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 =================
24 * ==================== TOOL CHANGES ===================
27 * ==================== FIXED BUGS ====================
29 The following bugs have been fixed or resolved. Note that "n-i-bz"
30 stands for "not in bugzilla" -- that is, a bug that was reported to us
31 but never got a bugzilla entry. We encourage you to file bugs in
32 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
33 than mailing the developers (or mailing lists) directly -- bugs that
34 are not entered into bugzilla tend to get forgotten about or ignored.
36 426751 Valgrind reports «still reachable» memory using musl (alpine running inside docker)
37 460192 Add epoll_pwait2
38 469049 link failure on ppc64 (big endian) valgrind 3.20
39 469146 massif --ignore-fn does not ignore inlined functions
40 469768 Make it possible to install gdb scripts in a different location
41 470121 Can't run callgrind_control with valgrind 3.21.0 because of perl errors
42 470132 s390x: Assertion failure on VGM instruction
43 470520 Multiple realloc zero errors crash in MC_(eq_Error)
44 470713 Failure on the Yosys project: valgrind: m_libcfile.c:1802
45 (Bool vgPlain_realpath(const HChar *, HChar *)):
46 Assertion 'resolved' failed
47 470830 Don't print actions vgdb me ... continue for vgdb --multi mode
48 470978 s390x: Valgrind cannot start qemu-kvm when "sysctl vm.allocate_pgste=0"
49 471311 gdb --multi mode stdout redirecting to stderr
50 471807 Add support for lazy reading and downloading of DWARF debuginfo
51 472219 Syscall param ppoll(ufds.events) points to uninitialised byte(s)
52 472963 Broken regular expression in configure.ac
53 473604 Fix bug472219.c compile failure with Clang 16
54 473677 make check compile failure with Clang 16 based on GCC 13.x
55 473745 must-be-redirected function - strlen
56 473870 FreeBSD 14 applications fail early at startup
57 473944 Handle mold linker split RW PT_LOAD segments correctly
58 n-i-bz Allow arguments with spaces in .valgrindrc files
60 To see details of a given bug, visit
61 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
62 where XXXXXX is the bug number as listed above.
64 Release 3.21.0 (28 Apr 2023)
65 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
67 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
68 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
69 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
70 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
71 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
72 AMD64/macOS 10.13 and nanoMIPS/Linux.
74 * ==================== CORE CHANGES ===================
76 * When GDB is used to debug a program running under valgrind using
77 the valgrind gdbserver, GDB will automatically load some
78 python code provided in valgrind defining GDB front end commands
79 corresponding to the valgrind monitor commands.
80 These GDB front end commands accept the same format as
81 the monitor commands directly sent to the Valgrind gdbserver.
82 These GDB front end commands provide a better integration
83 in the GDB command line interface, so as to use for example
84 GDB auto-completion, command specific help, searching for
85 a command or command help matching a regexp, ...
86 For relevant monitor commands, GDB will evaluate arguments
87 to make the use of monitor commands easier.
88 For example, instead of having to print the address of a variable
89 to pass it to a subsequent monitor command, the GDB front end
90 command will evaluate the address argument. It is for example
92 (gdb) memcheck who_points_at &some_struct sizeof(some_struct)
95 $2 = (some_struct_type *) 0x1130a0 <some_struct>
96 (gdb) p sizeof(some_struct)
98 (gdb) monitor who_point_at 0x1130a0 40
100 * The vgdb utility now supports extended-remote protocol when
101 invoked with --multi. In this mode the GDB run command is
102 supported. Which means you don't need to run gdb and valgrind
103 from different terminals. So for example to start your program
104 in gdb and run it under valgrind you can do:
106 (gdb) set remote exec-file prog
108 (gdb) target extended-remote | vgdb --multi
111 * The behaviour of realloc with a size of zero can now
112 be changed for tools that intercept malloc. Those
113 tools are memcheck, helgrind, drd, massif and dhat.
114 Realloc implementations generally do one of two things
115 - free the memory like free() and return NULL
116 (GNU libc and ptmalloc).
117 - either free the memory and then allocate a
118 minimum sized block or just return the
119 original pointer. Return NULL if the
120 allocation of the minimum sized block fails
121 (jemalloc, musl, snmalloc, Solaris, macOS).
122 When Valgrind is configured and built it will
123 try to match the OS and libc behaviour. However
124 if you are using a non-default library to replace
125 malloc and family (e.g., musl on a glibc Linux or
126 tcmalloc on FreeBSD) then you can use a command line
127 option to change the behaviour of Valgrind:
128 --realloc-zero-bytes-frees=yes|no [yes on Linux glibc, no otherwise]
130 * ================== PLATFORM CHANGES =================
132 * Make the address space limit on FreeBSD amd64 128Gbytes
133 (the same as Linux and Solaris, it was 32Gbytes)
135 * ==================== TOOL CHANGES ===================
138 - When doing a delta leak_search, it is now possible to only
139 output the new loss records compared to the previous leak search.
140 This is available in the memcheck monitor command 'leak_search'
141 by specifying the "new" keyword or in your program by using
142 the client request VALGRIND_DO_NEW_LEAK_CHECK.
143 Whenever a "delta" leak search is done (i.e. when specifying
144 "new" or "increased" or "changed" in the monitor command),
145 the new loss records have a "new" marker.
146 - Valgrind now contains python code that defines GDB memcheck
147 front end monitor commands. See CORE CHANGES.
148 - Performs checks for the use of realloc with a size of zero.
149 This is non-portable and a source of errors. If memcheck
150 detects such a usage it will generate an error
151 realloc() with size 0
152 followed by the usual callstacks.
153 A switch has been added to allow this to be turned off:
154 --show-realloc-size-zero=yes|no [yes]
157 - The option ---history-backtrace-size=<number> allows to configure
158 the number of entries to record in the stack traces of "old"
159 accesses. Previously, this number was hardcoded to 8.
160 - Valgrind now contains python code that defines GDB helgrind
161 front end monitor commands. See CORE CHANGES.
164 - `--cache-sim=no` is now the default. The cache simulation is old and
165 unlikely to match any real modern machine. This means only the `Ir`
166 event are gathered by default, but that is by far the most useful
168 - `cg_annotate`, `cg_diff`, and `cg_merge` have been rewritten in
169 Python. As a result, they all have more flexible command line
170 argument handling, e.g. supporting `--show-percs` and
171 `--no-show-percs` forms as well as the existing `--show-percs=yes`
172 and `--show-percs=no`.
173 - `cg_annotate` has some functional changes.
174 - It's much faster, e.g. 3-4x on common cases.
175 - It now supports diffing (with `--diff`, `--mod-filename`, and
176 `--mod-funcname`) and merging (by passing multiple data files).
177 - It now provides more information at the file and function level.
178 There are now "File:function" and "Function:file" sections. These
179 are very useful for programs that use inlining a lot.
180 - Support for user-annotated files and the `-I`/`--include` option
181 has been removed, because it was of little use and blocked other
183 - The `--auto` option is renamed `--annotate`, though the old
184 `--auto=yes`/`--auto=no` forms are still supported.
185 - `cg_diff` and `cg_merge` are now deprecated, because `cg_annotate`
186 now does a better job of diffing and merging.
187 - The Cachegrind output file format has changed very slightly, but in
188 ways nobody is likely to notice.
191 - Valgrind now contains python code that defines GDB callgrind
192 front end monitor commands. See CORE CHANGES.
195 - Valgrind now contains python code that defines GDB massif
196 front end monitor commands. See CORE CHANGES.
199 - A new kind of user request has been added which allows you to
200 override the 1024 byte limit on access count histograms for blocks
201 of memory. The client request is DHAT_HISTOGRAM_MEMORY.
203 * ==================== FIXED BUGS ====================
205 The following bugs have been fixed or resolved. Note that "n-i-bz"
206 stands for "not in bugzilla" -- that is, a bug that was reported to us
207 but never got a bugzilla entry. We encourage you to file bugs in
208 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
209 than mailing the developers (or mailing lists) directly -- bugs that
210 are not entered into bugzilla tend to get forgotten about or ignored.
212 170510 Don't warn about ioctl of size 0 without direction hint
213 241072 List tools in --help output
214 327548 false positive while destroying mutex
215 382034 Testcases build fixes for musl
216 351857 confusing error message about valid command line option
217 374596 inconsistent RDTSCP support on x86_64
218 392331 Spurious lock not held error from inside pthread_cond_timedwait
219 397083 Likely false positive "uninitialised value(s)" for __wmemchr_avx2 and __wmemcmp_avx2_movbe
220 400793 pthread_rwlock_timedwrlock false positive
221 419054 Unhandled syscall getcpu on arm32
222 433873 openat2 syscall unimplemented on Linux
223 434057 Add stdio mode to valgrind's gdbserver
224 435441 valgrind fails to interpose malloc on musl 1.2.2 due to weak symbol name and no libc soname
225 436413 Warn about realloc of size zero
226 439685 compiler warning in callgrind/main.c
227 444110 priv/guest_ppc_toIR.c:36198:31: warning: duplicated 'if' condition.
228 444487 hginfo test detects an extra lock inside data symbol "_rtld_local"
229 444488 Use glibc.pthread.stack_cache_size tunable
230 444568 drd/tests/pth_barrier_thr_cr fails on Fedora 38
231 445743 "The impossible happened: mutex is locked simultaneously by two threads"
232 while using mutexes with priority inheritance and signals
233 449309 Missing loopback device ioctl(s)
234 459476 vgdb: allow address reuse to avoid "address already in use" errorsuse" errors
235 460356 s390: Sqrt32Fx4 -- cannot reduce tree
236 462830 WARNING: unhandled amd64-freebsd syscall: 474
237 463027 broken check for MPX instruction support in assembler
238 464103 Enhancement: add a client request to DHAT to mark memory to be histogrammed
239 464476 Firefox fails to start under Valgrind
240 464609 Valgrind memcheck should support Linux pidfd_open
241 464680 Show issues caused by memory policies like selinux deny_execmem
242 464859 Build failures with GCC-13 (drd tsan_unittest)
243 464969 D language demangling
244 465435 m_libcfile.c:66 (vgPlain_safe_fd): Assertion 'newfd >= VG_(fd_hard_limit)' failed.
245 466104 aligned_alloc problems, part 1
246 467036 Add time cost statistics for Regtest
247 467482 Build failure on aarch64 Alpine
248 467714 fdleak_* and rlimit tests fail when parent process has more than
249 64 descriptors opened
250 467839 Gdbserver: Improve compatibility of library directory name
251 468401 [PATCH] Add a style file for clang-format
252 468556 Build failure for vgdb
253 468606 build: remove "Valgrind relies on GCC" check/output
254 469097 ppc64(be) doesn't support SCV syscall instruction
255 n-i-bz FreeBSD rfork syscall fail with EINVAL or ENOSYS rather than VG_(unimplemented)
257 To see details of a given bug, visit
258 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
259 where XXXXXX is the bug number as listed above.
261 * ==================== KNOWN ISSUES ===================
263 * configure --enable-lto=yes is know to not work in all setups.
264 See bug 469049. Workaround: Build without LTO.
266 (3.21.0.RC1: 14 Apr 2023)
267 (3.21.0.RC2: 21 Apr 2023)
269 Release 3.20.0 (24 Oct 2022)
270 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
272 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
273 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
274 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
275 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
276 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
277 AMD64/macOS 10.13 and nanoMIPS/Linux.
279 * ==================== CORE CHANGES ===================
281 * The option "--vgdb-stop-at=event1,event2,..." accepts the new value abexit.
282 This indicates to invoke gdbserver when your program exits abnormally
283 (i.e. with a non zero exit code).
284 * Fix Rust v0 name demangling.
285 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
286 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
287 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with
289 * The option --enable-debuginfod=<no|yes> [default: yes] has been added on
291 * More DWARF5 support as generated by clang14.
293 * ==================== FIXED BUGS ====================
295 The following bugs have been fixed or resolved. Note that "n-i-bz"
296 stands for "not in bugzilla" -- that is, a bug that was reported to us
297 but never got a bugzilla entry. We encourage you to file bugs in
298 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
299 than mailing the developers (or mailing lists) directly -- bugs that
300 are not entered into bugzilla tend to get forgotten about or ignored.
302 131186 writev reports error in (vector[...])
303 434764 iconv_open causes ld.so v2.28+ to use optimised strncmp
304 446754 Improve error codes from alloc functions under memcheck
305 452274 memcheck crashes with Assertion 'sci->status.what == SsIdle' failed
306 452779 Valgrind fails to build on FreeBSD 13.0 with llvm-devel (15.0.0)
307 453055 shared_timed_mutex drd test fails with "Lock shared failed" message
308 453602 Missing command line option to enable/disable debuginfod
309 452802 Handle lld 9+ split RW PT_LOAD segments correctly
310 454040 s390x: False-positive memcheck:cond in memmem on arch13 systems
311 456171 [PATCH] FreeBSD: Don't record address errors when accessing the 'kern.ps_strings' sysctl struct
312 n-i-bz Implement vgdb invoker on FreeBSD
313 458845 PowerPC: The L field for the dcbf and sync instruction should be
315 458915 Remove register cache to fix 458915 gdbserver causes wrong syscall return
316 459031 Documentation on --error-exitcode incomplete
317 459477 XERROR messages lacks ending '\n' in vgdb
318 462007 Implicit int in none/tests/faultstatus.c
320 To see details of a given bug, visit
321 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
322 where XXXXXX is the bug number as listed above.
324 (3.20.0.RC1: 20 Oct 2022)
327 Release 3.19.0 (11 Apr 2022)
328 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
330 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
331 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
332 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
333 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
334 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
335 AMD64/macOS 10.13 and nanoMIPS/Linux.
337 * ==================== CORE CHANGES ===================
339 * Fix Rust v0 name demangling.
340 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
341 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
342 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with HardenedBSD
344 * ================== PLATFORM CHANGES =================
347 - ignore the "v8.x" architecture levels, only look at actual CPU features
348 present. Fixes mismatch detected between RDMA and atomics features
349 preventing startup on some QEMU configurations.
350 - Implement LD{,A}XP and ST{,L}XP
351 - Fix incorrect code emitted for doubleword CAS.
354 - Fix sys_ipc semtimedop syscall
355 - Fix VFLRX and WFLRX instructions
356 - Fix EXRL instruction with negative offset
359 - Reimplement the vbpermq instruction support to generate less Iops and
360 avoid overflowing internal buffers.
361 - Fix checking for scv support to avoid "Facility 'SCV' unavailable (12),
362 exception" messages in dmsg.
363 - Fix setting condition code for Vector Compare quad word instructions.
364 - Fix fix lxsibzx, lxsihzx and lxsihzx instructions so they only load
365 their respective sized data.
366 - Fix the prefixed stq instruction in PC relative mode.
368 * ==================== TOOL CHANGES ===================
371 - Speed up --track-origins=yes for large (in the range of hundreds to
372 thousands of megabytes) mmap/munmaps.
374 - Several fixes for new versions of libstd++ using new posix try_lock
377 * ==================== FIXED BUGS ====================
379 The following bugs have been fixed or resolved. Note that "n-i-bz"
380 stands for "not in bugzilla" -- that is, a bug that was reported to us
381 but never got a bugzilla entry. We encourage you to file bugs in
382 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
383 than mailing the developers (or mailing lists) directly -- bugs that
384 are not entered into bugzilla tend to get forgotten about or ignored.
386 403802 leak_cpp_interior fails with some reachable blocks different than expected
387 435732 memcheck/tests/leak_cpp_interior fails with gcc11
388 444242 s390x: Valgrind crashes on EXRL with negative offset
389 444399 arm64: unhandled instruction 0xC87F2D89 (LD{,A}XP and ST{,L}XP).
391 444481 gdb_server test failures on s390x
392 444495 dhat/tests/copy fails on s390x
393 444552 memcheck/tests/sem fails on s390x with glibc 2.34
394 444571 PPC, fix the lxsibzx and lxsihzx so they only load their respective
396 444836 PPC, pstq instruction for R=1 is not storing to the correct address.
397 444925 fexecve syscall wrapper not properly implemented
398 445032 valgrind/memcheck crash with SIGSEGV when SIGVTALRM timer used and
400 445211 Fix out of tree builds
401 445300 [PATCH] Fix building tests with Musl
402 445011 SIGCHLD is sent when valgrind uses debuginfod-find
403 445354 arm64 backend: incorrect code emitted for doubleword CAS
404 445415 arm64 front end: alignment checks missing for atomic instructions
405 445504 Using C++ condition_variable results in bogus "mutex is locked simultaneously by two threads" warning
406 445607 Unhandled amd64-freebsd syscall: 247
407 445668 Inline stack frame generation is broken for Rust binaries
408 445916 Demangle Rust v0 symbols with .llvm suffix
409 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false positives
410 446138 DRD/Helgrind with std::timed_mutex::try_lock_until false positives
411 446281 Add a DRD suppression for fwrite
412 446103 Memcheck: `--track-origins=yes` causes extreme slowdowns for large mmap/munmap
413 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false
414 446251 TARGET_SIGNAL_THR added to enum target_signal
415 446823 FreeBSD - missing syscalls when using libzm4
416 447991 s390x: Valgrind indicates illegal instruction on wflrx
417 447995 Valgrind segfault on power10 due to hwcap checking code
418 449483 Powerpc: vcmpgtsq., vcmpgtuq,, vcmpequq. instructions not setting the
419 condition code correctly.
420 449672 ppc64 --track-origins=yes failures because of bad cmov addHRegUse
421 449838 sigsegv liburing the 'impossible' happened for io_uring_setup
422 450025 Powerc: ACC file not implemented as a logical overlay of the VSR
424 450437 Warn for execve syscall with argv or argv[0] being NULL
425 450536 Powerpc: valgrind throws 'facility scv unavailable exception'
426 451626 Syscall param bpf(attr->raw_tracepoint.name) points to unaddressable byte(s)
427 451827 [ppc64le] VEX temporary storage exhausted with several vbpermq instructions
428 451843 valgrind fails to start on a FreeBSD system which enforces W^X
430 To see details of a given bug, visit
431 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
432 where XXXXXX is the bug number as listed above.
434 (3.19.0.RC1: 02 Apr 2022)
435 (3.19.0.RC2: 08 Apr 2022)
438 Release 3.18.0 (15 Oct 2021)
439 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
441 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
442 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
443 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
444 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
445 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
446 AMD64/macOS 10.13 and nanoMIPS/Linux.
448 * ==================== CORE CHANGES ===================
450 * The libiberty demangler has been updated, which brings support for
451 Rust v0 name demangling. [Update: alas, due to a bug, this support
452 isn't working in 3.18.0.]
454 * __libc_freeres isn't called anymore after the program recieves a
455 fatal signal. Causing some internal glibc resources to hang around,
456 but preventing any crashes after the program has ended.
458 * The DWARF reader is now very much faster at startup when just
459 --read-inline-info=yes (the default in most cases) is given.
461 * glibc 2.34, which moved various functions from libpthread.so into
462 libc.so, is now supported.
464 * ================== PLATFORM CHANGES =================
468 - v8.2 scalar and vector FABD, FACGE, FACGT and FADD.
469 - v8.2 FP compare & conditional compare instructions.
470 - Zero variants of v8.2 FP compare instructions.
474 - Support the miscellaneous-instruction-extensions facility 3 and
475 the vector-enhancements facility 2. This enables programs
476 compiled with "-march=arch13" or "-march=z15" to be executed
481 - ISA 3.1 support is now complete
482 - ISA 3.0 support for the darn instruction added.
483 - ISA 3.0 support for the vector system call instruction scv added.
484 - ISA 3.0 support for the copy, paste and cpabort instructions added.
486 * Support for X86/FreeBSD and AMD64/FreeBSD has been added.
488 * ==================== OTHER CHANGES ====================
490 * Memcheck on amd64: minor fixes to remove some false positive
493 * ==================== FIXED BUGS ====================
495 The following bugs have been fixed or resolved. Note that "n-i-bz"
496 stands for "not in bugzilla" -- that is, a bug that was reported to us
497 but never got a bugzilla entry. We encourage you to file bugs in
498 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
499 than mailing the developers (or mailing lists) directly -- bugs that
500 are not entered into bugzilla tend to get forgotten about or ignored.
502 208531 [PATCH]: FreeBSD support for valgrind
503 368960 WARNING: unhandled amd64-linux syscall: 163 (acct)
504 407589 [Linux] Add support for C11 aligned_alloc() and GNU reallocarray()
505 423963 Error in child thread when CLONE_PIDFD is used
506 426148 crash with "impossible happened" when running BPF CO-RE programs
507 429375 PPC ISA 3.1 support is missing, part 9
508 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
509 431306 Update demangler to support Rust v0 name mangling
510 432387 s390x: z15 instructions support
511 433437 FreeBSD support, part 1
512 433438 FreeBSD support, part 2
513 433439 FreeBSD support, part 3
514 433469 FreeBSD support, part 4
515 433473 FreeBSD support, part 5
516 433477 FreeBSD support, part 6
517 433479 FreeBSD support, part 7
518 433504 FreeBSD support, part 8
519 433506 FreeBSD support, part 9
520 433507 FreeBSD support, part 10
521 433508 FreeBSD support, part 11
522 433510 FreeBSD support, part 12
523 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
524 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
525 434296 s390x: False-positive memcheck diagnostics from vector string
527 434840 PPC64 darn instruction not supported
528 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
529 435908 valgrind tries to fetch from deubginfod for files which already
530 have debug information
531 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2
532 439046 valgrind is unusably large when linked with lld
533 439090 Implement close_range(2)
534 439326 Valgrind 3.17.0 won't compile with Intel 2021 oneAPI compilers
535 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
536 440670 unhandled ppc64le-linux syscall: 252 statfs64 and 253 fstatfs64
537 440906 Fix impossible constraint issue in P10 testcase.
