4692 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4693 3.4.0 is a feature release with many significant improvements and the
4694 usual collection of bug fixes. This release supports X86/Linux,
4695 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4696 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4698 3.4.0 brings some significant tool improvements. Memcheck can now
4699 report the origin of uninitialised values, the thread checkers
4700 Helgrind and DRD are much improved, and we have a new experimental
4701 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4702 global arrays. In detail:
4704 * Memcheck is now able to track the origin of uninitialised values.
4705 When it reports an uninitialised value error, it will try to show
4706 the origin of the value, as either a heap or stack allocation.
4707 Origin tracking is expensive and so is not enabled by default. To
4708 use it, specify --track-origins=yes. Memcheck's speed will be
4709 essentially halved, and memory usage will be significantly
4710 increased. Nevertheless it can drastically reduce the effort
4711 required to identify the root cause of uninitialised value errors,
4712 and so is often a programmer productivity win, despite running more
4715 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4716 3.4.0, will be released shortly.
4718 * Helgrind's race detection algorithm has been completely redesigned
4719 and reimplemented, to address usability and scalability concerns:
4721 - The new algorithm has a lower false-error rate: it is much less
4722 likely to report races that do not really exist.
4724 - Helgrind will display full call stacks for both accesses involved
4725 in a race. This makes it easier to identify the root causes of
4728 - Limitations on the size of program that can run have been removed.
4730 - Performance has been modestly improved, although that is very
4733 - Direct support for Qt4 threading has been added.
4735 - pthread_barriers are now directly supported.
4737 - Helgrind works well on all supported Linux targets.
4739 * The DRD thread debugging tool has seen major improvements:
4741 - Greatly improved performance and significantly reduced memory
4744 - Support for several major threading libraries (Boost.Thread, Qt4,
4745 glib, OpenMP) has been added.
4747 - Support for atomic instructions, POSIX semaphores, barriers and
4748 reader-writer locks has been added.
4750 - Works now on PowerPC CPUs too.
4752 - Added support for printing thread stack usage at thread exit time.
4754 - Added support for debugging lock contention.
4756 - Added a manual for Drd.
4758 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4759 checks for misuses of pointers. In that sense it is a bit like
4760 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4761 detect overruns of stack and global arrays, it can detect
4762 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4763 detect accesses heap blocks that have been freed a very long time
4764 ago (millions of blocks in the past).
4766 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4767 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4768 of the main Valgrind documentation. As this is an experimental
4769 tool, we would be particularly interested in hearing about your
4770 experiences with it.
4772 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4773 longer built by default, although the code remains in the repository
4774 and the tarball. This is due to three factors: a perceived lack of
4775 users, a lack of maintenance, and concerns that it may not be
4776 possible to achieve reliable operation using the existing design.
4778 * As usual, support for the latest Linux distros and toolchain
4779 components has been added. It should work well on Fedora Core 10,
4780 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4781 state) is supported, as is glibc-2.9. The C++ demangler has been
4782 updated so as to work well with C++ compiled by even the most recent
4785 * You can now use frame-level wildcards in suppressions. This was a
4786 frequently-requested enhancement. A line "..." in a suppression now
4787 matches zero or more frames. This makes it easier to write
4788 suppressions which are precise yet insensitive to changes in
4791 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4793 * Very basic support for IBM Power6 has been added (64-bit processes only).
4795 * Valgrind is now cross-compilable. For example, it is possible to
4796 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4797 on a ppc32/64-linux target.
4799 * You can set the main thread's stack size at startup using the
4800 new --main-stacksize= flag (subject of course to ulimit settings).
4801 This is useful for running apps that need a lot of stack space.
4803 * The limitation that you can't use --trace-children=yes together
4804 with --db-attach=yes has been removed.
4806 * The following bugs have been fixed. Note that "n-i-bz" stands for
4807 "not in bugzilla" -- that is, a bug that was reported to us but
4808 never got a bugzilla entry. We encourage you to file bugs in
4809 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4810 mailing the developers (or mailing lists) directly.
4812 n-i-bz Make return types for some client requests 64-bit clean
4813 n-i-bz glibc 2.9 support
4814 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4815 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4816 n-i-bz Building in an env without gdb gives bogus gdb attach
4817 92456 Tracing the origin of uninitialised memory
4818 106497 Valgrind does not demangle some C++ template symbols
4820 151612 Suppression with "..." (frame-level wildcards in .supp files)
4821 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4822 159285 unhandled syscall:25 (stime, on x86-linux)
4823 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4824 160954 ppc build of valgrind crashes with illegal instruction (isel)
4825 160956 mallinfo implementation, w/ patch
4826 162092 Valgrind fails to start gnome-system-monitor
4827 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4828 163794 assertion failure with "--track-origins=yes"
4829 163933 sigcontext.err and .trapno must be set together
4830 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4831 164476 Missing kernel module loading system calls
4832 164669 SVN regression: mmap() drops posix file locks
4833 166581 Callgrind output corruption when program forks
4834 167288 Patch file for missing system calls on Cell BE
4835 168943 unsupported scas instruction pentium
4836 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4837 172417 x86->IR: 0x82 ...
4838 172563 amd64->IR: 0xD9 0xF5 - fprem1
4839 173099 .lds linker script generation error
4840 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4841 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4843 174908 --log-file value not expanded correctly for core file
4844 175044 Add lookup_dcookie for amd64
4845 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4847 Developer-visible changes:
4849 * Valgrind's debug-info reading machinery has been majorly overhauled.
4850 It can now correctly establish the addresses for ELF data symbols,
4851 which is something that has never worked properly before now.
4853 Also, Valgrind can now read DWARF3 type and location information for
4854 stack and global variables. This makes it possible to use the
4855 framework to build tools that rely on knowing the type and locations
4856 of stack and global variables, for example exp-Ptrcheck.
4858 Reading of such information is disabled by default, because most
4859 tools don't need it, and because it is expensive in space and time.
4860 However, you can force Valgrind to read it, using the
4861 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4862 make use of such information, if present, to provide source-level
4863 descriptions of data addresses in the error messages they create.
4865 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4866 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).