1 Release 3.22.0 (?? Oct 2023)
2 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
5 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
6 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
7 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
8 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
9 AMD64/macOS 10.13 and nanoMIPS/Linux.
11 * ==================== CORE CHANGES ===================
13 * A new configure option --with-gdbscripts-dir lets you install
14 the gdb valgrind python monitor scripts in a specific location.
15 For example a distro could use it to install the scripts in a
16 safe load location --with-gdbscripts-dir=%{_datadir}/gdb/auto-load
17 It is also possible to configure --without-gdb-scripts-dir so no
18 .debug_gdb_scripts section is added to the vgpreload library and
19 no valgrind-monitor python scripts are installed at all.
21 * ================== PLATFORM CHANGES =================
23 * support has been added for FreeBSD 14 and FreeBSD 15.
25 * ==================== TOOL CHANGES ===================
27 * Memcheck now tests and warns about the values used for
28 alignment and size. These apply to various functions: memalign,
29 posix_memalign and aligned_alloc in C and various overloads
30 of operators new and delete in C++. The kinds of error that can
32 - invalid alignment, for instance the alignment is usually required
34 - mismatched alignment between aligned allocation and aligned
36 - mismatched size when sized delete is used
37 - bad size for functions that have implementation defined behaviour
38 when the requested size is zero
40 * ==================== FIXED BUGS ====================
42 The following bugs have been fixed or resolved. Note that "n-i-bz"
43 stands for "not in bugzilla" -- that is, a bug that was reported to us
44 but never got a bugzilla entry. We encourage you to file bugs in
45 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
46 than mailing the developers (or mailing lists) directly -- bugs that
47 are not entered into bugzilla tend to get forgotten about or ignored.
49 426751 Valgrind reports "still reachable" memory using musl (alpine running inside docker)
50 433857 Add validation to C++17 aligned new/delete alignment size
51 433859 Add mismatched detection to C++ 17 aligned new/delete
52 460192 Add epoll_pwait2
53 466105 aligned_alloc problems, part 2
54 467441 Add mismatched detection to C++ 14 sized delete
55 469049 link failure on ppc64 (big endian) valgrind 3.20
56 469146 massif --ignore-fn does not ignore inlined functions
57 469768 Make it possible to install gdb scripts in a different location
58 470121 Can't run callgrind_control with valgrind 3.21.0 because of perl errors
59 470132 s390x: Assertion failure on VGM instruction
60 470520 Multiple realloc zero errors crash in MC_(eq_Error)
61 470713 Failure on the Yosys project: valgrind: m_libcfile.c:1802
62 (Bool vgPlain_realpath(const HChar *, HChar *)):
63 Assertion 'resolved' failed
64 470830 Don't print actions vgdb me ... continue for vgdb --multi mode
65 470978 s390x: Valgrind cannot start qemu-kvm when "sysctl vm.allocate_pgste=0"
66 471311 gdb --multi mode stdout redirecting to stderr
67 471807 Add support for lazy reading and downloading of DWARF debuginfo
68 472219 Syscall param ppoll(ufds.events) points to uninitialised byte(s)
69 472963 Broken regular expression in configure.ac
70 473604 Fix bug472219.c compile failure with Clang 16
71 473677 make check compile failure with Clang 16 based on GCC 13.x
72 473745 must-be-redirected function - strlen
73 473870 FreeBSD 14 applications fail early at startup
74 473944 Handle mold linker split RW PT_LOAD segments correctly
75 n-i-bz Allow arguments with spaces in .valgrindrc files
77 To see details of a given bug, visit
78 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
79 where XXXXXX is the bug number as listed above.
81 Release 3.21.0 (28 Apr 2023)
82 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
84 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
85 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
86 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
87 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
88 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
89 AMD64/macOS 10.13 and nanoMIPS/Linux.
91 * ==================== CORE CHANGES ===================
93 * When GDB is used to debug a program running under valgrind using
94 the valgrind gdbserver, GDB will automatically load some
95 python code provided in valgrind defining GDB front end commands
96 corresponding to the valgrind monitor commands.
97 These GDB front end commands accept the same format as
98 the monitor commands directly sent to the Valgrind gdbserver.
99 These GDB front end commands provide a better integration
100 in the GDB command line interface, so as to use for example
101 GDB auto-completion, command specific help, searching for
102 a command or command help matching a regexp, ...
103 For relevant monitor commands, GDB will evaluate arguments
104 to make the use of monitor commands easier.
105 For example, instead of having to print the address of a variable
106 to pass it to a subsequent monitor command, the GDB front end
107 command will evaluate the address argument. It is for example
109 (gdb) memcheck who_points_at &some_struct sizeof(some_struct)
112 $2 = (some_struct_type *) 0x1130a0 <some_struct>
113 (gdb) p sizeof(some_struct)
115 (gdb) monitor who_point_at 0x1130a0 40
117 * The vgdb utility now supports extended-remote protocol when
118 invoked with --multi. In this mode the GDB run command is
119 supported. Which means you don't need to run gdb and valgrind
120 from different terminals. So for example to start your program
121 in gdb and run it under valgrind you can do:
123 (gdb) set remote exec-file prog
125 (gdb) target extended-remote | vgdb --multi
128 * The behaviour of realloc with a size of zero can now
129 be changed for tools that intercept malloc. Those
130 tools are memcheck, helgrind, drd, massif and dhat.