538 441512 Remove a unneeded / unnecessary prefix check.
539 441534 Update the expected output for test_isa_3_1_VRT.
540 442061 very slow execution under Fedora 34 (readdwarf3)
541 443031 Gcc -many change requires explicit .machine directives
542 443033 Add support for the ISA 3.0 mcrxrx instruction
543 443034 Sraw, srawi, srad, sradi, mfs
544 443178 Powerpc, test jm-mfspr expected output needs to be updated.
545 443179 Need new test for the lxvx and stxvx instructions on ISA 2.07 and
547 443180 The subnormal test and the ISA 3.0 test generate compiler warnings
548 443314 In the latest GIT version, Valgrind with "--trace-flags" crashes
550 443605 Don't call final_tidyup (__libc_freeres) on FatalSignal
552 To see details of a given bug, visit
553 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
554 where XXXXXX is the bug number as listed below.
556 (3.18.0.RC1: 12 Oct 2021)
557 (3.18.0: 15 Oct 2021)
561 Release 3.17.0 (19 Mar 2021)
562 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
564 3.17.0 fixes a number of bugs and adds some functional changes: support for
565 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
566 some new instructions for Arm64, S390 and POWER. There are also some tool
569 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
570 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
571 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
572 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
573 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
575 * ==================== CORE CHANGES ===================
577 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
580 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
581 debugging information. When a debuginfo file cannot be found locally,
582 Valgrind is able to query debuginfod servers for the file using its
583 build-id. See the user manual for more information about debuginfod support.
585 * ================== PLATFORM CHANGES =================
589 - Inaccuracies resulting from double-rounding in the simulation of
590 floating-point multiply-add/subtract instructions have been fixed. These
591 should now behave exactly as the hardware does.
593 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
594 ongoing. Support for the half-word variants of at least the following
595 instructions has been added:
597 FABS <Vd>.<T>, <Vn>.<T>
599 FNEG <Vd>.<T>, <Vn>.<T>
601 FSQRT <Vd>.<T>, <Vn>.<T>
606 - Implement the new instructions/features that were added to z/Architecture
607 with the vector-enhancements facility 1. Also cover the instructions from
608 the vector-packed-decimal facility that are defined outside the chapter
609 "Vector Decimal Instructions", but not the ones from that chapter itself.
611 For a detailed list of newly supported instructions see the updates to
612 `docs/internals/s390-opcodes.csv'.
614 Since the miscellaneous instruction extensions facility 2 was already
615 added in Valgrind 3.16.0, this completes the support necessary to run
616 general programs built with `--march=z14' under Valgrind. The
617 vector-packed-decimal facility is currently not exploited by the standard
618 toolchain and libraries.
622 - Various bug fixes. Fix for the sync field to limit setting just two of
623 the two bits in the L-field. Fix the write size for the stxsibx and
624 stxsihx instructions. Fix the modsw and modsd instructions.
626 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
627 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
628 conversion instructions are still missing.
630 * ==================== TOOL CHANGES ====================
632 * General tool changes
634 - All the tools and their vgpreload libraries are now installed under
635 libexec because they cannot be executed directly and should be run through
636 the valgrind executable. This should be an internal, not user visible,
637 change, but might impact valgrind packagers.
639 - The --track-fds option now respects -q, --quiet and won't output anything
640 if no file descriptors are leaked. It also won't report the standard stdin
641 (0), stdout (1) or stderr (2) descriptors as being leaked with
642 --trace-fds=yes anymore. To track whether the standard file descriptors
643 are still open at the end of the program run use --trace-fds=all.
647 - DHAT has been extended, with two new modes of operation. The new
648 --mode=copy flag triggers copy profiling, which records calls to memcpy,
649 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
650 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
651 the new dhat/dhat.h file. This is useful for learning more about hot code
652 paths. See the user manual for more information about the new modes.
654 - Because of these changes, DHAT's file format has changed. DHAT output
655 files produced with earlier versions of DHAT will not work with this
656 version of DHAT's viewer, and DHAT output files produced with this version
657 of DHAT will not work with earlier versions of DHAT's viewer.
659 * ==================== FIXED BUGS ====================
661 The following bugs have been fixed or resolved. Note that "n-i-bz"
662 stands for "not in bugzilla" -- that is, a bug that was reported to us
663 but never got a bugzilla entry. We encourage you to file bugs in
664 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
665 than mailing the developers (or mailing lists) directly -- bugs that
666 are not entered into bugzilla tend to get forgotten about or ignored.
668 To see details of a given bug, visit
669 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
670 where XXXXXX is the bug number as listed below.
672 140178 open("/proc/self/exe", ...); doesn't quite work
673 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
674 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
675 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
676 345077 linux syscall execveat support (linux 3.19)
677 361770 Missing F_ADD_SEALS
678 369029 handle linux syscalls sched_getattr and sched_setattr
679 384729 __libc_freeres inhibits cross-platform valgrind
680 388787 Support for C++17 new/delete
681 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
682 396656 Warnings while reading debug info
683 397605 ioctl FICLONE mishandled
684 401416 Compile failure with openmpi 4.0
685 408663 Suppression file for musl libc
686 404076 s390x: z14 vector instructions not implemented
687 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
688 (actually affected all x86 and nanomips regardless of host bitness)
689 413547 regression test does not check for Arm 64 features.
690 414268 Enable AArch64 feature detection and decoding for v8.x instructions
691 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
692 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
693 422261 platform selection fails for unqualified client name
694 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
695 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
696 423195 PPC ISA 3.1 support is missing, part 1
697 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
698 424012 crash with readv/writev having invalid but not NULL arg2 iovec
699 424298 amd64: Implement RDSEED
700 425232 PPC ISA 3.1 support is missing, part 2
701 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
702 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
703 426123 PPC ISA 3.1 support is missing, part 3
704 426144 Fix "condition variable has not been initialized" on Fedora 33.
705 427400 PPC ISA 3.1 support is missing, part 4
706 427401 PPC ISA 3.1 support is missing, part 5
707 427404 PPC ISA 3.1 support is missing, part 6
708 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
709 427787 Support new faccessat2 linux syscall (439)
710 427969 debuginfo section duplicates a section in the main ELF file
711 428035 drd: Unbreak the musl build
712 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
713 428716 cppcheck detects potential leak in VEX/useful/smchash.c
714 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
715 429352 PPC ISA 3.1 support is missing, part 7
716 429354 PPC ISA 3.1 support is missing, part 8
717 429692 unhandled ppc64le-linux syscall: 147 (getsid)
718 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
720 429952 Errors when building regtest with clang
721 430354 ppc stxsibx and stxsihx instructions write too much data
722 430429 valgrind.h doesn't compile on s390x with clang
723 430485 expr_is_guardable doesn't handle Iex_Qop
724 431556 Complete arm64 FADDP v8.2 instruction support
725 432102 Add support for DWARF5 as produced by GCC11
726 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
727 432381 drd: Process STACK_REGISTER client requests
728 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
729 432672 vg_regtest: test-specific environment variables not reset between tests
730 432809 VEX should support REX.W + POPF
731 432861 PPC modsw and modsd give incorrect results for 1 mod 12
732 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
733 432215 Add debuginfod functionality
734 433323 Use pkglibexecdir as vglibdir
735 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
736 433629 valgrind/README has type "abd" instead of "and"
737 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
738 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
739 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
740 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
741 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
743 (3.17.0.RC1: 13 Mar 2021)
744 (3.17.0.RC2: 17 Mar 2021)
745 (3.17.0: 19 Mar 2021)
749 Release 3.16.1 (22 June 2020)
750 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
752 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
753 fixes character encoding problems in the documentation HTML.
755 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
756 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
758 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
762 Release 3.16.0 (27 May 2020)
763 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
765 3.16.0 is a feature release with many improvements and the usual collection of
768 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
769 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
770 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
771 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
772 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
774 * ==================== CORE CHANGES ===================
776 * It is now possible to dynamically change the value of many command line
777 options while your program (or its children) are running under Valgrind.
779 To see the list of dynamically changeable options, run
780 "valgrind --help-dyn-options".
782 You can change the options from the shell by using vgdb to launch
783 the monitor command "v.clo <clo option>...".
784 The same monitor command can be used from a gdb connected
785 to the valgrind gdbserver.
786 Your program can also change the dynamically changeable options using
787 the client request VALGRIND_CLO_CHANGE(option).
789 * ================== PLATFORM CHANGES =================
791 * MIPS: preliminary support for nanoMIPS instruction set has been added.
793 * ==================== TOOL CHANGES ====================
797 - The implicit memcpy done by each call to realloc now counts towards the
798 read and write counts of resized heap blocks, making those counts higher
803 - cg_annotate's --auto and --show-percs options now default to 'yes', because
804 they are usually wanted.
808 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
809 because they are usually wanted.
811 - The command option --collect-systime has been enhanced to specify
812 the unit used to record the elapsed time spent during system calls.
813 The command option now accepts the values no|yes|msec|usec|nsec,
814 where yes is a synonym of msec. When giving the value nsec, the
815 system cpu time of system calls is also recorded.
819 - Several memcheck options are now dynamically changeable.
820 Use valgrind --help-dyn-options to list them.
822 - The release 3.15 introduced a backward incompatible change for
823 some suppression entries related to preadv and pwritev syscalls.
824 When reading a suppression entry using the unsupported 3.14 format,
825 valgrind will now produce a warning to say the suppression entry will not
826 work, and suggest the needed change.
828 - Significantly fewer false positive errors on optimised code generated by
829 Clang and GCC. In particular, Memcheck now deals better with the
830 situation where the compiler will transform C-level "A && B" into "B && A"
831 under certain circumstances (in which the transformation is valid).
832 Handling of integer equality/non-equality checks on partially defined
833 values is also improved on some architectures.
837 - The exprimental Stack and Global Array Checking tool has been removed.
838 It only ever worked on x86 and amd64, and even on those it had a
839 high false positive rate and was slow. An alternative for detecting
840 stack and global array overruns is using the AddressSanitizer (ASAN)
841 facility of the GCC and Clang compilers, which require you to rebuild
842 your code with -fsanitize=address.
844 * ==================== OTHER CHANGES ====================
846 * New and modified GDB server monitor features:
848 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
850 - The gdbserver monitor commands that require an address and an optional
851 length argument now accepts the alternate 'C like' syntax "address[length]".
852 For example, the memcheck command "monitor who_points_at 0x12345678 120"
853 can now also be given as "monitor who_points_at 0x12345678[120]".
855 * ==================== FIXED BUGS ====================
857 The following bugs have been fixed or resolved. Note that "n-i-bz"
858 stands for "not in bugzilla" -- that is, a bug that was reported to us
859 but never got a bugzilla entry. We encourage you to file bugs in
860 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
861 than mailing the developers (or mailing lists) directly -- bugs that
862 are not entered into bugzilla tend to get forgotten about or ignored.
864 To see details of a given bug, visit
865 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
866 where XXXXXX is the bug number as listed below.
868 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
869 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
870 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
871 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
872 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
873 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
874 400872 Add nanoMIPS support to Valgrind
875 403212 drd/tests/trylock hangs on FreeBSD
876 404406 s390x: z14 miscellaneous instructions not implemented
877 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
878 406561 mcinfcallWSRU gdbserver_test fails on ppc64
879 406824 Unsupported baseline
880 407218 Add support for the copy_file_range syscall
881 407307 Intercept stpcpy also in ld.so for arm64
882 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
884 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
885 408009 Expose rdrand and f16c even on avx if host cpu supports them
886 408091 Missing pkey syscalls
887 408414 Add support for missing for preadv2 and pwritev2 syscalls
888 409141 Valgrind hangs when SIGKILLed
889 409206 Support for Linux PPS and PTP ioctls
890 409367 exit_group() after signal to thread waiting in futex() causes hangs
891 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
892 409780 References to non-existent configure.in
893 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
894 410599 Non-deterministic behaviour of pth_self_kill_15_other test
895 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
896 411134 Allow the user to change a set of command line options during execution
897 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
898 412344 Problem setting mips flags with specific paths
899 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
900 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
901 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
902 413603 callgrind_annotate/cg_annotate truncate function names at '#'
903 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
904 415136 ARMv8.1 Compare-and-Swap instructions are not supported
905 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
906 416239 valgrind crashes when handling clock_adjtime
907 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
908 416286 DRD reports "conflicting load" error on std::mutex::lock()
909 416301 s390x: "compare and signal" not supported
910 416387 finit_module and bpf syscalls are unhandled on arm64
911 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
912 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
913 416753 new 32bit time syscalls for 2038+
914 417075 pwritev(vector[...]) suppression ignored
915 417075 is not fixed, but incompatible supp entries are detected
916 and a warning is produced for these.
917 417187 [MIPS] Conditional branch problem since 'grail' changes
918 417238 Test memcheck/tests/vbit-test fails on mips64 BE
919 417266 Make memcheck/tests/linux/sigqueue usable with musl
920 417281 s390x: /bin/true segfaults with "grail" enabled
921 417427 commit to fix vki_siginfo_t definition created numerous regression
923 417452 s390_insn_store_emit: dst->tag for HRcVec128
924 417578 Add suppressions for glibc DTV leaks
925 417906 clone with CLONE_VFORK and no CLONE_VM fails
926 418004 Grail code additions break ppc64.
927 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
928 418997 s390x: Support Iex_ITE for float and vector types
929 419503 s390x: Avoid modifying registers returned from isel functions
930 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
931 421570 std_mutex fails on Arm v8.1 h/w
932 434035 vgdb might crash if valgrind is killed
933 n-i-bz Fix minor one time leaks in dhat.
934 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
935 n-i-bz Add support for the Linux io_uring system calls
936 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
937 n-i-bz Fix non-glibc build of test suite with s390x_features
938 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
939 423195 PPC ISA 3.1 support is missing, part 1
941 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
942 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
943 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
947 Release 3.15.0 (12 April 2019)
948 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
950 3.15.0 is a feature release with many improvements and the usual collection of
953 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
954 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
955 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
956 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
957 support for X86/macOS 10.13 and AMD64/macOS 10.13.
959 * ==================== CORE CHANGES ===================
961 * The XTree Massif output format now makes use of the information obtained
962 when specifying --read-inline-info=yes.
964 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
966 * ==================== TOOL CHANGES ====================
970 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
971 it has been promoted from an experimental tool to a regular tool. Run it
972 with --tool=dhat instead of --tool=exp-dhat.
974 - DHAT now prints only minimal data when the program ends, instead writing
975 the bulk of the profiling data to a file. As a result, the --show-top-n
976 and --sort-by options have been removed.
978 - Profile results can be viewed with the new viewer, dh_view.html. When
979 a run ends, a short message is printed, explaining how to view the result.
981 - See the documentation for more details.
985 - cg_annotate has a new option, --show-percs, which prints percentages next
990 - callgrind_annotate has a new option, --show-percs, which prints percentages
991 next to all event counts.
993 - callgrind_annotate now inserts commas in call counts, and
994 sort the caller/callee lists in the call tree.
998 - The default value for --read-inline-info is now "yes" on
999 Linux/Android/Solaris. It is still "no" on other OS.
1003 - The option --xtree-leak=yes (to output leak result in xtree format)
1004 automatically activates the option --show-leak-kinds=all, as xtree
1005 visualisation tools such as kcachegrind can in any case select what kind
1006 of leak to visualise.
1008 - There has been further work to avoid false positives. In particular,
1009 integer equality on partially defined inputs (C == and !=) is now handled
1012 * ==================== OTHER CHANGES ====================
1014 * The new option --show-error-list=no|yes displays, at the end of the run, the
1015 list of detected errors and the used suppressions. Prior to this change,
1016 showing this information could only be done by specifying "-v -v", but that
1017 also produced a lot of other possibly-non-useful messages. The option -s is
1018 equivalent to --show-error-list=yes.
1020 * ==================== FIXED BUGS ====================
1022 The following bugs have been fixed or resolved. Note that "n-i-bz"
1023 stands for "not in bugzilla" -- that is, a bug that was reported to us
1024 but never got a bugzilla entry. We encourage you to file bugs in
1025 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1026 than mailing the developers (or mailing lists) directly -- bugs that
1027 are not entered into bugzilla tend to get forgotten about or ignored.
1029 To see details of a given bug, visit
1030 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1031 where XXXXXX is the bug number as listed below.
1033 385411 s390x: z13 vector floating-point instructions not implemented
1034 397187 z13 vector register support for vgdb gdbserver
1035 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
1036 398870 Please add support for instruction vcvtps2ph
1037 399287 amd64 front end: Illegal Instruction vcmptrueps
1038 399301 Use inlined frames in Massif XTree output.
1039 399322 Improve callgrind_annotate output
1040 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
1041 400164 helgrind test encounters mips x-compiler warnings and assembler error
1042 400490 s390x: VRs allocated as if separate from FPRs
1043 400491 s390x: Operand of LOCH treated as unsigned integer
1044 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
1045 options, which specify a mips64 processor
1046 401112 LLVM 5.0 generates comparison against partially initialized data
1047 401277 More bugs in z13 support
1048 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
1049 401578 drd: crashes sometimes on fork()
1050 401627 memcheck errors with glibc avx2 optimized wcsncmp
1051 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
1052 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
1053 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
1055 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
1056 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
1057 402123 invalid assembler opcodes for mips32r2
1058 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
1059 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
1060 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
1061 402351 mips64 libvexmultiarch_test fails on s390x
1062 402369 Overhaul DHAT
1063 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
1064 402480 Do not use %rsp in clobber list
1065 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
1066 402515 Implement new option --show-error-list=no|yes / -s
1067 402519 POWER 3.0 addex instruction incorrectly implemented
1068 402781 Redo the cache used to process indirect branch targets
1069 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
1070 403552 s390x: wrong facility bit checked for vector facility
1071 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
1072 404638 Add VG_(replaceIndexXA)
1073 404843 s390x: backtrace sometimes ends prematurely
1074 404888 autotools cleanup series
1075 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
1076 405182 Valgrind fails to build with Clang
1077 405205 filter_libc: remove the line holding the futex syscall error entirely
1078 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
1079 the upper and lower 32-bits of the 64-bit result
1080 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
1081 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
1082 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
1084 405403 s390x disassembler cannot be used on x86
1085 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
1086 405458 MIPS mkFormVEC arguments swapped?
1087 405716 drd: Fix an integer overflow in the stack margin calculation
1088 405722 Support arm64 core dump
1089 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
1090 of the 64-bit destination field.
1091 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
1092 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
1093 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
1095 406256 PPC64, vector floating point instructions don't handle subnormal
1096 according to VSCR[NJ] bit setting.
1097 406352 cachegrind/callgrind fails ann tests because of missing a.c
1098 406354 dhat is broken on x86 (32bit)
1099 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
1100 406357 gdbserver_tests fails because of gdb output change
1101 406360 memcheck/tests/libstdc++.supp needs more supression variants
1102 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
1103 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
1104 407340 PPC64, does not support the vlogefp, vexptefp instructions.
1105 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
1106 n-i-bz Fix callgrind_annotate non deterministic order for equal total
1107 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
1108 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
1109 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
1111 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
1112 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
1113 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
1117 Release 3.14.0 (9 October 2018)
1118 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1120 3.14.0 is a feature release with many improvements and the usual collection of
1123 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1124 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1125 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1126 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
1127 support for X86/macOS 10.13, AMD64/macOS 10.13.
1129 * ==================== CORE CHANGES ===================
1131 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
1132 debug info for unloaded code. This allows saved stack traces (e.g. for
1133 memory leaks) to include file/line info for code that has been dlclose'd (or
1134 similar). See the user manual for more information and known limitations.
1136 * Ability to specify suppressions based on source file name and line number.
1138 * Majorly overhauled register allocator. No end-user changes, but the JIT
1139 generates code a bit more quickly now.
1141 * ================== PLATFORM CHANGES =================
1143 * Preliminary support for macOS 10.13 has been added.
1145 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
1147 * mips: support for MIPS SIMD architecture (MSA) has been added.
1149 * mips: support for MIPS N32 ABI has been added.
1151 * s390: partial support for vector instructions (integer and string) has been
1154 * ==================== TOOL CHANGES ====================
1156 * Helgrind: Addition of a flag
1157 --delta-stacktrace=no|yes [yes on linux amd64/x86]
1158 which specifies how full history stack traces should be computed.
1159 Setting this to =yes can speed up Helgrind by 25% when using
1160 --history-level=full.
1162 * Memcheck: reduced false positive rate for optimised code created by Clang 6
1163 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
1164 blocks more carefully to determine where it can avoid expensive definedness
1165 checks without loss of precision. This is controlled by the flag
1166 --expensive-definedness-checks=no|auto|yes [auto].
1168 * ==================== OTHER CHANGES ====================
1170 * Valgrind is now buildable with link-time optimisation (LTO). A new
1171 configure option --enable-lto=yes allows building Valgrind with LTO. If the
1172 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
1173 Note that if you are doing Valgrind development, --enable-lto=yes massively
1174 slows down the build process.
1176 * ==================== FIXED BUGS ====================
1178 The following bugs have been fixed or resolved. Note that "n-i-bz"
1179 stands for "not in bugzilla" -- that is, a bug that was reported to us
1180 but never got a bugzilla entry. We encourage you to file bugs in
1181 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1182 than mailing the developers (or mailing lists) directly -- bugs that
1183 are not entered into bugzilla tend to get forgotten about or ignored.