131 Realloc implementations generally do one of two things
132 - free the memory like free() and return NULL
133 (GNU libc and ptmalloc).
134 - either free the memory and then allocate a
135 minimum sized block or just return the
136 original pointer. Return NULL if the
137 allocation of the minimum sized block fails
138 (jemalloc, musl, snmalloc, Solaris, macOS).
139 When Valgrind is configured and built it will
140 try to match the OS and libc behaviour. However
141 if you are using a non-default library to replace
142 malloc and family (e.g., musl on a glibc Linux or
143 tcmalloc on FreeBSD) then you can use a command line
144 option to change the behaviour of Valgrind:
145 --realloc-zero-bytes-frees=yes|no [yes on Linux glibc, no otherwise]
147 * ================== PLATFORM CHANGES =================
149 * Make the address space limit on FreeBSD amd64 128Gbytes
150 (the same as Linux and Solaris, it was 32Gbytes)
152 * ==================== TOOL CHANGES ===================
155 - When doing a delta leak_search, it is now possible to only
156 output the new loss records compared to the previous leak search.
157 This is available in the memcheck monitor command 'leak_search'
158 by specifying the "new" keyword or in your program by using
159 the client request VALGRIND_DO_NEW_LEAK_CHECK.
160 Whenever a "delta" leak search is done (i.e. when specifying
161 "new" or "increased" or "changed" in the monitor command),
162 the new loss records have a "new" marker.
163 - Valgrind now contains python code that defines GDB memcheck
164 front end monitor commands. See CORE CHANGES.
165 - Performs checks for the use of realloc with a size of zero.
166 This is non-portable and a source of errors. If memcheck
167 detects such a usage it will generate an error
168 realloc() with size 0
169 followed by the usual callstacks.
170 A switch has been added to allow this to be turned off:
171 --show-realloc-size-zero=yes|no [yes]
174 - The option ---history-backtrace-size=<number> allows to configure
175 the number of entries to record in the stack traces of "old"
176 accesses. Previously, this number was hardcoded to 8.
177 - Valgrind now contains python code that defines GDB helgrind
178 front end monitor commands. See CORE CHANGES.
181 - `--cache-sim=no` is now the default. The cache simulation is old and
182 unlikely to match any real modern machine. This means only the `Ir`
183 event are gathered by default, but that is by far the most useful
185 - `cg_annotate`, `cg_diff`, and `cg_merge` have been rewritten in
186 Python. As a result, they all have more flexible command line
187 argument handling, e.g. supporting `--show-percs` and
188 `--no-show-percs` forms as well as the existing `--show-percs=yes`
189 and `--show-percs=no`.
190 - `cg_annotate` has some functional changes.
191 - It's much faster, e.g. 3-4x on common cases.
192 - It now supports diffing (with `--diff`, `--mod-filename`, and
193 `--mod-funcname`) and merging (by passing multiple data files).
194 - It now provides more information at the file and function level.
195 There are now "File:function" and "Function:file" sections. These
196 are very useful for programs that use inlining a lot.
197 - Support for user-annotated files and the `-I`/`--include` option
198 has been removed, because it was of little use and blocked other
200 - The `--auto` option is renamed `--annotate`, though the old
201 `--auto=yes`/`--auto=no` forms are still supported.
202 - `cg_diff` and `cg_merge` are now deprecated, because `cg_annotate`
203 now does a better job of diffing and merging.
204 - The Cachegrind output file format has changed very slightly, but in
205 ways nobody is likely to notice.
208 - Valgrind now contains python code that defines GDB callgrind
209 front end monitor commands. See CORE CHANGES.
212 - Valgrind now contains python code that defines GDB massif
213 front end monitor commands. See CORE CHANGES.
216 - A new kind of user request has been added which allows you to
217 override the 1024 byte limit on access count histograms for blocks
218 of memory. The client request is DHAT_HISTOGRAM_MEMORY.
220 * ==================== FIXED BUGS ====================
222 The following bugs have been fixed or resolved. Note that "n-i-bz"
223 stands for "not in bugzilla" -- that is, a bug that was reported to us
224 but never got a bugzilla entry. We encourage you to file bugs in
225 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
226 than mailing the developers (or mailing lists) directly -- bugs that
227 are not entered into bugzilla tend to get forgotten about or ignored.