1185 To see details of a given bug, visit
1186 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1187 where XXXXXX is the bug number as listed below.
1189 79362 Debug info is lost for .so files when they are dlclose'd
1190 208052 strlcpy error when n = 0
1191 255603 exp-sgcheck Assertion '!already_present' failed
1192 338252 building valgrind with -flto (link time optimisation) fails
1193 345763 MIPS N32 ABI support
1194 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
1195 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
1196 372347 Replacement problem of the additional c++14/c++17 new/delete operators
1197 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1198 376257 helgrind history full speed up using a cached stack
1199 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
1201 379748 Fix missing pselect syscall (OS X 10.11)
1202 379754 Fix missing syscall ulock_wait (OS X 10.12)
1203 380397 s390x: __GI_strcspn() replacemenet needed
1204 381162 possible array overrun in VEX register allocator
1205 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
1206 381274 powerpc too chatty even with --sigill-diagnostics=no
1207 381289 epoll_pwait can have a NULL sigmask
1208 381553 VEX register allocator v3
1209 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
1210 381769 Use ucontext_t instead of struct ucontext
1211 381805 arm32 needs ld.so index hardwire for new glibc security fixes
1212 382256 gz compiler flag test doesn't work for gold
1213 382407 vg_perf needs "--terse" command line option
1214 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
1215 382563 MIPS MSA ASE support
1216 382998 xml-socket doesn't work
1217 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
1218 383723 Fix missing kevent_qos syscall (macOS 10.11)
1219 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
1220 384096 Mention AddrCheck at Memcheck's command line option [..]
1221 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
1222 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
1223 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
1224 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
1225 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
1226 384337 performance improvements to VEX register allocator v2 and v3
1227 384526 reduce number of spill insns generated by VEX register allocator v3
1228 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
1229 384631 Sanitise client args as printed with -v
1230 384633 Add a simple progress-reporting facility
1231 384987 VEX regalloc: allocate caller-save registers for short lived vregs
1232 385055 PPC VEX temporary storage exhausted
1233 385182 PPC64 is missing support for the DSCR
1234 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
1235 385207 PPC64, generate_store_FPRF() generates too many Iops
1236 385208 PPC64, xxperm instruction exhausts temporary memory
1237 385210 PPC64, vpermr instruction could exhaust temporary memory
1238 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
1239 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
1240 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
1241 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
1242 385408 s390x: z13 vector "support" instructions not implemented
1243 385409 s390x: z13 vector integer instructions not implemented
1244 385410 s390x: z13 vector string instructions not implemented
1245 385412 s390x: new non-vector z13 instructions not implemented
1246 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
1247 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
1248 385939 Optionally exit on the first error
1249 386318 valgrind.org/info/tools.html is missing SGCheck
1250 386425 running valgrind + wine on armv7l gives illegal opcode
1251 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
1252 387410 MIPSr6 support
1253 387664 Memcheck: make expensive-definedness-checks be the default
1254 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
1255 387766 asm shifts cause false positive "Conditional jump or move depends
1256 on uninitialised value"
1257 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
1258 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
1259 388786 Support bpf syscall in amd64 Linux
1260 388862 Add replacements for wmemchr and wcsnlen on Linux
1261 389065 valgrind meets gcc flag -Wlogical-op
1262 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
1263 390471 suppression by specification of source-file line number
1264 390723 make xtree dump files world wide readable, similar to log files
1265 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
1266 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
1267 392118 unhandled amd64-linux syscall: 332 (statx)
1268 392449 callgrind not clearing the number of calls properly
1269 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
1270 lxssp, stxssp and stxvl instructions.
1271 393023 callgrind_control risks using the wrong vgdb
1272 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
1273 393099 posix_memalign() invalid write if alignment == 0
1274 393146 failing assert "is_DebugInfo_active(di)"
1275 395709 PPC64 is missing support for the xvnegsp instruction
1276 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
1278 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
1279 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
1280 396839 s390x: Trap instructions not implemented
1281 396887 arch_prctl should return EINVAL on unknown option
1282 == 397286 crash before launching binary (Unsupported arch_prctl option)
1283 == 397393 valgrind: the 'impossible' happened: (Archlinux)
1284 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
1285 396906 compile tests failure on mips32-linux: broken inline asm in tests on
1287 397012 glibc ld.so uses arch_prctl on i386
1288 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
1289 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
1290 397424 glibc 2.27 and gdb_server tests
1291 398028 Assertion `cfsi_fits` failing in simple C program
1292 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
1294 n-i-bz Fix missing workq_ops operations (macOS)
1295 n-i-bz fix bug in strspn replacement
1296 n-i-bz Add support for the Linux BLKFLSBUF ioctl
1297 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
1298 n-i-bz Fix possible stack trashing by semctl syscall wrapping
1299 n-i-bz Add support for the Linux membarrier() system call
1300 n-i-bz x86 front end: recognise and handle UD2 correctly
1301 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
1302 correctly aligned before entering the handler.
1304 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
1305 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
1306 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
1310 Release 3.13.0 (15 June 2017)
1311 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1313 3.13.0 is a feature release with many improvements and the usual collection of
1316 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1317 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1318 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1319 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
1321 * ==================== CORE CHANGES ===================
1323 * The translation cache size has been increased to keep up with the demands of
1324 large applications. The maximum number of sectors has increased from 24 to
1325 48. The default number of sectors has increased from 16 to 32 on all
1326 targets except Android, where the increase is from 6 to 12.
1328 * The amount of memory that Valgrind can use has been increased from 64GB to
1329 128GB. In particular this means your application can allocate up to about
1330 60GB when running on Memcheck.
1332 * Valgrind's default load address has been changed from 0x3800'0000 to
1333 0x5800'0000, so as to make it possible to load larger executables. This
1334 should make it possible to load executables of size at least 1200MB.
1336 * A massive spaceleak caused by reading compressed debuginfo files has been
1337 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
1340 * The C++ demangler has been updated.
1342 * Support for demangling Rust symbols has been added.
1344 * A new representation of stack traces, the "XTree", has been added. An XTree
1345 is a tree of stacktraces with data associated with the stacktraces. This is
1346 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
1347 consumption of your program. Reporting is controlled by the new options
1348 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
1350 A report can also be produced on demand using the gdbserver monitor command
1351 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
1352 format' and 'massif format. The existing visualisers for these formats (e.g.
1353 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
1354 analyse these reports.
1356 Memcheck can also produce XTree leak reports using the Callgrind file
1357 format. For more details, see the user manual.
1359 * ================== PLATFORM CHANGES =================
1361 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
1363 * amd64: fixes for JIT failure problems on long AVX2 code blocks
1365 * amd64 and x86: support for CET prefixes has been added
1367 * arm32: a few missing ARMv8 instructions have been implemented
1369 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
1370 Store-Conditional instructions has been added. This is to deal with
1371 processor implementations that implement the LL/SC specifications strictly
1372 and as a result cause Valgrind to hang in certain situations. The
1373 alternative implementation is automatically enabled at startup, as required.
1374 You can use the option --sim-hints=fallback-llsc to force-enable it if you
1377 * Support for OSX 10.12 has been improved.
1379 * On Linux, clone handling has been improved to honour CLONE_VFORK that
1380 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
1381 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
1382 CLONE_VM exact semantics will (still) not work.
1384 * The TileGX/Linux port has been removed because it appears to be both unused
1387 * ==================== TOOL CHANGES ====================
1391 - Memcheck should give fewer false positives when running optimised
1392 Clang/LLVM generated code.
1394 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1396 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
1397 to produce the end of execution leak report in a xtree callgrind format
1400 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
1401 the leak report in an xtree file.
1405 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1407 - For some workloads (typically, for big applications), Massif memory
1408 consumption and CPU consumption has decreased significantly.
1412 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1414 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
1415 for Ada gnat compiled applications.
1417 * ==================== OTHER CHANGES ====================
1419 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
1420 append the inner guest stacktrace to the inner host stacktrace. This helps
1421 to investigate the errors reported by the outer, when they are caused by the
1422 inner guest program (such as an inner regtest). See README_DEVELOPERS for
1425 * To allow fast detection of callgrind files by desktop environments and file
1426 managers, the format was extended to have an optional first line that
1427 uniquely identifies the format ("# callgrind format"). Callgrind creates
1428 this line now, as does the new xtree functionality.
1430 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
1431 have a new %n format letter that is replaced by a sequence number.
1433 * "--version -v" now shows the SVN revision numbers from which Valgrind was
1436 * ==================== FIXED BUGS ====================
1438 The following bugs have been fixed or resolved. Note that "n-i-bz"
1439 stands for "not in bugzilla" -- that is, a bug that was reported to us
1440 but never got a bugzilla entry. We encourage you to file bugs in
1441 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1442 than mailing the developers (or mailing lists) directly -- bugs that
1443 are not entered into bugzilla tend to get forgotten about or ignored.
1445 To see details of a given bug, visit
1446 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1447 where XXXXXX is the bug number as listed below.
1449 162848 --log-file output isn't split when a program forks
1450 340777 Illegal instruction on mips (ar71xx)
1451 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
1452 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
1453 to a different stack.
1454 344139 x86 stack-seg overrides, needed by the Wine people
1455 344524 store conditional of guest applications always fail - observed on
1457 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
1458 352395 Please provide SVN revision info in --version -v
1459 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
1460 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
1461 INVALID_POSIX_THREADID' failed
1462 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
1463 barrier implementation
1464 358697 valgrind.h: Some code remains even when defining NVALGRIND
1465 359202 Add musl libc configure/compile
1466 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
1467 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
1468 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
1469 362223 assertion failed when .valgrindrc is a directory instead of a file
1470 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
1471 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
1472 368507 can't malloc chunks larger than about 34GB
1473 368529 Android arm target link error, missing atexit and pthread_atfork
1474 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
1475 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
1476 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
1477 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
1478 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
1479 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
1480 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
1481 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
1482 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
1483 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
1484 370028 Reduce the number of compiler warnings on MIPS platforms
1485 370635 arm64 missing syscall getcpu
1486 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
1487 371227 Clean AArch64 syscall table
1488 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
1489 371471 Valgrind complains about non legit memory leaks on placement new (C++)
1490 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
1491 371503 disInstr(arm64): unhandled instruction 0xF89F0000
1492 371869 support '%' in symbol Z-encoding
1493 371916 execution tree xtree concept
1494 372120 c++ demangler demangles symbols which are not c++
1495 372185 Support of valgrind on ARMv8 with 32 bit executable
1496 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1497 372195 Power PC, xxsel instruction is not always recognized.
1498 372504 Hanging on exit_group
1499 372600 process loops forever when fatal signals are arriving quickly
1500 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1501 373046 Stacks registered by core are never deregistered
1502 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1503 373086 Implement additional Xen hypercalls
1504 373192 Calling posix_spawn in glibc 2.24 completely broken
1505 373488 Support for fanotify API on ARM64 architecture
1506 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1507 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1508 373938 const IRExpr arguments for matchIRExpr()
1509 374719 some spelling fixes
1510 374963 increase valgrind's load address to prevent mmap failure
1511 375514 valgrind_get_tls_addr() does not work in case of static TLS
1512 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1513 for ML_(find_rx_mapping)()
1514 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1515 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1516 == 377159 "vex: the `impossible' happened" still present
1517 == 375150 Assertion 'tres.status == VexTransOK' failed
1518 == 378068 valgrind crashes on AVX2 function in FFmpeg
1519 376142 Segfaults on MIPS Cavium Octeon boards
1520 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1521 376455 Solaris: unhandled syscall lgrpsys(180)
1522 376518 Solaris: unhandled fast trap getlgrp(6)
1523 376611 ppc64 and arm64 don't know about prlimit64 syscall
1524 376729 PPC64, remove R2 from the clobber list
1526 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1527 to be wrongly marked as addressable
1528 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1529 PIE enabled by default
1530 377376 memcheck/tests/linux/getregset fails with glibc2.24
1531 377427 PPC64, lxv instruction failing on odd destination register
1532 377478 PPC64: ISA 3.0 setup fixes
1533 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1534 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1535 and 2 args for FUTEX_TRYLOCK_PI
1536 377717 Fix massive space leak when reading compressed debuginfo sections
1537 377891 Update Xen 4.6 domctl wrappers
1538 377930 fcntl syscall wrapper is missing flock structure check
1539 378524 libvexmultiarch_test regression on s390x and ppc64
1540 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1541 378673 Update libiberty demangler
1542 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1543 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1544 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1545 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1546 (task_register_dyld_image_infos)
1547 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1548 (task_register_dyld_shared_cache_image_info)
1549 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1550 379473 MIPS: add support for rdhwr cycle counter register
1551 379504 remove TileGX/Linux port
1552 379525 Support more x86 nop opcodes
1553 379838 disAMode(x86): not an addr!
1554 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1555 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1556 379895 clock_gettime does not execute POST syscall wrapper
1557 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1558 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1559 380200 xtree generated callgrind files refer to files without directory name
1560 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1561 380397 s390x: __GI_strcspn() replacement needed
1562 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1564 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1565 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1566 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1570 Release 3.12.0 (20 October 2016)
1571 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1573 3.12.0 is a feature release with many improvements and the usual
1574 collection of bug fixes.
1576 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1577 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1578 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1579 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1580 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1581 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1583 * ================== PLATFORM CHANGES =================
1585 * POWER: Support for ISA 3.0 has been added
1587 * mips: support for O32 FPXX ABI has been added.
1588 * mips: improved recognition of different processors
1589 * mips: determination of page size now done at run time
1591 * amd64: Partial support for AMD FMA4 instructions.
1593 * arm, arm64: Support for v8 crypto and CRC instructions.
1595 * Improvements and robustification of the Solaris port.
1597 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1599 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1600 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1601 where possible. Valgrind's support for 32-bit x86 has stagnated in
1602 recent years and has fallen far behind that for 64-bit x86
1603 instructions. By contrast 64-bit x86 is well supported, up to and
1606 * ==================== TOOL CHANGES ====================
1610 - Added meta mempool support for describing a custom allocator which:
1611 - Auto-frees all chunks assuming that destroying a pool destroys all
1613 - Uses itself to allocate other memory blocks
1615 - New flag --ignore-range-below-sp to ignore memory accesses below
1616 the stack pointer, if you really have to. The related flag
1617 --workaround-gcc296-bugs=yes is now deprecated. Use
1618 --ignore-range-below-sp=1024-1 as a replacement.
1622 - Improved thread startup time significantly on non-Linux platforms.
1626 - Added collection of the metric "tot-blocks-allocd"
1628 * ==================== OTHER CHANGES ====================
1630 * Replacement/wrapping of malloc/new related functions is now done not just
1631 for system libraries by default, but for any globally defined malloc/new
1632 related function (both in shared libraries and statically linked alternative
1633 malloc implementations). The dynamic (runtime) linker is excluded, though.
1634 To only intercept malloc/new related functions in
1635 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1636 "nouserintercepts" can be any non-existing library name).
1637 This new functionality is not implemented for MacOS X.
1639 * The maximum number of callers in a suppression entry is now equal to
1640 the maximum size for --num-callers (500).
1641 Note that --gen-suppressions=yes|all similarly generates suppressions
1642 containing up to --num-callers frames.
1644 * New and modified GDB server monitor features:
1646 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1647 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1650 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1651 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1654 * Valgrind is able to read compressed debuginfo sections in two formats:
1655 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1656 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1658 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1659 for the most common use case (x86_64-linux, Memcheck) has been
1662 * Improved performance for programs that do a lot of discarding of
1663 instruction address ranges of 8KB or less.
1665 * The C++ symbol demangler has been updated.
1667 * More robustness against invalid syscall parameters on Linux.
1669 * ==================== FIXED BUGS ====================
1671 The following bugs have been fixed or resolved. Note that "n-i-bz"
1672 stands for "not in bugzilla" -- that is, a bug that was reported to us
1673 but never got a bugzilla entry. We encourage you to file bugs in
1674 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1675 than mailing the developers (or mailing lists) directly -- bugs that
1676 are not entered into bugzilla tend to get forgotten about or ignored.
1678 To see details of a given bug, visit
1679 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1680 where XXXXXX is the bug number as listed below.
1682 191069 Exiting due to signal not reported in XML output
1683 199468 Suppressions: stack size limited to 25
1684 while --num-callers allows more frames
1685 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1686 278744 cvtps2pd with redundant RexW
1687 303877 valgrind doesn't support compressed debuginfo sections.
1688 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1689 348345 Assertion fails for negative lineno
1690 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1691 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1692 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1693 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1694 352197 mips: mmap2() not wrapped correctly for page size > 4096
1695 353083 arm64 doesn't implement various xattr system calls
1696 353084 arm64 doesn't support sigpending system call
1697 353137 www: update info for Supported Platforms
1698 353138 www: update "The Valgrind Developers" page
1699 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1702 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1703 353398 WARNING: unhandled amd64-solaris syscall: 207
1704 353660 XML in auxwhat tag not escaping reserved symbols properly
1705 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1706 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1707 353802 ELF debug info reader confused with multiple .rodata sections
1708 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1709 353917 unhandled amd64-solaris syscall fchdir(120)
1710 353920 unhandled amd64-solaris syscall: 170
1711 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1712 354392 unhandled amd64-solaris syscall: 171
1713 354797 Vbit test does not include Iops for Power 8 instruction support
1714 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1718 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1719 355188 valgrind should intercept all malloc related global functions
1720 355454 do not intercept malloc related symbols from the runtime linker
1721 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1722 356044 Dwarf line info reader misinterprets is_stmt register
1723 356112 mips: replace addi with addiu
1724 356393 valgrind (vex) crashes because isZeroU happened
1727 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1728 356678 arm64-linux: unhandled syscall 232 (mincore)
1729 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1730 356823 Unsupported ARM instruction: stlex
1731 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1732 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1733 357673 crash if I try to run valgrind with a binary link with libcurl
1734 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1735 357871 pthread_spin_destroy not properly wrapped
1736 357887 Calls to VG_(fclose) do not close the file descriptor
1737 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1738 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1739 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1740 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1741 359181 Buffer Overflow during Demangling
1742 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1743 359289 s390x: popcnt (B9E1) not implemented
1744 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1745 359503 Add missing syscalls for aarch64 (arm64)
1746 359645 "You need libc6-dbg" help message could be more helpful
1747 359703 s390: wire up separate socketcalls system calls
1748 359724 getsockname might crash - deref_UInt should call safe_to_deref
1749 359733 amd64 implement ld.so strchr/index override like x86
1750 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1751 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1753 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1754 359871 Incorrect mask handling in ppoll
1755 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1756 360008 Contents of Power vr registers contents is not printed correctly when
1757 the --vgdb-shadow-registers=yes option is used
1758 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1759 non-zero shadow bits
1760 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1761 360425 arm64 unsupported instruction ldpsw
1763 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1764 360571 Error about the Android Runtime reading below the stack pointer on ARM
1765 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1766 360749 kludge for multiple .rodata sections on Solaris no longer needed
1767 360752 raise the number of reserved fds in m_main.c from 10 to 12
1768 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1769 361226 s390x: risbgn (EC59) not implemented
1770 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1771 361354 ppc64[le]: wire up separate socketcalls system calls
1772 361615 Inconsistent termination for multithreaded process terminated by signal
1773 361926 Unhandled Solaris syscall: sysfs(84)
1774 362009 V dumps core on unimplemented functionality before threads are created
1775 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1776 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1777 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1778 362953 Request for an update to the Valgrind Developers page
1779 363680 add renameat2() support
1780 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1781 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1782 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1783 364058 clarify in manual limitations of array overruns detections
1784 364413 pselect sycallwrapper mishandles NULL sigmask
1785 364728 Power PC, missing support for several HW registers in
1786 get_otrack_shadow_offset_wrk()
1787 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1788 365273 Invalid write to stack location reported after signal handler runs
1789 365912 ppc64BE segfault during jm-insns test (RELRO)
1790 366079 FPXX Support for MIPS32 Valgrind
1791 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1792 366344 Multiple unhandled instruction for Aarch64
1793 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1794 367995 Integration of memcheck with custom memory allocator
1795 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1796 368412 False positive result for altivec capability check
1797 368416 Add tc06_two_races_xml.exp output for ppc64
1798 368419 Perf Events ioctls not implemented
1799 368461 mmapunmap test fails on ppc64
1800 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1801 369000 AMD64 fma4 instructions unsupported.
1802 369169 ppc64 fails jm_int_isa_2_07 test
1803 369175 jm_vec_isa_2_07 test crashes on ppc64
1804 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1805 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1806 369359 msghdr_foreachfield can crash when handling bad iovec
1807 369360 Bad sigprocmask old or new sets can crash valgrind
1808 369361 vmsplice syscall wrapper crashes on bad iovec
1809 369362 Bad sigaction arguments crash valgrind
1810 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1811 369402 Bad set/get_thread_area pointer crashes valgrind
1812 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1813 369446 valgrind crashes on unknown fcntl command
1814 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1815 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1816 370265 ISA 3.0 HW cap stuff needs updating
1817 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1818 372195 Power PC, xxsel instruction is not always recognized
1820 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1821 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1822 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1823 n-i-bz Fix ppoll_alarm exclusion on OS X
1824 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1825 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1826 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1827 n-i-bz amd64: memcheck false positive with shr %edx
1828 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1829 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1830 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1831 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1833 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1834 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1835 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1839 Release 3.11.0 (22 September 2015)
1840 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1842 3.11.0 is a feature release with many improvements and the usual
1843 collection of bug fixes.
1845 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1846 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1847 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1848 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1849 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1850 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1852 * ================== PLATFORM CHANGES =================
1854 * Support for Solaris/x86 and Solaris/amd64 has been added.
1856 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1858 * Preliminary support for the Tilera TileGX architecture has been added.
1860 * s390x: It is now required for the host to have the "long displacement"
1861 facility. The oldest supported machine model is z990.
1863 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1864 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1866 * The JIT's register allocator is significantly faster, making the JIT
1867 as a whole somewhat faster, so JIT-intensive activities, for example
1868 program startup, are modestly faster, around 5%.
1870 * There have been changes to the default settings of several command
1871 line flags, as detailed below.
1873 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1874 capable hosts, the simulated CPUID will now indicate AVX2 support.
1876 * ==================== TOOL CHANGES ====================
1880 - The default value for --leak-check-heuristics has been changed from
1881 "none" to "all". This helps to reduce the number of possibly
1882 lost blocks, in particular for C++ applications.
1884 - The default value for --keep-stacktraces has been changed from
1885 "malloc-then-free" to "malloc-and-free". This has a small cost in
1886 memory (one word per malloc-ed block) but allows Memcheck to show the
1887 3 stacktraces of a dangling reference: where the block was allocated,
1888 where it was freed, and where it is acccessed after being freed.
1890 - The default value for --partial-loads-ok has been changed from "no" to
1891 "yes", so as to avoid false positive errors resulting from some kinds
1892 of vectorised loops.
1894 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1895 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1896 than get_vbits when you need to associate byte data value with
1897 their corresponding validity bits.
1899 - The 'block_list' monitor command has been enhanced:
1900 o it can print a range of loss records
1901 o it now accepts an optional argument 'limited <max_blocks>'
1902 to control the number of blocks printed.
1903 o if a block has been found using a heuristic, then
1904 'block_list' now shows the heuristic after the block size.
1905 o the loss records/blocks to print can be limited to the blocks
1906 found via specified heuristics.
1908 - The C helper functions used to instrument loads on
1909 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1910 replaced by handwritten assembly sequences. This gives speedups
1911 in the region of 0% to 7% for those targets only.
1913 - A new command line option, --expensive-definedness-checks=yes|no,
1914 has been added. This is useful for avoiding occasional invalid
1915 uninitialised-value errors in optimised code. Watch out for
1916 runtime degradation, as this can be up to 25%. As always, though,
1917 the slowdown is highly application specific. The default setting
1922 - A new monitor command 'all_snapshots <filename>' dumps all
1923 snapshots taken so far.
1927 - Significant memory reduction and moderate speedups for
1928 --history-level=full for applications accessing a lot of memory
1929 with many different stacktraces.
1931 - The default value for --conflict-cache-size=N has been doubled to
1932 2000000. Users that were not using the default value should
1933 preferably also double the value they give.
1935 The default was changed due to the changes in the "full history"
1936 implementation. Doubling the value gives on average a slightly more
1937 complete history and uses similar memory (or significantly less memory
1938 in the worst case) than the previous implementation.
1940 - The Helgrind monitor command 'info locks' now accepts an optional
1941 argument 'lock_addr', which shows information about the lock at the
1944 - When using --history-level=full, the new Helgrind monitor command
1945 'accesshistory <addr> [<len>]' will show the recorded accesses for
1946 <len> (or 1) bytes at <addr>.
1948 * ==================== OTHER CHANGES ====================
1950 * The default value for the --smc-check option has been changed from
1951 "stack" to "all-non-file" on targets that provide automatic D-I
1952 cache coherence (x86, amd64 and s390x). The result is to provide,
1953 by default, transparent support for JIT generated and self-modifying
1954 code on all targets.
1956 * Mac OS X only: the default value for the --dsymutil option has been
1957 changed from "no" to "yes", since any serious usage on Mac OS X
1958 always required it to be "yes".
1960 * The command line options --db-attach and --db-command have been removed.
1961 They were deprecated in 3.10.0.
1963 * When a process dies due to a signal, Valgrind now shows the signal
1964 and the stacktrace at default verbosity (i.e. verbosity 1).
1966 * The address description logic used by Memcheck and Helgrind now
1967 describes addresses in anonymous segments, file mmap-ed segments,
1968 shared memory segments and the brk data segment.
1970 * The new option --error-markers=<begin>,<end> can be used to mark the
1971 begin/end of errors in textual output mode, to facilitate
1972 searching/extracting errors in output files that mix valgrind errors
1973 with program output.
1975 * The new option --max-threads=<number> can be used to change the number
1976 of threads valgrind can handle. The default is 500 threads which
1977 should be more than enough for most applications.
1979 * The new option --valgrind-stacksize=<number> can be used to change the
1980 size of the private thread stacks used by Valgrind. This is useful
1981 for reducing memory use or increasing the stack size if Valgrind
1982 segfaults due to stack overflow.
1984 * The new option --avg-transtab-entry-size=<number> can be used to specify
1985 the expected instrumented block size, either to reduce memory use or
1986 to avoid excessive retranslation.
1988 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1990 * New and modified GDB server monitor features:
1992 - When a signal is reported in GDB, you can now use the GDB convenience
1993 variable $_siginfo to examine detailed signal information.
1995 - Valgrind's gdbserver now allows the user to change the signal
1996 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1997 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1998 continue without passing the signal to the process.
2000 - With GDB >= 7.10, the command 'target remote'
2001 will automatically load the executable file of the process running
2002 under Valgrind. This means you do not need to specify the executable
2003 file yourself, GDB will discover it itself. See GDB documentation about
2004 'qXfer:exec-file:read' packet for more info.
2006 * ==================== FIXED BUGS ====================
2008 The following bugs have been fixed or resolved. Note that "n-i-bz"
2009 stands for "not in bugzilla" -- that is, a bug that was reported to us
2010 but never got a bugzilla entry. We encourage you to file bugs in
2011 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2012 than mailing the developers (or mailing lists) directly -- bugs that
2013 are not entered into bugzilla tend to get forgotten about or ignored.
2015 To see details of a given bug, visit
2016 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2017 where XXXXXX is the bug number as listed below.
2019 116002 VG_(printf): Problems with justification of strings and integers
2020 155125 avoid cutting away file:lineno after long function name
2021 197259 Unsupported arch_prtctl PR_SET_GS option
2022 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
2023 201216 Fix Valgrind does not support pthread_sigmask() on OS X
2024 201435 Fix Darwin: -v does not show kernel version
2025 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
2026 211256 Fixed an outdated comment regarding the default platform.
2027 211529 Incomplete call stacks for code compiled by newer versions of MSVC
2028 211926 Avoid compilation warnings in valgrind.h with -pedantic
2029 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
2031 226609 Crediting upstream authors in man page
2032 231257 Valgrind omits path when executing script from shebang line
2033 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
2034 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
2035 269360 s390x: Fix addressing mode selection for compare-and-swap
2036 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
2038 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
2039 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
2040 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
2041 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
2042 330147 libmpiwrap PMPI_Get_count returns undefined value
2043 333051 mmap of huge pages fails due to incorrect alignment
2045 334802 valgrind does not always explain why a given option is bad
2046 335618 mov.w rN, pc/sp (ARM32)
2047 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
2052 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
2053 338602 AVX2 bit in CPUID missing
2054 338606 Strange message for scripts with invalid interpreter
2055 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
2056 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
2057 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
2059 339156 gdbsrv not called for fatal signal
2060 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
2061 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
2062 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
2063 339442 Fix testsuite build failure on OS X 10.9
2064 339542 Enable compilation with Intel's ICC compiler
2065 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
2066 339688 Mac-specific ASM does not support .version directive (cpuid,
2067 tronical and pushfpopf tests)
2068 339745 Valgrind crash when check Marmalade app (partial fix)
2069 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
2070 339778 Linux/TileGx platform support for Valgrind
2071 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
2072 339789 Fix none/tests/execve test on Mac OS X 10.9
2073 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
2074 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
2075 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
2076 340392 Allow user to select more accurate definedness checking in memcheck
2077 to avoid invalid complaints on optimised code
2078 340430 Fix some grammatical weirdness in the manual.
2079 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
2080 341419 Signal handler ucontext_t not filled out correctly on OS X
2081 341539 VG_(describe_addr) should not describe address as belonging to client
2082 segment if it is past the heap end
2083 341613 Enable building of manythreads and thread-exits tests on Mac OS X
2084 341615 Fix none/tests/darwin/access_extended test on Mac OS X
2085 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
2086 341789 aarch64: shmat fails with valgrind on ARMv8
2087 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
2088 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
2089 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
2090 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
2091 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
2092 342221 socket connect false positive uninit memory for unknown af family
2093 342353 Allow dumping full massif output while valgrind is still running
2094 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
2098 342603 Add I2C_SMBUS ioctl support
2099 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
2100 342683 Mark memory past the initial brk limit as unaddressable
2101 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
2102 342795 Internal glibc __GI_mempcpy call should be intercepted
2103 342841 s390x: Support instructions fiebr(a) and fidbr(a)
2104 343012 Unhandled syscall 319 (memfd_create)
2105 343069 Patch updating v4l2 API support
2106 343173 helgrind crash during stack unwind
2107 343219 fix GET_STARTREGS for arm
2108 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
2109 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
2110 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
2111 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
2112 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
2113 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
2114 343597 ppc64le: incorrect use of offseof macro
2115 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
2116 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
2117 343732 Unhandled syscall 144 (setgid) on aarch64
2118 343733 Unhandled syscall 187 (msgctl and related) on aarch64
2119 343802 s390x: False positive "conditional jump or move depends on [..]
2120 343902 --vgdb=yes doesn't break when --xml=yes is used
2121 343967 Don't warn about setuid/setgid/setcap executable for directories
2122 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
2123 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
2124 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
2125 344054 www - update info for Solaris/illumos
2126 344416 'make regtest' does not work cleanly on OS X
2127 344235 Remove duplicate include of pub_core_aspacemgr.h
2128 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
2129 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
2130 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
2131 344314 callgrind_annotate ... warnings about commands containing newlines
2132 344318 socketcall should wrap recvmmsg and sendmmsg
2133 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
2134 344416 Fix 'make regtest' does not work cleanly on OS X
2135 344499 Fix compilation for Linux kernel >= 4.0.0
2136 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
2137 unix:349 (__pthread_fchdir)
2138 344559 Garbage collection of unused segment names in address space manager
2139 344560 Fix stack traces missing penultimate frame on OS X
2140 344621 Fix memcheck/tests/err_disable4 test on OS X
2141 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
2142 344702 Fix missing libobjc suppressions on OS X 10.10
2144 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
2145 344939 Fix memcheck/tests/xml1 on OS X 10.10
2146 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
2147 345079 Fix build problems in VEX/useful/test_main.c
2148 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
2149 345177 arm64: prfm (reg) not implemented
2150 345215 Performance improvements for the register allocator
2151 345248 add support for Solaris OS in valgrind
2152 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
2153 345394 Fix memcheck/tests/strchr on OS X
2154 345637 Fix memcheck/tests/sendmsg on OS X
2155 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
2156 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
2157 345887 Fix an assertion in the address space manager
2158 345928 amd64: callstack only contains current function for small stacks
2159 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
2160 345987 MIPS64: Implement cavium LHX instruction
2161 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
2162 346185 Fix typo saving altivec register v24
2163 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
2164 and LibVEX_GuestPPC64_get_CR()
2165 346270 Regression tests none/tests/jm_vec/isa_2_07 and
2166 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
2167 346307 fuse filesystem syscall deadlocks
2168 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
2169 346411 MIPS: SysRes::_valEx handling is incorrect
2170 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
2171 346474 PPC64 Power 8, spr TEXASRU register not supported
2172 346487 Compiler generates "note" about a future ABI change for PPC64
2173 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
2174 and causing spurious "invalid read of size 8" warnings
2175 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
2176 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
2177 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
2178 347322 Power PC regression test cleanup
2179 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
2181 347389 unhandled syscall: 373 (Linux ARM syncfs)
2182 347686 Patch set to cleanup PPC64 regtests
2183 347978 Remove bash dependencies where not needed
2184 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
2185 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
2187 348102 Patch updating v4l2 API support
2188 348247 amd64 front end: jno jumps wrongly when overflow is not set
2189 348269 Improve mmap MAP_HUGETLB support.
2190 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
2191 348345 Assertion fails for negative lineno
2192 348377 Unsupported ARM instruction: yield
2193 348565 Fix detection of command line option availability for clang
2194 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
2195 348728 Fix broken check for VIDIOC_G_ENC_INDEX
2196 348748 Fix redundant condition
2197 348890 Fix clang warning about unsupported --param inline-unit-growth=900
2198 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
2199 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
2200 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
2201 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
2202 349626 Implemented additional Xen hypercalls
2203 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
2204 349790 Clean up of the hardware capability checking utilities.
2205 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
2206 349874 Fix typos in source code
2207 349879 memcheck: add handwritten assembly for helperc_LOADV*
2208 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
2209 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
2210 350202 Add limited param to 'monitor block_list'
2211 350290 s390x: Support instructions fixbr(a)
2212 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
2213 350809 Fix none/tests/async-sigs for Solaris
2214 350811 Remove reference to --db-attach which has been removed.
2215 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
2216 350854 hard-to-understand code in VG_(load_ELF)()
2217 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2218 351386 Solaris: Cannot run ld.so.1 under Valgrind
2219 351474 Fix VG_(iseqsigset) as obvious
2220 351531 Typo in /include/vki/vki-xen-physdev.h header guard
2221 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
2222 351858 ldsoexec support on Solaris
2223 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
2224 352130 helgrind reports false races for printfs using mempcpy on FILE* state
2225 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
2226 352320 arm64 crash on none/tests/nestedfs
2227 352765 Vbit test fails on Power 6
2228 352768 The mbar instruction is missing from the Power PC support
2229 352769 Power PC program priority register (PPR) is not supported
2230 n-i-bz Provide implementations of certain compiler builtins to support
2231 compilers that may not provide those
2232 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
2233 n-i-bz Fix compilation on distros with glibc < 2.5
2234 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
2235 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
2236 n-i-bz mremap did not work properly on shared memory
2237 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
2238 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
2240 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
2241 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
2242 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
2246 Release 3.10.1 (25 November 2014)
2247 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2248 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
2249 and backports fixes for all reported missing AArch64 ARMv8 instructions
2250 and syscalls from the trunk. If you package or deliver 3.10.0 for others
2251 to use, you might want to consider upgrading to 3.10.1 instead.
2253 The following bugs have been fixed or resolved. Note that "n-i-bz"
2254 stands for "not in bugzilla" -- that is, a bug that was reported to us
2255 but never got a bugzilla entry. We encourage you to file bugs in
2256 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2257 than mailing the developers (or mailing lists) directly -- bugs that
2258 are not entered into bugzilla tend to get forgotten about or ignored.
2260 To see details of a given bug, visit
2261 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2262 where XXXXXX is the bug number as listed below.
2264 335440 arm64: ld1 (single structure) is not implemented
2265 335713 arm64: unhanded instruction: prfm (immediate)
2266 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
2267 339182 ppc64: AvSplat ought to load destination vector register with [..]
2268 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
2269 339433 ppc64 lxvw4x instruction uses four 32-byte loads
2270 339645 Use correct tag names in sys_getdents/64 wrappers
2271 339706 Fix false positive for ioctl(TIOCSIG) on linux
2272 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
2273 339853 arm64 times syscall unknown
2274 339855 arm64 unhandled getsid/setsid syscalls
2275 339858 arm64 dmb sy not implemented
2276 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
2277 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
2278 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
2280 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
2281 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
2282 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
2283 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
2284 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
2285 340509 arm64: unhandled instruction fcvtas
2286 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
2287 340632 arm64: unhandled instruction fcvtas
2288 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
2289 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
2290 340788 warning: unhandled syscall: 318 (getrandom)
2291 340807 disInstr(arm): unhandled instruction: 0xEE989B20
2292 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
2293 340922 arm64: unhandled getgroups/setgroups syscalls
2294 350251 Fix typo in VEX utility program (test_main.c).
2295 350407 arm64: unhandled instruction ucvtf (vector, integer)
2296 350809 none/tests/async-sigs breaks when run under cron on Solaris
2297 350811 update README.solaris after r15445
2298 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
2299 350854 strange code in VG_(load_ELF)()
2300 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2301 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
2302 n-i-bz Add missing ]] to terminate CDATA.
2303 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
2304 n-i-bz Enable sys_fadvise64_64 on arm32.
2305 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
2306 n-i-bz Add test cases for all known arm64 load/store instructions.
2307 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
2308 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
2309 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
2310 n-i-bz arm64: Implement "fcvtpu w, s".
2311 n-i-bz arm64: implement ADDP and various others
2312 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
2313 n-i-bz arm64: enable FCVT{A,N}S X,S.
2315 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
2319 Release 3.10.0 (10 September 2014)
2320 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2322 3.10.0 is a feature release with many improvements and the usual
2323 collection of bug fixes.
2325 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
2326 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
2327 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
2328 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
2329 significantly improved relative to the 3.9.0 release.
2331 * ================== PLATFORM CHANGES =================
2333 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
2334 is mostly complete, and is usable, but some SIMD instructions are as
2337 * Support for little-endian variant of the 64-bit POWER architecture.
2339 * Support for Android on MIPS32.
2341 * Support for 64bit FPU on MIPS32 platforms.
2343 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
2345 * Configuration for and running on Android targets has changed.
2346 See README.android in the source tree for details.
2348 * ================== DEPRECATED FEATURES =================
2350 * --db-attach is now deprecated and will be removed in the next
2351 valgrind feature release. The built-in GDB server capabilities are
2352 superior and should be used instead. Learn more here:
2353 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
2355 * ==================== TOOL CHANGES ====================
2359 - Client code can now selectively disable and re-enable reporting of
2360 invalid address errors in specific ranges using the new client
2361 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
2362 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
2364 - Leak checker: there is a new leak check heuristic called
2365 "length64". This is used to detect interior pointers pointing 8
2366 bytes inside a block, on the assumption that the first 8 bytes
2367 holds the value "block size - 8". This is used by
2368 sqlite3MemMalloc, for example.
2370 - Checking of system call parameters: if a syscall parameter
2371 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
2372 several fields not initialised, an error is now reported for each
2373 field. Previously, an error was reported only for the first
2374 uninitialised field.
2376 - Mismatched alloc/free checking: a new flag
2377 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
2378 such checks if necessary.
2382 - Improvements to error messages:
2384 o Race condition error message involving heap allocated blocks also
2385 show the thread number that allocated the raced-on block.
2387 o All locks referenced by an error message are now announced.
2388 Previously, some error messages only showed the lock addresses.
2390 o The message indicating where a lock was first observed now also
2391 describes the address/location of the lock.
2393 - Helgrind now understands the Ada task termination rules and
2394 creates a happens-before relationship between a terminated task
2395 and its master. This avoids some false positives and avoids a big
2396 memory leak when a lot of Ada tasks are created and terminated.
2397 The interceptions are only activated with forthcoming releases of
2398 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
2400 - A new GDB server monitor command "info locks" giving the list of
2401 locks, their location, and their status.
2405 - callgrind_control now supports the --vgdb-prefix argument,
2406 which is needed if valgrind was started with this same argument.
2408 * ==================== OTHER CHANGES ====================
2410 * Unwinding through inlined function calls. Stack unwinding can now
2411 make use of Dwarf3 inlined-unwind information if it is available.
2412 The practical effect is that inlined calls become visible in stack
2413 traces. The suppression matching machinery has been adjusted
2414 accordingly. This is controlled by the new option
2415 --read-inline-info=yes|no. Currently this is enabled by default
2416 only on Linux and Android targets and only for the tools Memcheck,
2419 * Valgrind can now read EXIDX unwind information on 32-bit ARM
2420 targets. If an object contains both CFI and EXIDX unwind
2421 information, Valgrind will prefer the CFI over the EXIDX. This
2422 facilitates unwinding through system libraries on arm-android
2425 * Address description logic has been improved and is now common
2426 between Memcheck and Helgrind, resulting in better address
2427 descriptions for some kinds of error messages.
2429 * Error messages about dubious arguments (eg, to malloc or calloc) are
2430 output like other errors. This means that they can be suppressed
2431 and they have a stack trace.
2433 * The C++ demangler has been updated for better C++11 support.
2435 * New and modified GDB server monitor features:
2437 - Thread local variables/storage (__thread) can now be displayed.
2439 - The GDB server monitor command "v.info location <address>"
2440 displays information about an address. The information produced
2441 depends on the tool and on the options given to valgrind.
2442 Possibly, the following are described: global variables, local
2443 (stack) variables, allocated or freed blocks, ...
2445 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
2446 ask the GDB server to stop at the start of program execution, at
2447 the end of the program execution and on Valgrind internal errors.
2449 - A new monitor command "v.info stats" shows various Valgrind core
2450 and tool statistics.
2452 - A new monitor command "v.set hostvisibility" allows the GDB server
2453 to provide access to Valgrind internal host status/memory.
2455 * A new option "--aspace-minaddr=<address>" can in some situations
2456 allow the use of more memory by decreasing the address above which
2457 Valgrind maps memory. It can also be used to solve address
2458 conflicts with system libraries by increasing the default value.
2459 See user manual for details.
2461 * The amount of memory used by Valgrind to store debug info (unwind
2462 info, line number information and symbol data) has been
2463 significantly reduced, even though Valgrind now reads more
2464 information in order to support unwinding of inlined function calls.
2466 * Dwarf3 handling with --read-var-info=yes has been improved:
2468 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
2470 - Code compiled with
2471 -ffunction-sections -fdata-sections -Wl,--gc-sections
2472 no longer causes assertion failures.
2474 * Improved checking for the --sim-hints= and --kernel-variant=
2475 options. Unknown strings are now detected and reported to the user
2478 * The semantics of stack start/end boundaries in the valgrind.h
2479 VALGRIND_STACK_REGISTER client request has been clarified and
2480 documented. The convention is that start and end are respectively
2481 the lowest and highest addressable bytes of the stack.
2483 * ==================== FIXED BUGS ====================
2485 The following bugs have been fixed or resolved. Note that "n-i-bz"
2486 stands for "not in bugzilla" -- that is, a bug that was reported to us
2487 but never got a bugzilla entry. We encourage you to file bugs in
2488 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2489 than mailing the developers (or mailing lists) directly -- bugs that
2490 are not entered into bugzilla tend to get forgotten about or ignored.
2492 To see details of a given bug, visit
2493 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2494 where XXXXXX is the bug number as listed below.
2496 175819 Support for ipv6 socket reporting with --track-fds
2497 232510 make distcheck fails
2498 249435 Analyzing wine programs with callgrind triggers a crash
2499 278972 support for inlined function calls in stacktraces and suppression
2501 291310 FXSAVE instruction marks memory as undefined on amd64
2502 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2503 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2504 315199 vgcore file for threaded app does not show which thread crashed
2505 315952 tun/tap ioctls are not supported
2506 323178 Unhandled instruction: PLDW register (ARM)
2507 323179 Unhandled instruction: PLDW immediate (ARM)
2508 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2509 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2510 325124 [MIPSEL] Compilation error
2511 325477 Phase 4 support for IBM Power ISA 2.07
2512 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2513 325628 Phase 5 support for IBM Power ISA 2.07
2514 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2515 325751 Missing the two privileged Power PC Transactional Memory Instructions
2516 325816 Phase 6 support for IBM Power ISA 2.07
2517 325856 Make SGCheck fail gracefully on unsupported platforms
2518 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2519 326436 DRD: False positive in libstdc++ std::list::push_back
2520 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2521 326462 Refactor vgdb to isolate invoker stuff into separate module
2522 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2523 326623 DRD: false positive conflict report in a field assignment
2524 326724 Valgrind does not compile on OSX 1.9 Mavericks
2525 326816 Intercept for __strncpy_sse2_unaligned missing?
2526 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2527 326983 Clear direction flag after tests on amd64.
2528 327212 Do not prepend the current directory to absolute path names.
2529 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2530 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2531 327284 s390x: Fix translation of the risbg instruction
2532 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2533 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2534 327916 DW_TAG_typedef may have no name
2535 327943 s390x: add a redirection for the 'index' function
2536 328100 XABORT not implemented
2537 328205 Implement additional Xen hypercalls
2538 328454 add support Backtraces with ARM unwind tables (EXIDX)
2539 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2540 328711 valgrind.1 manpage "memcheck options" section is badly generated
2541 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2542 329612 Incorrect handling of AT_BASE for image execution
2543 329694 clang warns about using uninitialized variable
2544 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2545 330228 mmap must align to VKI_SHMLBA on mips32
2546 330257 LLVM does not support `-mno-dynamic-no-pic` option
2547 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2548 330459 --track-fds=yes doesn't track eventfds
2549 330469 Add clock_adjtime syscall support
2550 330594 Missing sysalls on PowerPC / uClibc
2551 330622 Add test to regression suite for POWER instruction: dcbzl
2552 330939 Support for AMD's syscall instruction on x86
2554 330941 Typo in PRE(poll) syscall wrapper
2555 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2556 331254 Fix expected output for memcheck/tests/dw4
2557 331255 Fix race condition in test none/tests/coolo_sigaction
2558 331257 Fix type of jump buffer in test none/tests/faultstatus
2559 331305 configure uses bash specific syntax
2560 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2561 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2562 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2563 331829 Unexpected ioctl opcode sign extension
2564 331830 ppc64: WARNING: unhandled syscall: 96/97
2565 331839 drd/tests/sem_open specifies invalid semaphore name
2566 331847 outcome of drd/tests/thread_name is nondeterministic
2567 332037 Valgrind cannot handle Thumb "add pc, reg"
2568 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2569 consistency checks enabled
2570 332263 intercepts for pthread_rwlock_timedrdlock and
2571 pthread_rwlock_timedwrlock are incorrect
2572 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2574 332276 Implement additional Xen hypercalls
2575 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2576 332765 Fix ms_print to create temporary files in a proper directory
2577 333072 drd: Add semaphore annotations
2578 333145 Tests for missaligned PC+#imm access for arm
2579 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2580 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2581 333248 WARNING: unhandled syscall: unix:443
2582 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2583 333501 cachegrind: assertion: Cache set count is not a power of two.
2586 333666 Recognize MPX instructions and bnd prefix.
2587 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2588 333817 Valgrind reports the memory areas written to by the SG_IO
2590 334049 lzcnt fails silently (x86_32)
2591 334384 Valgrind does not have support Little Endian support for
2593 334585 recvmmsg unhandled (+patch) (arm)
2594 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2595 334727 Build fails with -Werror=format-security
2596 334788 clarify doc about --log-file initial program directory
2597 334834 PPC64 Little Endian support, patch 2
2598 334836 PPC64 Little Endian support, patch 3 testcase fixes
2599 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2600 335034 Unhandled ioctl: HCIGETDEVLIST
2601 335155 vgdb, fix error print statement.
2602 335262 arm64: movi 8bit version is not supported
2603 335263 arm64: dmb instruction is not implemented
2604 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2605 335496 arm64: sbc/abc instructions are not implemented
2606 335554 arm64: unhandled instruction: abs
2607 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2608 335735 arm64: unhandled instruction: cnt
2609 335736 arm64: unhandled instruction: uaddlv
2610 335848 arm64: unhandled instruction: {s,u}cvtf
2611 335902 arm64: unhandled instruction: sli
2612 335903 arm64: unhandled instruction: umull (vector)
2613 336055 arm64: unhandled instruction: mov (element)
2614 336062 arm64: unhandled instruction: shrn{,2}
2615 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2616 336189 arm64: unhandled Instruction: mvn
2617 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2618 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2619 336772 Make moans about unknown ioctls more informative
2620 336957 Add a section about the Solaris/illumos port on the webpage
2621 337094 ifunc wrapper is broken on ppc64
2622 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2623 337528 leak check heuristic for block prefixed by length as 64bit number
2624 337740 Implement additional Xen hypercalls
2625 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2626 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2627 337871 deprecate --db-attach
2628 338023 Add support for all V4L2/media ioctls
2629 338024 inlined functions are not shown if DW_AT_ranges is used
2630 338106 Add support for 'kcmp' syscall
2631 338115 DRD: computed conflict set differs from actual after fork
2632 338160 implement display of thread local storage in gdbsrv
2633 338205 configure.ac and check for -Wno-tautological-compare
2634 338300 coredumps are missing one byte of every segment
2635 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2636 338499 --sim-hints parsing broken due to wrong order in tokens
2637 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2638 338681 Unable to unwind through clone thread created on i386-linux
2639 338698 race condition between gdbsrv and vgdb on startup
2640 338703 helgrind on arm-linux gets false positives in dynamic loader
2641 338791 alt dwz files can be relative of debug/main file
2642 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2643 338932 build V-trunk with gcc-trunk
2644 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2645 345079 Fix build problems in VEX/useful/test_main.c
2646 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2647 n-i-bz s390x: Fix memory corruption for multithreaded applications
2648 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2649 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2650 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2651 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2652 n-i-bz Add command-line option --default-suppressions
2653 n-i-bz Add support for BLKDISCARDZEROES ioctl
2654 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2655 n-i-bz Add support for sys_pivot_root and sys_unshare
2657 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2658 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2659 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2663 Release 3.9.0 (31 October 2013)
2664 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2665 3.9.0 is a feature release with many improvements and the usual
2666 collection of bug fixes.
2668 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2669 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2670 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2671 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2673 * ================== PLATFORM CHANGES =================
2675 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2676 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2678 * Support for MIPS DSP ASE on MIPS32 platforms.
2680 * Support for s390x Decimal Floating Point instructions on hosts that
2681 have the DFP facility installed.
2683 * Support for POWER8 (Power ISA 2.07) instructions
2685 * Support for Intel AVX2 instructions. This is available only on 64
2688 * Initial support for Intel Transactional Synchronization Extensions,
2691 * Initial support for Hardware Transactional Memory on POWER.
2693 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2694 run large GUI apps tolerably well.
2696 * ==================== TOOL CHANGES ====================
2700 - Improvements in handling of vectorised code, leading to
2701 significantly fewer false error reports. You need to use the flag
2702 --partial-loads-ok=yes to get the benefits of these changes.
2704 - Better control over the leak checker. It is now possible to
2705 specify which leak kinds (definite/indirect/possible/reachable)
2706 should be displayed, which should be regarded as errors, and which
2707 should be suppressed by a given leak suppression. This is done
2708 using the options --show-leak-kinds=kind1,kind2,..,
2709 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2710 "match-leak-kinds:" line in suppression entries, respectively.
2712 Note that generated leak suppressions contain this new line and
2713 are therefore more specific than in previous releases. To get the
2714 same behaviour as previous releases, remove the "match-leak-kinds:"
2715 line from generated suppressions before using them.
2717 - Reduced "possible leak" reports from the leak checker by the use
2718 of better heuristics. The available heuristics provide detection
2719 of valid interior pointers to std::stdstring, to new[] allocated
2720 arrays with elements having destructors and to interior pointers
2721 pointing to an inner part of a C++ object using multiple
2722 inheritance. They can be selected individually using the
2723 option --leak-check-heuristics=heur1,heur2,...
2725 - Better control of stacktrace acquisition for heap-allocated
2726 blocks. Using the --keep-stacktraces option, it is possible to
2727 control independently whether a stack trace is acquired for each
2728 allocation and deallocation. This can be used to create better
2729 "use after free" errors or to decrease Valgrind's resource
2730 consumption by recording less information.
2732 - Better reporting of leak suppression usage. The list of used
2733 suppressions (shown when the -v option is given) now shows, for
2734 each leak suppressions, how many blocks and bytes it suppressed
2735 during the last leak search.
2739 - False errors resulting from the use of statically initialised
2740 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2743 - False errors resulting from the use of pthread_cond_waits that
2744 timeout, have been removed.
2746 * ==================== OTHER CHANGES ====================
2748 * Some attempt to tune Valgrind's space requirements to the expected
2749 capabilities of the target:
2751 - The default size of the translation cache has been reduced from 8
2752 sectors to 6 on Android platforms, since each sector occupies
2753 about 40MB when using Memcheck.
2755 - The default size of the translation cache has been increased to 16
2756 sectors on all other platforms, reflecting the fact that large
2757 applications require instrumentation and storage of huge amounts
2758 of code. For similar reasons, the number of memory mapped
2759 segments that can be tracked has been increased by a factor of 6.
2761 - In all cases, the maximum number of sectors in the translation
2762 cache can be controlled by the new flag --num-transtab-sectors.
2764 * Changes in how debug info (line numbers, etc) is read:
2766 - Valgrind no longer temporarily mmaps the entire object to read
2767 from it. Instead, reading is done through a small fixed sized
2768 buffer. This avoids virtual memory usage spikes when Valgrind
2769 reads debuginfo from large shared objects.
2771 - A new experimental remote debug info server. Valgrind can read
2772 debug info from a different machine (typically, a build host)
2773 where debuginfo objects are stored. This can save a lot of time
2774 and hassle when running Valgrind on resource-constrained targets
2775 (phones, tablets) when the full debuginfo objects are stored
2776 somewhere else. This is enabled by the --debuginfo-server=
2779 - Consistency checking between main and debug objects can be
2780 disabled using the --allow-mismatched-debuginfo option.
2782 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2783 scanning can recover stack traces in some cases when the normal
2784 unwind mechanisms fail. Stack scanning is best described as "a
2785 nasty, dangerous and misleading hack" and so is disabled by default.
2786 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2789 * Detection and merging of recursive stack frame cycles. When your
2790 program has recursive algorithms, this limits the memory used by
2791 Valgrind for recorded stack traces and avoids recording
2792 uninteresting repeated calls. This is controlled by the command
2793 line option --merge-recursive-frame and by the monitor command
2794 "v.set merge-recursive-frames".
2796 * File name and line numbers for used suppressions. The list of used
2797 suppressions (shown when the -v option is given) now shows, for each
2798 used suppression, the file name and line number where the suppression
2801 * New and modified GDB server monitor features:
2803 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2804 that can be used to execute gdbserver monitor commands from the
2807 - A new monitor command, "v.info open_fds", that gives the list of
2808 open file descriptors and additional details.
2810 - An optional message in the "v.info n_errs_found" monitor command,
2811 for example "v.info n_errs_found test 1234 finished", allowing a
2812 comment string to be added to the process output, perhaps for the
2813 purpose of separating errors of different tests or test phases.
2815 - A new monitor command "v.info execontext" that shows information
2816 about the stack traces recorded by Valgrind.
2818 - A new monitor command "v.do expensive_sanity_check_general" to run
2819 some internal consistency checks.
2821 * New flag --sigill-diagnostics to control whether a diagnostic
2822 message is printed when the JIT encounters an instruction it can't
2823 translate. The actual behavior -- delivery of SIGILL to the
2824 application -- is unchanged.
2826 * The maximum amount of memory that Valgrind can use on 64 bit targets
2827 has been increased from 32GB to 64GB. This should make it possible
2828 to run applications on Memcheck that natively require up to about 35GB.
2830 * ==================== FIXED BUGS ====================
2832 The following bugs have been fixed or resolved. Note that "n-i-bz"
2833 stands for "not in bugzilla" -- that is, a bug that was reported to us
2834 but never got a bugzilla entry. We encourage you to file bugs in
2835 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2836 than mailing the developers (or mailing lists) directly -- bugs that
2837 are not entered into bugzilla tend to get forgotten about or ignored.
2839 To see details of a given bug, visit
2840 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2841 where XXXXXX is the bug number as listed below.
2843 123837 system call: 4th argument is optional, depending on cmd
2844 135425 memcheck should tell you where Freed blocks were Mallocd
2845 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2846 207815 Adds some of the drm ioctls to syswrap-linux.c
2847 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2848 252955 Impossible to compile with ccache
2849 253519 Memcheck reports auxv pointer accesses as invalid reads.
2850 263034 Crash when loading some PPC64 binaries
2851 269599 Increase deepest backtrace
2852 274695 s390x: Support "compare to/from logical" instructions (z196)
2853 275800 s390x: Autodetect cache info (part 2)
2854 280271 Valgrind reports possible memory leaks on still-reachable std::string
2855 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2856 289578 Backtraces with ARM unwind tables (stack scan flags)
2857 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2858 304832 ppc32: build failure
2859 305431 Use find_buildid shdr fallback for separate .debug files
2860 305728 Add support for AVX2 instructions
2861 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2862 306035 s390x: Fix IR generation for LAAG and friends
2863 306054 s390x: Condition code computation for convert-to-int/logical
2864 306098 s390x: alternate opcode form for convert to/from fixed
2865 306587 Fix cache line detection from auxiliary vector for PPC.
2866 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2867 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2868 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2869 307101 sys_capget second argument can be NULL
2870 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2871 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2872 307113 s390x: DFP support
2873 307141 valgrind does't work in mips-linux system
2874 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2875 307285 x86_amd64 feature test for avx in test suite is wrong
2876 307290 memcheck overlap testcase needs memcpy version filter
2877 307463 Please add "&limit=0" to the "all open bugs" link
2878 307465 --show-possibly-lost=no should reduce the error count / exit code
2879 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2880 307729 pkgconfig support broken valgrind.pc
2881 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2882 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2883 308089 Unhandled syscall on ppc64: prctl
2884 308135 PPC32 MPC8xx has 16 bytes cache size
2885 308321 testsuite memcheck filter interferes with gdb_filter
2887 308341 vgdb should report process exit (or fatal signal)
2888 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2889 308495 Remove build dependency on installed Xen headers
2890 308573 Internal error on 64-bit instruction executed in 32-bit mode
2892 308627 pmovmskb validity bit propagation is imprecise
2893 308644 vgdb command for having the info for the track-fds option
2894 308711 give more info about aspacemgr and arenas in out_of_memory
2895 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2896 308718 ARM implement SMLALBB family of instructions
2897 308886 Missing support for PTRACE_SET/GETREGSET
2898 308930 syscall name_to_handle_at (303 on amd64) not handled
2899 309229 V-bit tester does not report number of tests generated
2900 309323 print unrecognized instuction on MIPS
2901 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2902 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2903 309430 Self hosting ppc64 encounters a vassert error on operand type
2904 309600 valgrind is a bit confused about 0-sized sections
2905 309823 Generate errors for still reachable blocks
2906 309921 PCMPISTRI validity bit propagation is imprecise
2907 309922 none/tests/ppc64/test_dfp5 sometimes fails
2908 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2909 310424 --read-var-info does not properly describe static variables
2910 310792 search additional path for debug symbols
2911 310931 s390x: Message-security assist (MSA) instruction extension [..]
2912 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2913 311318 ARM: "128-bit constant is not implemented" error message
2914 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2915 311690 V crashes because it redirects branches inside of a redirected function
2916 311880 x86_64: make regtest hangs at shell_valid1
2917 311922 WARNING: unhandled syscall: 170
2919 312171 ppc: insn selection for DFP
2920 312571 Rounding mode call wrong for the DFP Iops [..]
2921 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2922 312913 Dangling pointers error should also report the alloc stack trace
2923 312980 Building on Mountain Lion generates some compiler warnings
2924 313267 Adding MIPS64/Linux port to Valgrind
2927 313811 Buffer overflow in assert_fail
2928 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2929 314269 ppc: dead code in insn selection
2930 314718 ARM: implement integer divide instruction (sdiv and udiv)
2931 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2932 315441 sendmsg syscall should ignore unset msghdr msg_flags
2933 315534 msgrcv inside a thread causes valgrind to hang (block)
2934 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2935 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2936 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2937 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2938 316144 valgrind.1 manpage contains unknown ??? strings [..]
2939 316145 callgrind command line options in manpage reference (unknown) [..]
2940 316145 callgrind command line options in manpage reference [..]
2941 316181 drd: Fixed a 4x slowdown for certain applications
2942 316503 Valgrind does not support SSE4 "movntdqa" instruction
2943 316535 Use of |signed int| instead of |size_t| in valgrind messages
2944 316696 fluidanimate program of parsec 2.1 stuck
2945 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2946 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2947 317186 "Impossible happens" when occurs VCVT instruction on ARM
2948 317318 Support for Threading Building Blocks "scalable_malloc"
2949 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2950 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2951 317463 bmi testcase IR SANITY CHECK FAILURE
2952 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2953 318050 libmpiwrap fails to compile with out-of-source build
2954 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2955 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2956 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2957 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2958 318932 Add missing PPC64 and PPC32 system call support
2959 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2960 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2961 319494 VEX Makefile-gcc standalone build update after r2702
2962 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2963 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2964 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2965 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2966 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2967 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2968 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2969 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2970 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2971 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2972 320895 add fanotify support (patch included)
2973 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2974 321065 Valgrind updates for Xen 4.3
2975 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2976 321363 Unhandled instruction: SSAX (ARM + Thumb)
2977 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2978 321466 Unhandled instruction: SHASX (ARM + Thumb)
2979 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2980 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2981 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2982 321620 Unhandled instruction: UASX (ARM + Thumb)
2983 321621 Unhandled instruction: USAX (ARM + Thumb)
2984 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2985 321693 Unhandled instruction: LDRSBT (Thumb)
2986 321694 Unhandled instruction: UQASX (ARM + Thumb)
2987 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2988 321697 Unhandled instruction: UHASX (ARM + Thumb)
2989 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2990 321704 Unhandled instruction: REVSH (ARM + Thumb)
2991 321730 Add cg_diff and cg_merge man pages
2992 321738 Add vgdb and valgrind-listener man pages
2994 321891 Unhandled instruction: LDRHT (Thumb)
2995 321960 pthread_create() then alloca() causing invalid stack write errors
2996 321969 ppc32 and ppc64 don't support [lf]setxattr
2997 322254 Show threadname together with tid if set by application
2998 322294 Add initial support for IBM Power ISA 2.07
2999 322368 Assertion failure in wqthread_hijack under OS X 10.8
3000 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
3001 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
3002 322851 0bXXX binary literal syntax is not standard
3003 323035 Unhandled instruction: LDRSHT(Thumb)
3004 323036 Unhandled instruction: SMMLS (ARM and Thumb)
3005 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
3006 323175 Unhandled instruction: SMLALD (ARM + Thumb)
3007 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
3008 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
3009 323437 Phase 2 support for IBM Power ISA 2.07
3010 323713 Support mmxext (integer sse) subset on i386 (athlon)
3011 323803 Transactional memory instructions are not supported for Power
3012 323893 SSE3 not available on amd cpus in valgrind
3013 323905 Probable false positive from Valgrind/drd on close()
3014 323912 valgrind.h header isn't compatible for mingw64
3015 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
3016 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
3017 324181 mmap does not handle MAP_32BIT
3018 324227 memcheck false positive leak when a thread calls exit+block [..]
3019 324421 Support for fanotify API on ARM architecture
3020 324514 gdbserver monitor cmd output behaviour consistency [..]
3021 324518 ppc64: Emulation of dcbt instructions does not handle [..]
3022 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
3023 324582 When access is made to freed memory, report both allocation [..]
3024 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
3025 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
3026 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
3027 324834 Unhandled instructions in Microsoft C run-time for x86_64
3028 324894 Phase 3 support for IBM Power ISA 2.07
3029 326091 drd: Avoid false race reports from optimized strlen() impls
3030 326113 valgrind libvex hwcaps error on AMD64
3031 n-i-bz Some wrong command line options could be ignored
3032 n-i-bz patch to allow fair-sched on android
3033 n-i-bz report error for vgdb snapshot requested before execution
3034 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
3036 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
3040 Release 3.8.1 (19 September 2012)
3041 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3042 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
3043 that occur moderately frequently in real use cases, adds support for
3044 some missing instructions on ARM, and fixes a deadlock condition on
3045 MacOSX. If you package or deliver 3.8.0 for others to use, you might
3046 want to consider upgrading to 3.8.1 instead.
3048 The following bugs have been fixed or resolved. Note that "n-i-bz"
3049 stands for "not in bugzilla" -- that is, a bug that was reported to us
3050 but never got a bugzilla entry. We encourage you to file bugs in
3051 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3052 than mailing the developers (or mailing lists) directly -- bugs that
3053 are not entered into bugzilla tend to get forgotten about or ignored.
3055 To see details of a given bug, visit
3056 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3057 where XXXXXX is the bug number as listed below.
3060 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
3061 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
3062 298281 wcslen causes false(?) uninitialised value warnings
3063 301281 valgrind hangs on OS X when the process calls system()
3064 304035 disInstr(arm): unhandled instruction 0xE1023053
3065 304867 implement MOVBE instruction in x86 mode
3066 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
3067 305042 amd64: implement 0F 7F encoding of movq between two registers
3068 305199 ARM: implement QDADD and QDSUB
3069 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
3070 305513 killed by fatal signal: SIGSEGV
3071 305690 DRD reporting invalid semaphore when sem_trywait fails
3072 305926 Invalid alignment checks for some AVX instructions
3073 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
3074 306310 3.8.0 release tarball missing some files
3075 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
3076 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
3077 n-i-bz shmat of a segment > 4Gb does not work
3078 n-i-bz simulate_control_c script wrong USR1 signal number on mips
3079 n-i-bz vgdb ptrace calls wrong on mips [...]
3080 n-i-bz Fixes for more MPI false positives
3081 n-i-bz exp-sgcheck's memcpy causes programs to segfault
3082 n-i-bz OSX build w/ clang: asserts at startup
3083 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
3084 n-i-bz fix a couple of union tag-vs-field mixups
3085 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
3087 The following bugs were fixed in 3.8.0 but not listed in this NEWS
3090 254088 Valgrind should know about UD2 instruction
3093 304754 NEWS blows TeX's little mind
3095 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
3099 Release 3.8.0 (10 August 2012)
3100 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3101 3.8.0 is a feature release with many improvements and the usual
3102 collection of bug fixes.
3104 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3105 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
3106 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
3107 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
3108 There is initial support for MacOSX 10.8, but it is not usable for
3109 serious work at present.
3111 * ================== PLATFORM CHANGES =================
3113 * Support for MIPS32 platforms running Linux. Valgrind has been
3114 tested on MIPS32 and MIPS32r2 platforms running different Debian
3115 Squeeze and MeeGo distributions. Both little-endian and big-endian
3116 cores are supported. The tools Memcheck, Massif and Lackey have
3117 been tested and are known to work. See README.mips for more details.
3119 * Preliminary support for Android running on x86.
3121 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
3123 * Support for Intel AVX instructions and for AES instructions. This
3124 support is available only for 64 bit code.
3126 * Support for POWER Decimal Floating Point instructions.
3128 * ==================== TOOL CHANGES ====================
3130 * Non-libc malloc implementations are now supported. This is useful
3131 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
3132 Using the new option --soname-synonyms, such tools can be informed
3133 that the malloc implementation is either linked statically into the
3134 executable, or is present in some other shared library different
3135 from libc.so. This makes it possible to process statically linked
3136 programs, and programs using other malloc libraries, for example
3137 TCMalloc or JEMalloc.
3139 * For tools that provide their own replacement for malloc et al, the
3140 option --redzone-size=<number> allows users to specify the size of
3141 the padding blocks (redzones) added before and after each client
3142 allocated block. Smaller redzones decrease the memory needed by
3143 Valgrind. Bigger redzones increase the chance to detect blocks
3144 overrun or underrun. Prior to this change, the redzone size was
3145 hardwired to 16 bytes in Memcheck.
3149 - The leak_check GDB server monitor command now can
3150 control the maximum nr of loss records to output.
3152 - Reduction of memory use for applications allocating
3153 many blocks and/or having many partially defined bytes.
3155 - Addition of GDB server monitor command 'block_list' that lists
3156 the addresses/sizes of the blocks of a leak search loss record.
3158 - Addition of GDB server monitor command 'who_points_at' that lists
3159 the locations pointing at a block.
3161 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
3162 detect an invalid access of these redzones, by marking them
3163 noaccess. Similarly, if a redzone size is given for a memory
3164 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
3165 This still allows to find some bugs if the user has forgotten to
3166 mark the pool superblock noaccess.
3168 - Performance of memory leak check has been improved, especially in
3169 cases where there are many leaked blocks and/or many suppression
3170 rules used to suppress leak reports.
3172 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
3173 more precise analysis, which is important for LLVM/Clang
3174 generated code. This is at the cost of somewhat reduced
3175 performance. Note there is no change to analysis precision or
3176 costs on Linux targets.
3180 - Added even more facilities that can help finding the cause of a data
3181 race, namely the command-line option --ptrace-addr and the macro
3182 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
3184 - Fixed a subtle bug that could cause false positive data race reports.
3186 * ==================== OTHER CHANGES ====================
3188 * The C++ demangler has been updated so as to work well with C++
3189 compiled by up to at least g++ 4.6.
3191 * Tool developers can make replacement/wrapping more flexible thanks
3192 to the new option --soname-synonyms. This was reported above, but
3193 in fact is very general and applies to all function
3194 replacement/wrapping, not just to malloc-family functions.
3196 * Round-robin scheduling of threads can be selected, using the new
3197 option --fair-sched= yes. Prior to this change, the pipe-based
3198 thread serialisation mechanism (which is still the default) could
3199 give very unfair scheduling. --fair-sched=yes improves
3200 responsiveness of interactive multithreaded applications, and
3201 improves repeatability of results from the thread checkers Helgrind
3204 * For tool developers: support to run Valgrind on Valgrind has been
3205 improved. We can now routinely Valgrind on Helgrind or Memcheck.
3207 * gdbserver now shows the float shadow registers as integer
3208 rather than float values, as the shadow values are mostly
3209 used as bit patterns.
3211 * Increased limit for the --num-callers command line flag to 500.
3213 * Performance improvements for error matching when there are many
3214 suppression records in use.
3216 * Improved support for DWARF4 debugging information (bug 284184).
3218 * Initial support for DWZ compressed Dwarf debug info.
3220 * Improved control over the IR optimiser's handling of the tradeoff
3221 between performance and precision of exceptions. Specifically,
3222 --vex-iropt-precise-memory-exns has been removed and replaced by
3223 --vex-iropt-register-updates, with extended functionality. This
3224 allows the Valgrind gdbserver to always show up to date register
3227 * Modest performance gains through the use of translation chaining for
3230 * ==================== FIXED BUGS ====================
3232 The following bugs have been fixed or resolved. Note that "n-i-bz"
3233 stands for "not in bugzilla" -- that is, a bug that was reported to us
3234 but never got a bugzilla entry. We encourage you to file bugs in
3235 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3236 than mailing the developers (or mailing lists) directly -- bugs that
3237 are not entered into bugzilla tend to get forgotten about or ignored.
3239 To see details of a given bug, visit
3240 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3241 where XXXXXX is the bug number as listed below.
3243 197914 Building valgrind from svn now requires automake-1.10
3244 203877 increase to 16Mb maximum allowed alignment for memalign et al
3245 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
3246 247386 make perf does not run all performance tests
3247 270006 Valgrind scheduler unfair
3248 270777 Adding MIPS/Linux port to Valgrind
3249 270796 s390x: Removed broken support for the TS insn
3250 271438 Fix configure for proper SSE4.2 detection
3251 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
3252 273475 Add support for AVX instructions
3253 274078 improved configure logic for mpicc
3254 276993 fix mremap 'no thrash checks'
3255 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
3256 281482 memcheck incorrect byte allocation count in realloc() for silly argument
3257 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
3258 283413 Fix wrong sanity check
3259 283671 Robustize alignment computation in LibVEX_Alloc
3260 283961 Adding support for some HCI IOCTLs
3261 284124 parse_type_DIE: confused by: DWARF 4
3262 284864 == 273475 (Add support for AVX instructions)
3263 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
3264 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
3265 285725 == 273475 (Add support for AVX instructions)
3266 286261 add wrapper for linux I2C_RDWR ioctl
3267 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
3268 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
3269 286384 configure fails "checking for a supported version of gcc"
3270 286497 == 273475 (Add support for AVX instructions)
3271 286596 == 273475 (Add support for AVX instructions)
3272 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
3273 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
3274 287260 Incorrect conditional jump or move depends on uninitialised value(s)
3275 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
3276 287307 == 273475 (Add support for AVX instructions)
3277 287858 VG_(strerror): unknown error
3278 288298 (MacOSX) unhandled syscall shm_unlink
3279 288995 == 273475 (Add support for AVX instructions)
3280 289470 Loading of large Mach-O thin binaries fails.
3281 289656 == 273475 (Add support for AVX instructions)
3282 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
3283 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
3284 289839 s390x: Provide support for unicode conversion instructions
3285 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
3286 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
3287 290655 Add support for AESKEYGENASSIST instruction
3288 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
3289 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
3290 291253 ES register not initialised in valgrind simulation
3291 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
3292 291865 s390x: Support the "Compare Double and Swap" family of instructions
3293 292300 == 273475 (Add support for AVX instructions)
3294 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
3295 292493 == 273475 (Add support for AVX instructions)
3296 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
3297 292627 Missing support for some SCSI ioctls
3298 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
3299 292841 == 273475 (Add support for AVX instructions)
3300 292993 implement the getcpu syscall on amd64-linux
3301 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
3302 293088 Add some VEX sanity checks for ppc64 unhandled instructions
3303 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
3304 293754 PCMPxSTRx not implemented for 16-bit characters
3305 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
3306 293808 CLFLUSH not supported by latest VEX for amd64
3307 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
3308 294048 MPSADBW instruction not implemented
3309 294055 regtest none/tests/shell fails when locale is not set to C
3310 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
3311 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
3312 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
3313 294260 disInstr_AMD64: disInstr miscalculated next %rip
3314 294523 --partial-loads-ok=yes causes false negatives
3315 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
3316 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
3317 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
3318 295089 can not annotate source for both helgrind and drd
3319 295221 POWER Processor decimal floating point instruction support missing
3320 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
3321 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
3322 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
3323 295617 ARM - Add some missing syscalls
3324 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
3325 296229 Linux user input device ioctls missing wrappers
3326 296318 ELF Debug info improvements (more than one rx/rw mapping)
3327 296422 Add translation chaining support
3328 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
3329 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
3330 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
3331 297078 gdbserver signal handling problems [..]
3332 297147 drd false positives on newly allocated memory
3333 297329 disallow decoding of IBM Power DFP insns on some machines
3334 297497 POWER Processor decimal floating point instruction support missing
3335 297701 Another alias for strncasecmp_l in libc-2.13.so
3336 297911 'invalid write' not reported when using APIs for custom mem allocators.
3337 297976 s390x: revisit EX implementation
3338 297991 Valgrind interferes with mmap()+ftell()
3339 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
3340 297993 Fix compilation of valgrind with gcc -g3.
3341 298080 POWER Processor DFP support missing, part 3
3342 298227 == 273475 (Add support for AVX instructions)
3343 298335 == 273475 (Add support for AVX instructions)
3344 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
3345 298394 s390x: Don't bail out on an unknown machine model. [..]
3346 298421 accept4() syscall (366) support is missing for ARM
3347 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
3348 298732 valgrind installation problem in ubuntu with kernel version 3.x
3349 298862 POWER Processor DFP instruction support missing, part 4
3350 298864 DWARF reader mis-parses DW_FORM_ref_addr
3351 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
3352 299053 Support DWARF4 DW_AT_high_pc constant form
3353 299104 == 273475 (Add support for AVX instructions)
3354 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
3355 299629 dup3() syscall (358) support is missing for ARM
3356 299694 POWER Processor DFP instruction support missing, part 5
3357 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
3358 299803 == 273475 (Add support for AVX instructions)
3359 299804 == 273475 (Add support for AVX instructions)
3360 299805 == 273475 (Add support for AVX instructions)
3361 300140 ARM - Missing (T1) SMMUL
3362 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
3363 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
3364 300414 FCOM and FCOMP unimplemented for amd64 guest
3365 301204 infinite loop in canonicaliseSymtab with ifunc symbol
3366 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
3367 301265 add x86 support to Android build
3368 301984 configure script doesn't detect certain versions of clang
3369 302205 Fix compiler warnings for POWER VEX code and POWER test cases
3370 302287 Unhandled movbe instruction on Atom processors
3371 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
3372 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
3373 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
3374 302656 == 273475 (Add support for AVX instructions)
3375 302709 valgrind for ARM needs extra tls support for android emulator [..]
3376 302827 add wrapper for CDROM_GET_CAPABILITY
3377 302901 Valgrind crashes with dwz optimized debuginfo
3378 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
3379 303116 Add support for the POWER instruction popcntb
3380 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
3381 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
3382 303466 == 273475 (Add support for AVX instructions)
3383 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
3384 303963 strstr() function produces wrong results under valgrind callgrind
3385 304054 CALL_FN_xx macros need to enforce stack alignment
3386 304561 tee system call not supported
3387 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
3388 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
3389 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
3390 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
3391 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
3392 n-i-bz s390x: Shadow registers can now be examined using vgdb
3394 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
3395 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
3399 Release 3.7.0 (5 November 2011)
3400 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3401 3.7.0 is a feature release with many significant improvements and the
3402 usual collection of bug fixes.
3404 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3405 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
3406 Support for recent distros and toolchain components (glibc 2.14, gcc
3407 4.6, MacOSX 10.7) has been added.
3409 * ================== PLATFORM CHANGES =================
3411 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
3412 analyse 64-bit programs running on z/Architecture. Most user space
3413 instructions up to and including z10 are supported. Valgrind has
3414 been tested extensively on z9, z10, and z196 machines running SLES
3415 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
3416 known to work well. Callgrind, Helgrind, and DRD work reasonably
3417 well on z9 and later models. See README.s390 for more details.
3419 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
3420 64-bit processes are supported. Some complex threaded applications
3421 (Firefox) are observed to hang when run as 32 bit applications,
3422 whereas 64-bit versions run OK. The cause is unknown. Memcheck
3423 will likely report some false errors. In general, expect some rough
3424 spots. This release also supports MacOSX 10.6, but drops support
3427 * Preliminary support for Android (on ARM). Valgrind can now run
3428 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
3429 README.android for more details, plus instructions on how to get
3432 * Support for the IBM Power ISA 2.06 (Power7 instructions)
3434 * General correctness and performance improvements for ARM/Linux, and,
3435 by extension, ARM/Android.
3437 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
3438 instruction set support is under development but is not available in
3441 * Support for AIX5 has been removed.
3443 * ==================== TOOL CHANGES ====================
3445 * Memcheck: some incremental changes:
3447 - reduction of memory use in some circumstances
3449 - improved handling of freed memory, which in some circumstances
3450 can cause detection of use-after-free that would previously have
3453 - fix of a longstanding bug that could cause false negatives (missed
3454 errors) in programs doing vector saturated narrowing instructions.
3456 * Helgrind: performance improvements and major memory use reductions,
3457 particularly for large, long running applications which perform many
3458 synchronisation (lock, unlock, etc) events. Plus many smaller
3461 - display of locksets for both threads involved in a race
3463 - general improvements in formatting/clarity of error messages
3465 - addition of facilities and documentation regarding annotation
3466 of thread safe reference counted C++ classes
3468 - new flag --check-stack-refs=no|yes [yes], to disable race checking
3469 on thread stacks (a performance hack)
3471 - new flag --free-is-write=no|yes [no], to enable detection of races
3472 where one thread accesses heap memory but another one frees it,
3473 without any coordinating synchronisation event
3475 * DRD: enabled XML output; added support for delayed thread deletion
3476 in order to detect races that occur close to the end of a thread
3477 (--join-list-vol); fixed a memory leak triggered by repeated client
3478 memory allocatation and deallocation; improved Darwin support.
3480 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
3482 * exp-sgcheck: this tool has been reduced in scope so as to improve
3483 performance and remove checking that Memcheck does better.
3484 Specifically, the ability to check for overruns for stack and global
3485 arrays is unchanged, but the ability to check for overruns of heap
3486 blocks has been removed. The tool has accordingly been renamed to
3487 exp-sgcheck ("Stack and Global Array Checking").
3489 * ==================== OTHER CHANGES ====================
3491 * GDB server: Valgrind now has an embedded GDB server. That means it
3492 is possible to control a Valgrind run from GDB, doing all the usual
3493 things that GDB can do (single stepping, breakpoints, examining
3494 data, etc). Tool-specific functionality is also available. For
3495 example, it is possible to query the definedness state of variables
3496 or memory from within GDB when running Memcheck; arbitrarily large
3497 memory watchpoints are supported, etc. To use the GDB server, start
3498 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3501 * Improved support for unfriendly self-modifying code: a new option
3502 --smc-check=all-non-file is available. This adds the relevant
3503 consistency checks only to code that originates in non-file-backed
3504 mappings. In effect this confines the consistency checking only to
3505 code that is or might be JIT generated, and avoids checks on code
3506 that must have been compiled ahead of time. This significantly
3507 improves performance on applications that generate code at run time.
3509 * It is now possible to build a working Valgrind using Clang-2.9 on
3512 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3513 These enable and disable error reporting on a per-thread, and
3514 nestable, basis. This is useful for hiding errors in particularly
3515 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3516 now uses this facility.
3518 * Added the --mod-funcname option to cg_diff.
3520 * ==================== FIXED BUGS ====================
3522 The following bugs have been fixed or resolved. Note that "n-i-bz"
3523 stands for "not in bugzilla" -- that is, a bug that was reported to us
3524 but never got a bugzilla entry. We encourage you to file bugs in
3525 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3526 mailing the developers (or mailing lists) directly -- bugs that are
3527 not entered into bugzilla tend to get forgotten about or ignored.
3529 To see details of a given bug, visit
3530 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3531 where XXXXXX is the bug number as listed below.
3533 79311 malloc silly arg warning does not give stack trace
3534 210935 port valgrind.h (not valgrind) to win32 to support client requests
3535 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3536 243404 Port to zSeries
3537 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3538 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3539 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3540 253206 Some fixes for the faultstatus testcase
3541 255223 capget testcase fails when running as root
3542 256703 xlc_dbl_u32.c testcase broken
3543 256726 Helgrind tests have broken inline asm
3544 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3545 264800 testcase compile failure on zseries
3546 265762 make public VEX headers compilable by G++ 3.x
3547 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3548 266753 configure script does not give the user the option to not use QtCore
3549 266931 gen_insn_test.pl is broken
3550 266961 ld-linux.so.2 i?86-linux strlen issues
3551 266990 setns instruction causes false positive
3552 267020 Make directory for temporary files configurable at run-time.
3553 267342 == 267997 (segmentation fault on Mac OS 10.6)
3554 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3555 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3556 267488 regtest: darwin support for 64-bit build
3557 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3558 267630 Add support for IBM Power ISA 2.06 -- stage 1
3559 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3560 267819 Add client request for informing the core about reallocation
3561 267925 laog data structure quadratic for a single sequence of lock
3562 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3563 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3564 268513 missed optimizations in fold_Expr
3565 268619 s390x: fpr - gpr transfer facility
3566 268620 s390x: reconsider "long displacement" requirement
3567 268621 s390x: improve IR generation for XC
3568 268715 s390x: FLOGR is not universally available
3569 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3570 268930 s390x: MHY is not universally available
3571 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3572 269079 Support ptrace system call on ARM
3573 269144 missing "Bad option" error message
3574 269209 conditional load and store facility (z196)
3575 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3576 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3577 269736 s390x: minor code generation tweaks
3578 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3579 269863 s390x: remove unused function parameters
3580 269864 s390x: tweak s390_emit_load_cc
3581 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3582 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3583 270115 s390x: rewrite some testcases
3584 270309 == 267997 (valgrind crash on startup)
3585 270320 add support for Linux FIOQSIZE ioctl() call
3586 270326 segfault while trying to sanitize the environment passed to execle
3587 270794 IBM POWER7 support patch causes regression in none/tests
3588 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3589 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3590 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3591 270959 s390x: invalid use of R0 as base register
3592 271042 VSX configure check fails when it should not
3593 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3594 271259 s390x: fix code confusion
3595 271337 == 267997 (Valgrind segfaults on MacOS X)
3596 271385 s390x: Implement Ist_MBE
3597 271501 s390x: misc cleanups
3598 271504 s390x: promote likely and unlikely
3599 271579 ppc: using wrong enum type
3600 271615 unhandled instruction "popcnt" (arch=amd10h)
3601 271730 Fix bug when checking ioctls: duplicate check
3602 271776 s390x: provide STFLE instruction support
3603 271779 s390x: provide clock instructions like STCK
3604 271799 Darwin: ioctls without an arg report a memory error
3605 271820 arm: fix type confusion
3606 271917 pthread_cond_timedwait failure leads to not-locked false positive
3607 272067 s390x: fix DISP20 macro
3608 272615 A typo in debug output in mc_leakcheck.c
3609 272661 callgrind_annotate chokes when run from paths containing regex chars
3610 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3611 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3612 272967 make documentation build-system more robust
3613 272986 Fix gcc-4.6 warnings with valgrind.h
3614 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3615 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3616 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3617 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3618 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3619 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3620 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3621 273778 exp-ptrcheck: unhandled sysno == 259
3622 274089 exp-ptrcheck: unhandled sysno == 208
3623 274378 s390x: Various dispatcher tweaks
3624 274447 WARNING: unhandled syscall: 340
3625 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3626 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3627 274926 valgrind does not build against linux-3
3628 275148 configure FAIL with glibc-2.14
3629 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3630 275168 Make Valgrind work for MacOSX 10.7 Lion
3631 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3632 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3633 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3634 275308 Fix implementation for ppc64 fres instruc
3635 275339 s390x: fix testcase compile warnings
3636 275517 s390x: Provide support for CKSM instruction
3637 275710 s390x: get rid of redundant address mode calculation
3638 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3639 275852 == 250101 (valgrind uses all swap space and is killed)
3640 276784 Add support for IBM Power ISA 2.06 -- stage 3
3641 276987 gdbsrv: fix tests following recent commits
3642 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3643 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3644 277471 Unhandled syscall: 340
3645 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3646 277653 ARM: support Thumb2 PLD instruction
3647 277663 ARM: NEON float VMUL by scalar incorrect
3648 277689 ARM: tests for VSTn with register post-index are broken
3649 277694 ARM: BLX LR instruction broken in ARM mode
3650 277780 ARM: VMOV.F32 (immediate) instruction is broken
3651 278057 fuse filesystem syscall deadlocks
3652 278078 Unimplemented syscall 280 on ppc32
3653 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3654 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3655 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3656 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3657 279027 Support for MVCL and CLCL instruction
3658 279027 s390x: Provide support for CLCL and MVCL instructions
3659 279062 Remove a redundant check in the insn selector for ppc.
3660 279071 JDK creates PTEST with redundant REX.W prefix
3661 279212 gdbsrv: add monitor cmd v.info scheduler.
3662 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3663 279698 memcheck discards valid-bits for packuswb
3664 279795 memcheck reports uninitialised values for mincore on amd64
3665 279994 Add support for IBM Power ISA 2.06 -- stage 3
3666 280083 mempolicy syscall check errors
3667 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3668 280710 s390x: config files for nightly builds
3669 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3670 280965 Valgrind breaks fcntl locks when program does mmap
3671 281138 WARNING: unhandled syscall: 340
3672 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3673 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3674 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3675 281468 s390x: handle do_clone and gcc clones in call traces
3676 281488 ARM: VFP register corruption
3677 281828 == 275284 (false memmove warning: "Source and destination overlap")
3678 281883 s390x: Fix system call wrapper for "clone".
3679 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3680 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3681 282238 SLES10: make check fails
3682 282979 strcasestr needs replacement with recent(>=2.12) glibc
3683 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3684 283243 Regression in ppc64 memcheck tests
3685 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3686 283427 re-connect epoll_pwait syscall on ARM linux
3687 283600 gdbsrv: android: port vgdb.c
3688 283709 none/tests/faultstatus needs to account for page size
3689 284305 filter_gdb needs enhancement to work on ppc64
3690 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3691 284472 Thumb2 ROR.W encoding T2 not implemented
3692 284621 XML-escape process command line in XML output
3693 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3694 that have non-power-of-2 sizes (also AMDs)
3695 n-i-bz don't be spooked by libraries mashed by elfhack
3696 n-i-bz don't be spooked by libxul.so linked with gold
3697 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3699 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3700 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3701 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3705 Release 3.6.1 (16 February 2011)
3706 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3707 3.6.1 is a bug fix release. It adds support for some SSE4
3708 instructions that were omitted in 3.6.0 due to lack of time. Initial
3709 support for glibc-2.13 has been added. A number of bugs causing
3710 crashing or assertion failures have been fixed.
3712 The following bugs have been fixed or resolved. Note that "n-i-bz"
3713 stands for "not in bugzilla" -- that is, a bug that was reported to us
3714 but never got a bugzilla entry. We encourage you to file bugs in
3715 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3716 mailing the developers (or mailing lists) directly -- bugs that are
3717 not entered into bugzilla tend to get forgotten about or ignored.
3719 To see details of a given bug, visit
3720 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3721 where XXXXXX is the bug number as listed below.
3723 188572 Valgrind on Mac should suppress setenv() mem leak
3724 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3725 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3726 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3727 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3728 254420 memory pool tracking broken
3729 254957 Test code failing to compile due to changes in memcheck.h
3730 255009 helgrind/drd: crash on chmod with invalid parameter
3731 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3732 255355 helgrind/drd: crash on threaded programs doing fork
3734 255418 (SSE4.x) rint call compiled with ICC
3735 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3736 255888 closing valgrindoutput tag outputted to log-stream on error
3737 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3738 255966 Slowness when using mempool annotations
3739 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3740 256600 super-optimized strcasecmp() false positive
3741 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3742 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3743 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3744 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3745 257276 Missing case in memcheck --track-origins=yes
3746 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3747 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3748 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3749 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3750 263099 callgrind_annotate counts Ir improperly [...]
3751 263877 undefined coprocessor instruction on ARMv7
3752 265964 configure FAIL with glibc-2.13
3753 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3754 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3755 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3756 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3757 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3759 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3763 Release 3.6.0 (21 October 2010)
3764 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3765 3.6.0 is a feature release with many significant improvements and the
3766 usual collection of bug fixes.
3768 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3769 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3770 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3772 -------------------------
3774 Here are some highlights. Details are shown further down:
3776 * Support for ARM/Linux.
3778 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3780 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3782 * Support for the SSE4.2 instruction set.
3784 * Enhancements to the Callgrind profiler, including the ability to
3785 handle CPUs with three levels of cache.
3787 * A new experimental heap profiler, DHAT.
3789 * A huge number of bug fixes and small enhancements.
3791 -------------------------
3793 Here are details of the above changes, together with descriptions of
3794 many other changes, and a list of fixed bugs.
3796 * ================== PLATFORM CHANGES =================
3798 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3799 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3800 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3802 This requires a CPU capable of running the ARMv7-A instruction set
3803 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3804 of the user space instruction set, including ARM and Thumb integer
3805 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3806 Cachegrind and Massif tools work properly; other tools work to
3809 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3810 with support for recent releases of the underlying toolchain
3811 components, notably gcc-4.5 and glibc-2.12.
3813 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3814 support also works much better on OS X 10.5, and is as solid as
3817 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3818 64-bit mode. In 32-bit mode, support is only available up to and
3819 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3820 supported in 64-bit mode, and 32-bit mode does in fact support the
3821 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3822 10.6 on 32-bit targets.
3824 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3825 and including version 2.05 is supported.
3827 * ==================== TOOL CHANGES ====================
3829 * Cachegrind has a new processing script, cg_diff, which finds the
3830 difference between two profiles. It's very useful for evaluating
3831 the performance effects of a change in a program.
3833 Related to this change, the meaning of cg_annotate's (rarely-used)
3834 --threshold option has changed; this is unlikely to affect many
3835 people, if you do use it please see the user manual for details.
3837 * Callgrind now can do branch prediction simulation, similar to
3838 Cachegrind. In addition, it optionally can count the number of
3839 executed global bus events. Both can be used for a better
3840 approximation of a "Cycle Estimation" as derived event (you need to
3841 update the event formula in KCachegrind yourself).
3843 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3844 rather than the L2 cache. This is to accommodate machines with
3845 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3846 cache configuration of such a machine it will run the simulation as
3847 if the L2 cache isn't present. This means the results are less
3848 likely to match the true result for the machine, but
3849 Cachegrind/Callgrind's results are already only approximate, and
3850 should not be considered authoritative. The results are still
3851 useful for giving a general idea about a program's locality.
3853 * Massif has a new option, --pages-as-heap, which is disabled by
3854 default. When enabled, instead of tracking allocations at the level
3855 of heap blocks (as allocated with malloc/new/new[]), it instead
3856 tracks memory allocations at the level of memory pages (as mapped by
3857 mmap, brk, etc). Each mapped page is treated as its own block.
3858 Interpreting the page-level output is harder than the heap-level
3859 output, but this option is useful if you want to account for every
3860 byte of memory used by a program.
3862 * DRD has two new command-line options: --free-is-write and
3863 --trace-alloc. The former allows to detect reading from already freed
3864 memory, and the latter allows tracing of all memory allocations and
3867 * DRD has several new annotations. Custom barrier implementations can
3868 now be annotated, as well as benign races on static variables.
3870 * DRD's happens before / happens after annotations have been made more
3871 powerful, so that they can now also be used to annotate e.g. a smart
3872 pointer implementation.
3874 * Helgrind's annotation set has also been drastically improved, so as
3875 to provide to users a general set of annotations to describe locks,
3876 semaphores, barriers and condition variables. Annotations to
3877 describe thread-safe reference counted heap objects have also been
3880 * Memcheck has a new command-line option, --show-possibly-lost, which
3881 is enabled by default. When disabled, the leak detector will not
3882 show possibly-lost blocks.
3884 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3885 has been added. DHAT keeps track of allocated heap blocks, and also
3886 inspects every memory reference to see which block (if any) is being
3887 accessed. This gives a lot of insight into block lifetimes,
3888 utilisation, turnover, liveness, and the location of hot and cold
3889 fields. You can use DHAT to do hot-field profiling.
3891 * ==================== OTHER CHANGES ====================
3893 * Improved support for unfriendly self-modifying code: the extra
3894 overhead incurred by --smc-check=all has been reduced by
3895 approximately a factor of 5 as compared with 3.5.0.
3897 * Ability to show directory names for source files in error messages.
3898 This is combined with a flexible mechanism for specifying which
3899 parts of the paths should be shown. This is enabled by the new flag
3902 * A new flag, --require-text-symbol, which will stop the run if a
3903 specified symbol is not found it a given shared object when it is
3904 loaded into the process. This makes advanced working with function
3905 intercepting and wrapping safer and more reliable.
3907 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3908 and control of Valgrind is now available for the tools Memcheck and
3909 Helgrind. XML output from Valgrind is available for Memcheck,
3910 Helgrind and exp-Ptrcheck.
3912 * More reliable stack unwinding on amd64-linux, particularly in the
3913 presence of function wrappers, and with gcc-4.5 compiled code.
3915 * Modest scalability (performance improvements) for massive
3916 long-running applications, particularly for those with huge amounts
3919 * Support for analyzing programs running under Wine with has been
3920 improved. The header files <valgrind/valgrind.h>,
3921 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3922 Windows-programs compiled with MinGW or one of the Microsoft Visual
3925 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3926 The 32-bit simulator was not affected. This did not occur often,
3927 but when it did would usually crash the program under test.
3930 * A large number of bugs were fixed. These are shown below.
3932 * A number of bugs were investigated, and were candidates for fixing,
3933 but are not fixed in 3.6.0, due to lack of developer time. They may
3934 get fixed in later releases. They are:
3936 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3937 212419 false positive "lock order violated" (A+B vs A)
3938 213685 Undefined value propagates past dependency breaking instruction
3939 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3940 237920 valgrind segfault on fork failure
3941 242137 support for code compiled by LLVM-2.8
3942 242423 Another unknown Intel cache config value
3943 243232 Inconsistent Lock Orderings report with trylock
3944 243483 ppc: callgrind triggers VEX assertion failure
3945 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3946 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3948 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3949 249435 Analyzing wine programs with callgrind triggers a crash
3950 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3951 250065 Handling large allocations
3952 250101 huge "free" memory usage due to m_mallocfree.c
3953 "superblocks fragmentation"
3954 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3955 252091 Callgrind on ARM does not detect function returns correctly
3956 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3957 254420 memory pool tracking broken
3958 n-i-bz support for adding symbols for JIT generated code
3961 The following bugs have been fixed or resolved. Note that "n-i-bz"
3962 stands for "not in bugzilla" -- that is, a bug that was reported to us
3963 but never got a bugzilla entry. We encourage you to file bugs in
3964 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3965 mailing the developers (or mailing lists) directly -- bugs that are
3966 not entered into bugzilla tend to get forgotten about or ignored.
3968 To see details of a given bug, visit
3969 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3970 where XXXXXX is the bug number as listed below.
3972 135264 dcbzl instruction missing
3974 153699 Valgrind should report unaligned reads with movdqa
3976 190429 Valgrind reports lost of errors in ld.so
3977 with x86_64 2.9.90 glibc
3978 197266 valgrind appears to choke on the xmms instruction
3980 197988 Crash when demangling very large symbol names
3981 202315 unhandled syscall: 332 (inotify_init1)
3982 203256 Add page-level profiling to Massif
3983 205093 dsymutil=yes needs quotes, locking (partial fix)
3984 205241 Snow Leopard 10.6 support (partial fix)
3985 206600 Leak checker fails to upgrade indirect blocks when their
3986 parent becomes reachable
3987 210935 port valgrind.h (not valgrind) to win32 so apps run under
3988 wine can make client requests
3989 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3990 within Linux ip-stack checksum functions
3991 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3993 213685 Undefined value propagates past dependency breaking instruction
3995 215914 Valgrind inserts bogus empty environment variable
3997 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3998 222545 shmat fails under valgind on some arm targets
3999 222560 ARM NEON support
4002 232509 Docs build fails with formatting inside <title></title> elements
4004 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
4005 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
4006 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
4007 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
4008 237485 symlink (syscall 57) is not supported on Mac OS
4009 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
4011 238208 is_just_below_ESP doesn't take into account red-zone
4012 238345 valgrind passes wrong $0 when executing a shell script
4013 238679 mq_timedreceive syscall doesn't flag the reception buffer
4015 238696 fcntl command F_DUPFD_CLOEXEC not supported
4016 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4017 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4018 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
4020 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
4026 242606 unhandled syscall: setegid (in Ptrcheck)
4027 242814 Helgrind "Impossible has happened" during
4028 QApplication::initInstance();
4029 243064 Valgrind attempting to read debug information from iso
4030 243270 Make stack unwinding in Valgrind wrappers more reliable
4031 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
4032 sysno = 277 (mq_open)
4033 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
4034 244493 ARM VFP d16-d31 registers support
4035 244670 add support for audit_session_self syscall on Mac OS 10.6
4036 244921 The xml report of helgrind tool is not well format
4037 244923 In the xml report file, the <preamble> not escape the
4038 xml char, eg '<','&','>'
4039 245535 print full path names in plain text reports
4040 245925 x86-64 red zone handling problem
4041 246258 Valgrind not catching integer underruns + new [] s
4042 246311 reg/reg cmpxchg doesn't work on amd64
4043 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
4044 246888 Improve Makefile.vex.am
4045 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
4046 to [f]chmod_extended
4047 247526 IBM POWER6 (ISA 2.05) support is incomplete
4048 247561 Some leak testcases fails due to reachable addresses in
4050 247875 sizeofIRType to handle Ity_I128
4051 247894 [PATCH] unhandled syscall sys_readahead
4052 247980 Doesn't honor CFLAGS passed to configure
4053 248373 darwin10.supp is empty in the trunk
4054 248822 Linux FIBMAP ioctl has int parameter instead of long
4055 248893 [PATCH] make readdwarf.c big endianess safe to enable
4056 unwinding on big endian systems
4057 249224 Syscall 336 not supported (SYS_proc_info)
4059 249775 Incorrect scheme for detecting NEON capabilities of host CPU
4060 249943 jni JVM init fails when using valgrind
4061 249991 Valgrind incorrectly declares AESKEYGENASSIST support
4063 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
4064 250799 frexp$fenv_access_off function generates SIGILL
4065 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
4066 251251 support pclmulqdq insn
4067 251362 valgrind: ARM: attach to debugger either fails or provokes
4069 251674 Unhandled syscall 294
4072 254257 Add support for debugfiles found by build-id
4073 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
4074 254646 Wrapped functions cause stack misalignment on OS X
4075 (and possibly Linux)
4076 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
4078 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
4082 Release 3.5.0 (19 August 2009)
4083 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4084 3.5.0 is a feature release with many significant improvements and the
4085 usual collection of bug fixes. The main improvement is that Valgrind
4086 now works on Mac OS X.
4088 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
4089 and X86/Darwin. Support for recent distros and toolchain components
4090 (glibc 2.10, gcc 4.5) has been added.
4092 -------------------------
4094 Here is a short summary of the changes. Details are shown further
4097 * Support for Mac OS X (10.5.x).
4099 * Improvements and simplifications to Memcheck's leak checker.
4101 * Clarification and simplifications in various aspects of Valgrind's
4104 * XML output for Helgrind and Ptrcheck.
4106 * Performance and stability improvements for Helgrind and DRD.
4108 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
4110 * A new experimental tool, BBV, useful for computer architecture
4113 * Improved Wine support, including ability to read Windows PDB
4116 -------------------------
4118 Here are details of the above changes, followed by descriptions of
4119 many other minor changes, and a list of fixed bugs.
4122 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
4123 called "Darwin" because that is the name of the OS core, which is the
4124 level that Valgrind works at.)
4128 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
4129 because it would require work and 10.4 is only becoming less common.
4131 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
4132 fairly well. For 10.5.x, 32-bit programs are the default even on
4133 64-bit machines, so it handles most current programs.
4135 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
4136 officially supported, but simple programs at least will probably work.
4137 However, start-up is slow.
4139 - PowerPC machines are not supported.
4141 Things that don't work:
4143 - The Ptrcheck tool.
4145 - Objective-C garbage collection.
4149 - If you have Rogue Amoeba's "Instant Hijack" program installed,
4150 Valgrind will fail with a SIGTRAP at start-up. See
4151 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
4156 - You will likely find --dsymutil=yes a useful option, as error
4157 messages may be imprecise without it.
4159 - Mac OS X support is new and therefore will be less robust than the
4160 Linux support. Please report any bugs you find.
4162 - Threaded programs may run more slowly than on Linux.
4164 Many thanks to Greg Parker for developing this port over several years.
4167 * Memcheck's leak checker has been improved.
4169 - The results for --leak-check=summary now match the summary results
4170 for --leak-check=full. Previously they could differ because
4171 --leak-check=summary counted "indirectly lost" blocks and
4172 "suppressed" blocks as "definitely lost".
4174 - Blocks that are only reachable via at least one interior-pointer,
4175 but are directly pointed to by a start-pointer, were previously
4176 marked as "still reachable". They are now correctly marked as
4179 - The default value for the --leak-resolution option has been
4180 changed from "low" to "high". In general, this means that more
4181 leak reports will be produced, but each leak report will describe
4182 fewer leaked blocks.
4184 - With --leak-check=full, "definitely lost" and "possibly lost"
4185 leaks are now considered as proper errors, ie. they are counted
4186 for the "ERROR SUMMARY" and affect the behaviour of
4187 --error-exitcode. These leaks are not counted as errors if
4188 --leak-check=summary is specified, however.
4190 - Documentation for the leak checker has been improved.
4193 * Various aspects of Valgrind's text output have changed.
4195 - Valgrind's start-up message has changed. It is shorter but also
4196 includes the command being run, which makes it easier to use
4197 --trace-children=yes. An example:
4199 - Valgrind's shut-down messages have also changed. This is most
4200 noticeable with Memcheck, where the leak summary now occurs before
4201 the error summary. This change was necessary to allow leaks to be
4202 counted as proper errors (see the description of the leak checker
4203 changes above for more details). This was also necessary to fix a
4204 longstanding bug in which uses of suppressions against leaks were
4205 not "counted", leading to difficulties in maintaining suppression
4206 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
4208 - Behavior of -v has changed. In previous versions, -v printed out
4209 a mixture of marginally-user-useful information, and tool/core
4210 statistics. The statistics printing has now been moved to its own
4211 flag, --stats=yes. This means -v is less verbose and more likely
4212 to convey useful end-user information.
4214 - The format of some (non-XML) stack trace entries has changed a
4215 little. Previously there were six possible forms:
4217 0x80483BF: really (a.c:20)
4218 0x80483BF: really (in /foo/a.out)
4220 0x80483BF: (within /foo/a.out)
4221 0x80483BF: ??? (a.c:20)
4224 The third and fourth of these forms have been made more consistent
4225 with the others. The six possible forms are now:
4227 0x80483BF: really (a.c:20)
4228 0x80483BF: really (in /foo/a.out)
4229 0x80483BF: really (in ???)
4230 0x80483BF: ??? (in /foo/a.out)
4231 0x80483BF: ??? (a.c:20)
4234 Stack traces produced when --xml=yes is specified are different
4238 * Helgrind and Ptrcheck now support XML output, so they can be used
4239 from GUI tools. Also, the XML output mechanism has been
4242 - The XML format has been overhauled and generalised, so it is more
4243 suitable for error reporting tools in general. The Memcheck
4244 specific aspects of it have been removed. The new format, which
4245 is an evolution of the old format, is described in
4246 docs/internals/xml-output-protocol4.txt.
4248 - Memcheck has been updated to use the new format.
4250 - Helgrind and Ptrcheck are now able to emit output in this format.
4252 - The XML output mechanism has been overhauled. XML is now output
4253 to its own file descriptor, which means that:
4255 * Valgrind can output text and XML independently.
4257 * The longstanding problem of XML output being corrupted by
4258 unexpected un-tagged text messages is solved.
4260 As before, the destination for text output is specified using
4261 --log-file=, --log-fd= or --log-socket=.
4263 As before, XML output for a tool is enabled using --xml=yes.
4265 Because there's a new XML output channel, the XML output
4266 destination is now specified by --xml-file=, --xml-fd= or
4269 Initial feedback has shown this causes some confusion. To
4270 clarify, the two envisaged usage scenarios are:
4272 (1) Normal text output. In this case, do not specify --xml=yes
4273 nor any of --xml-file=, --xml-fd= or --xml-socket=.
4275 (2) XML output. In this case, specify --xml=yes, and one of
4276 --xml-file=, --xml-fd= or --xml-socket= to select the XML
4277 destination, one of --log-file=, --log-fd= or --log-socket=
4278 to select the destination for any remaining text messages,
4279 and, importantly, -q.
4281 -q makes Valgrind completely silent on the text channel,
4282 except in the case of critical failures, such as Valgrind
4283 itself segfaulting, or failing to read debugging information.
4284 Hence, in this scenario, it suffices to check whether or not
4285 any output appeared on the text channel. If yes, then it is
4286 likely to be a critical error which should be brought to the
4287 attention of the user. If no (the text channel produced no
4288 output) then it can be assumed that the run was successful.
4290 This allows GUIs to make the critical distinction they need to
4291 make (did the run fail or not?) without having to search or
4292 filter the text output channel in any way.
4294 It is also recommended to use --child-silent-after-fork=yes in
4298 * Improvements and changes in Helgrind:
4300 - XML output, as described above
4302 - Checks for consistent association between pthread condition
4303 variables and their associated mutexes are now performed.
4305 - pthread_spinlock functions are supported.
4307 - Modest performance improvements.
4309 - Initial (skeletal) support for describing the behaviour of
4310 non-POSIX synchronisation objects through ThreadSanitizer
4311 compatible ANNOTATE_* macros.
4313 - More controllable tradeoffs between performance and the level of
4314 detail of "previous" accesses in a race. There are now three
4317 * --history-level=full. This is the default, and was also the
4318 default in 3.4.x. It shows both stacks involved in a race, but
4319 requires a lot of memory and can be very slow in programs that
4320 do many inter-thread synchronisation events.
4322 * --history-level=none. This only shows the later stack involved
4323 in a race. This can be much faster than --history-level=full,
4324 but makes it much more difficult to find the other access
4325 involved in the race.
4327 The new intermediate setting is
4329 * --history-level=approx
4331 For the earlier (other) access, two stacks are presented. The
4332 earlier access is guaranteed to be somewhere in between the two
4333 program points denoted by those stacks. This is not as useful
4334 as showing the exact stack for the previous access (as per
4335 --history-level=full), but it is better than nothing, and it's
4336 almost as fast as --history-level=none.
4339 * New features and improvements in DRD:
4341 - The error messages printed by DRD are now easier to interpret.
4342 Instead of using two different numbers to identify each thread
4343 (Valgrind thread ID and DRD thread ID), DRD does now identify
4344 threads via a single number (the DRD thread ID). Furthermore
4345 "first observed at" information is now printed for all error
4346 messages related to synchronization objects.
4348 - Added support for named semaphores (sem_open() and sem_close()).
4350 - Race conditions between pthread_barrier_wait() and
4351 pthread_barrier_destroy() calls are now reported.
4353 - Added support for custom allocators through the macros
4354 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
4355 in <valgrind/valgrind.h>). An alternative for these two macros is
4356 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
4359 - Added support for annotating non-POSIX synchronization objects
4360 through several new ANNOTATE_*() macros.
4362 - OpenMP: added support for the OpenMP runtime (libgomp) included
4363 with gcc versions 4.3.0 and 4.4.0.
4367 - Added two new command-line options (--first-race-only and
4368 --segment-merging-interval).
4371 * Genuinely atomic support for x86/amd64/ppc atomic instructions
4373 Valgrind will now preserve (memory-access) atomicity of LOCK-
4374 prefixed x86/amd64 instructions, and any others implying a global
4375 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
4377 This means that Valgrinded processes will "play nicely" in
4378 situations where communication with other processes, or the kernel,
4379 is done through shared memory and coordinated with such atomic
4380 instructions. Prior to this change, such arrangements usually
4381 resulted in hangs, races or other synchronisation failures, because
4382 Valgrind did not honour atomicity of such instructions.
4385 * A new experimental tool, BBV, has been added. BBV generates basic
4386 block vectors for use with the SimPoint analysis tool, which allows
4387 a program's overall behaviour to be approximated by running only a
4388 fraction of it. This is useful for computer architecture
4389 researchers. You can run BBV by specifying --tool=exp-bbv (the
4390 "exp-" prefix is short for "experimental"). BBV was written by
4394 * Modestly improved support for running Windows applications under
4395 Wine. In particular, initial support for reading Windows .PDB debug
4396 information has been added.
4399 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
4400 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
4404 * The Valgrind client requests VALGRIND_PRINTF and
4405 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
4406 the string was always printed immediately on its own line. Now, the
4407 string will be added to a buffer but not printed until a newline is
4408 encountered, or other Valgrind output is printed (note that for
4409 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
4410 "other Valgrind output"). This allows you to use multiple
4411 VALGRIND_PRINTF calls to build up a single output line, and also to
4412 print multiple output lines with a single request (by embedding
4413 multiple newlines in the string).
4416 * The graphs drawn by Massif's ms_print program have changed slightly:
4418 - The half-height chars '.' and ',' are no longer drawn, because
4419 they are confusing. The --y option can be used if the default
4420 y-resolution is not high enough.
4422 - Horizontal lines are now drawn after the top of a snapshot if
4423 there is a gap until the next snapshot. This makes it clear that
4424 the memory usage has not dropped to zero between snapshots.
4427 * Something that happened in 3.4.0, but wasn't clearly announced: the
4428 option --read-var-info=yes can be used by some tools (Memcheck,
4429 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
4430 variable type and location information. This makes those tools
4431 start up more slowly and increases memory consumption, but
4432 descriptions of data addresses in error messages become more
4436 * exp-Omega, an experimental instantaneous leak-detecting tool, was
4437 disabled in 3.4.0 due to a lack of interest and maintenance,
4438 although the source code was still in the distribution. The source
4439 code has now been removed from the distribution. For anyone
4440 interested, the removal occurred in SVN revision r10247.
4443 * Some changes have been made to the build system.
4445 - VEX/ is now integrated properly into the build system. This means
4446 that dependency tracking within VEX/ now works properly, "make
4447 install" will work without requiring "make" before it, and
4448 parallel builds (ie. 'make -j') now work (previously a
4449 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
4450 was effectively ignored).
4452 - The --with-vex configure option has been removed. It was of
4453 little use and removing it simplified the build system.
4455 - The location of some install files has changed. This should not
4456 affect most users. Those who might be affected:
4458 * For people who use Valgrind with MPI programs, the installed
4459 libmpiwrap.so library has moved from
4460 $(INSTALL)/<platform>/libmpiwrap.so to
4461 $(INSTALL)/libmpiwrap-<platform>.so.
4463 * For people who distribute standalone Valgrind tools, the
4464 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
4465 have moved to $(INSTALL)/libcoregrind-<platform>.a.
4467 These changes simplify the build system.
4469 - Previously, all the distributed suppression (*.supp) files were
4470 installed. Now, only default.supp is installed. This should not
4471 affect users as the other installed suppression files were not
4472 read; the fact that they were installed was a mistake.
4475 * KNOWN LIMITATIONS:
4477 - Memcheck is unusable with the Intel compiler suite version 11.1,
4478 when it generates code for SSE2-and-above capable targets. This
4479 is because of icc's use of highly optimised inlined strlen
4480 implementations. It causes Memcheck to report huge numbers of
4481 false errors even in simple programs. Helgrind and DRD may also
4484 Versions 11.0 and earlier may be OK, but this has not been
4488 The following bugs have been fixed or resolved. Note that "n-i-bz"
4489 stands for "not in bugzilla" -- that is, a bug that was reported to us
4490 but never got a bugzilla entry. We encourage you to file bugs in
4491 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4492 mailing the developers (or mailing lists) directly -- bugs that are
4493 not entered into bugzilla tend to get forgotten about or ignored.
4495 To see details of a given bug, visit
4496 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4497 where XXXXXX is the bug number as listed below.
4499 84303 How about a LockCheck tool?
4500 91633 dereference of null ptr in vgPlain_st_basetype
4501 97452 Valgrind doesn't report any pthreads problems
4502 100628 leak-check gets assertion failure when using
4503 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4504 108528 NPTL pthread cleanup handlers not called
4505 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4506 110128 mallinfo is not implemented...
4507 110770 VEX: Generated files not always updated when making valgrind
4508 111102 Memcheck: problems with large (memory footprint) applications
4509 115673 Vex's decoder should never assert
4510 117564 False positive: Syscall param clone(child_tidptr) contains
4511 uninitialised byte(s)
4512 119404 executing ssh from inside valgrind fails
4513 133679 Callgrind does not write path names to sources with dwarf debug
4515 135847 configure.in problem with non gnu compilers (and possible fix)
4516 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4517 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4518 136230 memcheck reports "possibly lost", should be "still reachable"
4519 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4520 137904 Valgrind reports a memory leak when using POSIX threads,
4522 139076 valgrind VT_GETSTATE error
4523 142228 complaint of elf_dynamic_do_rela in trivial usage
4524 145347 spurious warning with USBDEVFS_REAPURB
4525 148441 (wine) can't find memory leak in Wine, win32 binary
4527 148742 Leak-check fails assert on exit
4528 149878 add (proper) check for calloc integer overflow
4529 150606 Call graph is broken when using callgrind control
4530 152393 leak errors produce an exit code of 0. I need some way to
4531 cause leak errors to result in a nonzero exit code.
4532 157154 documentation (leak-resolution doc speaks about num-callers
4533 def=4) + what is a loss record
4534 159501 incorrect handling of ALSA ioctls
4535 162020 Valgrinding an empty/zero-byte file crashes valgrind
4536 162482 ppc: Valgrind crashes while reading stabs information
4537 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4538 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4539 163560 VEX/test_main.c is missing from valgrind-3.3.1
4540 164353 malloc_usable_size() doesn't return a usable size
4541 165468 Inconsistent formatting in memcheck manual -- please fix
4542 169505 main.c:286 (endOfInstr):
4543 Assertion 'ii->cost_offset == *cost_offset' failed
4544 177206 Generate default.supp during compile instead of configure
4545 177209 Configure valt_load_address based on arch+os
4546 177305 eventfd / syscall 323 patch lost
4547 179731 Tests fail to build because of inlining of non-local asm labels
4548 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4549 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4550 181594 Bogus warning for empty text segment
4551 181707 dwarf doesn't require enumerations to have name
4552 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4553 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4554 Assertion '!already_present' failed.
4555 185359 exp-ptrcheck: unhandled syscall getresuid()
4556 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4557 185816 Valgrind is unable to handle debug info for files with split
4558 debug info that are prelinked afterwards
4559 185980 [darwin] unhandled syscall: sem_open
4560 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4561 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4562 186790 Suppression pattern used for leaks are not reported
4563 186796 Symbols with length>200 in suppression files are ignored
4564 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4565 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4566 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4567 188046 bashisms in the configure script
4568 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4569 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4570 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4571 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4572 assertion fail "!lock->heldBy"
4573 188427 Add support for epoll_create1 (with patch)
4574 188530 Support for SIOCGSTAMPNS
4575 188560 Include valgrind.spec in the tarball
4576 188572 Valgrind on Mac should suppress setenv() mem leak
4577 189054 Valgrind fails to build because of duplicate non-local asm labels
4578 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4579 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4580 189763 drd assertion failure: s_threadinfo[tid].is_recording
4581 190219 unhandled syscall: 328 (x86-linux)
4582 190391 dup of 181394; see above
4583 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4584 190820 No debug information on powerpc-linux
4585 191095 PATCH: Improve usbdevfs ioctl handling
4586 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4588 191189 --xml=yes should obey --gen-suppressions=all
4589 191192 syslog() needs a suppression on macosx
4590 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4591 191761 getrlimit on MacOSX
4592 191992 multiple --fn-skip only works sometimes; dependent on order
4593 192634 V. reports "aspacem sync_check_mapping_callback:
4594 segment mismatch" on Darwin
4595 192954 __extension__ missing on 2 client requests
4596 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4597 194474 "INSTALL" file has different build instructions than "README"
4598 194671 Unhandled syscall (sem_wait?) from mac valgrind
4599 195069 memcheck: reports leak (memory still reachable) for
4601 195169 drd: (vgDrd_barrier_post_wait):
4602 Assertion 'r->sg[p->post_iteration]' failed.
4603 195268 valgrind --log-file doesn't accept ~/...
4604 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4605 195860 WARNING: unhandled syscall: unix:223
4606 196528 need a error suppression for pthread_rwlock_init under os x?
4607 197227 Support aio_* syscalls on Darwin
4608 197456 valgrind should reject --suppressions=(directory)
4609 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4610 197591 unhandled syscall 27 (mincore)
4611 197793 Merge DCAS branch to the trunk == 85756, 142103
4612 197794 Avoid duplicate filenames in Vex
4613 197898 make check fails on current SVN
4614 197901 make check fails also under exp-ptrcheck in current SVN
4615 197929 Make --leak-resolution=high the default
4616 197930 Reduce spacing between leak reports
4617 197933 Print command line of client at start-up, and shorten preamble
4618 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4619 198395 add BBV to the distribution as an experimental tool
4620 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4621 198649 callgrind_annotate doesn't cumulate counters
4622 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4623 199977 Valgrind complains about an unrecognized instruction in the
4624 atomic_incs test program
4625 200029 valgrind isn't able to read Fedora 12 debuginfo
4626 200760 darwin unhandled syscall: unix:284
4627 200827 DRD doesn't work on Mac OS X
4628 200990 VG_(read_millisecond_timer)() does not work correctly
4629 201016 Valgrind does not support pthread_kill() on Mac OS
4630 201169 Document --read-var-info
4631 201323 Pre-3.5.0 performance sanity checking
4632 201384 Review user manual for the 3.5.0 release
4633 201585 mfpvr not implemented on ppc
4634 201708 tests failing because x86 direction flag is left set
4635 201757 Valgrind doesn't handle any recent sys_futex additions
4636 204377 64-bit valgrind can not start a shell script
4637 (with #!/path/to/shell) if the shell is a 32-bit executable
4638 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4639 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4640 about memory allocation events with memory access tracing enabled
4641 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4643 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4647 Release 3.4.1 (28 February 2009)
4648 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4649 3.4.1 is a bug-fix release that fixes some regressions and assertion
4650 failures in debug info reading in 3.4.0, most notably incorrect stack
4651 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4652 other debug info problems are also fixed. A number of bugs in the
4653 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4655 In view of the fact that 3.4.0 contains user-visible regressions
4656 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4657 encouraged to ship 3.4.1 in preference to 3.4.0.
4659 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4660 bugzilla" -- that is, a bug that was reported to us but never got a
4661 bugzilla entry. We encourage you to file bugs in bugzilla
4662 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4663 developers (or mailing lists) directly -- bugs that are not entered
4664 into bugzilla tend to get forgotten about or ignored.
4666 n-i-bz Fix various bugs reading icc-11 generated debug info
4667 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4668 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4669 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4670 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4671 179618 exp-ptrcheck crashed / exit prematurely
4672 179624 helgrind: false positive races with pthread_create and
4673 recv/open/close/read
4674 134207 pkg-config output contains @VG_PLATFORM@
4675 176926 floating point exception at valgrind startup with PPC 440EPX
4676 181594 Bogus warning for empty text segment
4677 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4678 181707 Dwarf3 doesn't require enumerations to have name
4679 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4680 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4681 Assertion '!already_present' failed.
4682 185359 exp-ptrcheck unhandled syscall getresuid()
4684 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4685 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4689 Release 3.4.0 (2 January 2009)
4690 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4691 3.4.0 is a feature release with many significant improvements and the
4692 usual collection of bug fixes. This release supports X86/Linux,
4693 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4694 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4696 3.4.0 brings some significant tool improvements. Memcheck can now
4697 report the origin of uninitialised values, the thread checkers
4698 Helgrind and DRD are much improved, and we have a new experimental
4699 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4700 global arrays. In detail:
4702 * Memcheck is now able to track the origin of uninitialised values.
4703 When it reports an uninitialised value error, it will try to show
4704 the origin of the value, as either a heap or stack allocation.
4705 Origin tracking is expensive and so is not enabled by default. To
4706 use it, specify --track-origins=yes. Memcheck's speed will be
4707 essentially halved, and memory usage will be significantly
4708 increased. Nevertheless it can drastically reduce the effort
4709 required to identify the root cause of uninitialised value errors,
4710 and so is often a programmer productivity win, despite running more
4713 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4714 3.4.0, will be released shortly.
4716 * Helgrind's race detection algorithm has been completely redesigned
4717 and reimplemented, to address usability and scalability concerns:
4719 - The new algorithm has a lower false-error rate: it is much less
4720 likely to report races that do not really exist.
4722 - Helgrind will display full call stacks for both accesses involved
4723 in a race. This makes it easier to identify the root causes of
4726 - Limitations on the size of program that can run have been removed.
4728 - Performance has been modestly improved, although that is very
4731 - Direct support for Qt4 threading has been added.
4733 - pthread_barriers are now directly supported.
4735 - Helgrind works well on all supported Linux targets.
4737 * The DRD thread debugging tool has seen major improvements:
4739 - Greatly improved performance and significantly reduced memory
4742 - Support for several major threading libraries (Boost.Thread, Qt4,
4743 glib, OpenMP) has been added.
4745 - Support for atomic instructions, POSIX semaphores, barriers and
4746 reader-writer locks has been added.
4748 - Works now on PowerPC CPUs too.
4750 - Added support for printing thread stack usage at thread exit time.
4752 - Added support for debugging lock contention.
4754 - Added a manual for Drd.
4756 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4757 checks for misuses of pointers. In that sense it is a bit like
4758 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4759 detect overruns of stack and global arrays, it can detect
4760 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4761 detect accesses heap blocks that have been freed a very long time
4762 ago (millions of blocks in the past).
4764 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4765 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4766 of the main Valgrind documentation. As this is an experimental
4767 tool, we would be particularly interested in hearing about your
4768 experiences with it.
4770 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4771 longer built by default, although the code remains in the repository
4772 and the tarball. This is due to three factors: a perceived lack of
4773 users, a lack of maintenance, and concerns that it may not be
4774 possible to achieve reliable operation using the existing design.
4776 * As usual, support for the latest Linux distros and toolchain
4777 components has been added. It should work well on Fedora Core 10,
4778 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4779 state) is supported, as is glibc-2.9. The C++ demangler has been
4780 updated so as to work well with C++ compiled by even the most recent
4783 * You can now use frame-level wildcards in suppressions. This was a
4784 frequently-requested enhancement. A line "..." in a suppression now
4785 matches zero or more frames. This makes it easier to write
4786 suppressions which are precise yet insensitive to changes in
4789 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4791 * Very basic support for IBM Power6 has been added (64-bit processes only).
4793 * Valgrind is now cross-compilable. For example, it is possible to
4794 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4795 on a ppc32/64-linux target.
4797 * You can set the main thread's stack size at startup using the
4798 new --main-stacksize= flag (subject of course to ulimit settings).
4799 This is useful for running apps that need a lot of stack space.
4801 * The limitation that you can't use --trace-children=yes together
4802 with --db-attach=yes has been removed.
4804 * The following bugs have been fixed. Note that "n-i-bz" stands for
4805 "not in bugzilla" -- that is, a bug that was reported to us but
4806 never got a bugzilla entry. We encourage you to file bugs in
4807 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4808 mailing the developers (or mailing lists) directly.
4810 n-i-bz Make return types for some client requests 64-bit clean
4811 n-i-bz glibc 2.9 support
4812 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4813 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4814 n-i-bz Building in an env without gdb gives bogus gdb attach
4815 92456 Tracing the origin of uninitialised memory
4816 106497 Valgrind does not demangle some C++ template symbols
4818 151612 Suppression with "..." (frame-level wildcards in .supp files)
4819 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4820 159285 unhandled syscall:25 (stime, on x86-linux)
4821 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4822 160954 ppc build of valgrind crashes with illegal instruction (isel)
4823 160956 mallinfo implementation, w/ patch
4824 162092 Valgrind fails to start gnome-system-monitor
4825 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4826 163794 assertion failure with "--track-origins=yes"
4827 163933 sigcontext.err and .trapno must be set together
4828 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4829 164476 Missing kernel module loading system calls
4830 164669 SVN regression: mmap() drops posix file locks
4831 166581 Callgrind output corruption when program forks
4832 167288 Patch file for missing system calls on Cell BE
4833 168943 unsupported scas instruction pentium
4834 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4835 172417 x86->IR: 0x82 ...
4836 172563 amd64->IR: 0xD9 0xF5 - fprem1
4837 173099 .lds linker script generation error
4838 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4839 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4841 174908 --log-file value not expanded correctly for core file
4842 175044 Add lookup_dcookie for amd64
4843 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4845 Developer-visible changes:
4847 * Valgrind's debug-info reading machinery has been majorly overhauled.
4848 It can now correctly establish the addresses for ELF data symbols,
4849 which is something that has never worked properly before now.
4851 Also, Valgrind can now read DWARF3 type and location information for
4852 stack and global variables. This makes it possible to use the
4853 framework to build tools that rely on knowing the type and locations
4854 of stack and global variables, for example exp-Ptrcheck.
4856 Reading of such information is disabled by default, because most
4857 tools don't need it, and because it is expensive in space and time.
4858 However, you can force Valgrind to read it, using the
4859 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4860 make use of such information, if present, to provide source-level
4861 descriptions of data addresses in the error messages they create.
4863 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4864 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).