229 170510 Don't warn about ioctl of size 0 without direction hint
230 241072 List tools in --help output
231 327548 false positive while destroying mutex
232 382034 Testcases build fixes for musl
233 351857 confusing error message about valid command line option
234 374596 inconsistent RDTSCP support on x86_64
235 392331 Spurious lock not held error from inside pthread_cond_timedwait
236 397083 Likely false positive "uninitialised value(s)" for __wmemchr_avx2 and __wmemcmp_avx2_movbe
237 400793 pthread_rwlock_timedwrlock false positive
238 419054 Unhandled syscall getcpu on arm32
239 433873 openat2 syscall unimplemented on Linux
240 434057 Add stdio mode to valgrind's gdbserver
241 435441 valgrind fails to interpose malloc on musl 1.2.2 due to weak symbol name and no libc soname
242 436413 Warn about realloc of size zero
243 439685 compiler warning in callgrind/main.c
244 444110 priv/guest_ppc_toIR.c:36198:31: warning: duplicated 'if' condition.
245 444487 hginfo test detects an extra lock inside data symbol "_rtld_local"
246 444488 Use glibc.pthread.stack_cache_size tunable
247 444568 drd/tests/pth_barrier_thr_cr fails on Fedora 38
248 445743 "The impossible happened: mutex is locked simultaneously by two threads"
249 while using mutexes with priority inheritance and signals
250 449309 Missing loopback device ioctl(s)
251 459476 vgdb: allow address reuse to avoid "address already in use" errorsuse" errors
252 460356 s390: Sqrt32Fx4 -- cannot reduce tree
253 462830 WARNING: unhandled amd64-freebsd syscall: 474
254 463027 broken check for MPX instruction support in assembler
255 464103 Enhancement: add a client request to DHAT to mark memory to be histogrammed
256 464476 Firefox fails to start under Valgrind
257 464609 Valgrind memcheck should support Linux pidfd_open
258 464680 Show issues caused by memory policies like selinux deny_execmem
259 464859 Build failures with GCC-13 (drd tsan_unittest)
260 464969 D language demangling
261 465435 m_libcfile.c:66 (vgPlain_safe_fd): Assertion 'newfd >= VG_(fd_hard_limit)' failed.
262 466104 aligned_alloc problems, part 1
263 467036 Add time cost statistics for Regtest
264 467482 Build failure on aarch64 Alpine
265 467714 fdleak_* and rlimit tests fail when parent process has more than
266 64 descriptors opened
267 467839 Gdbserver: Improve compatibility of library directory name
268 468401 [PATCH] Add a style file for clang-format
269 468556 Build failure for vgdb
270 468606 build: remove "Valgrind relies on GCC" check/output
271 469097 ppc64(be) doesn't support SCV syscall instruction
272 n-i-bz FreeBSD rfork syscall fail with EINVAL or ENOSYS rather than VG_(unimplemented)
274 To see details of a given bug, visit
275 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
276 where XXXXXX is the bug number as listed above.
278 * ==================== KNOWN ISSUES ===================
280 * configure --enable-lto=yes is know to not work in all setups.
281 See bug 469049. Workaround: Build without LTO.
283 (3.21.0.RC1: 14 Apr 2023)
284 (3.21.0.RC2: 21 Apr 2023)
286 Release 3.20.0 (24 Oct 2022)
287 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
289 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
290 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
291 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
292 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
293 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
294 AMD64/macOS 10.13 and nanoMIPS/Linux.
296 * ==================== CORE CHANGES ===================
298 * The option "--vgdb-stop-at=event1,event2,..." accepts the new value abexit.
299 This indicates to invoke gdbserver when your program exits abnormally
300 (i.e. with a non zero exit code).
301 * Fix Rust v0 name demangling.
302 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
303 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
304 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with
306 * The option --enable-debuginfod=<no|yes> [default: yes] has been added on
308 * More DWARF5 support as generated by clang14.
310 * ==================== FIXED BUGS ====================
312 The following bugs have been fixed or resolved. Note that "n-i-bz"
313 stands for "not in bugzilla" -- that is, a bug that was reported to us
314 but never got a bugzilla entry. We encourage you to file bugs in
315 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
316 than mailing the developers (or mailing lists) directly -- bugs that
317 are not entered into bugzilla tend to get forgotten about or ignored.
319 131186 writev reports error in (vector[...])
320 434764 iconv_open causes ld.so v2.28+ to use optimised strncmp
321 446754 Improve error codes from alloc functions under memcheck
322 452274 memcheck crashes with Assertion 'sci->status.what == SsIdle' failed
323 452779 Valgrind fails to build on FreeBSD 13.0 with llvm-devel (15.0.0)
324 453055 shared_timed_mutex drd test fails with "Lock shared failed" message
325 453602 Missing command line option to enable/disable debuginfod
326 452802 Handle lld 9+ split RW PT_LOAD segments correctly
327 454040 s390x: False-positive memcheck:cond in memmem on arch13 systems
328 456171 [PATCH] FreeBSD: Don't record address errors when accessing the 'kern.ps_strings' sysctl struct
329 n-i-bz Implement vgdb invoker on FreeBSD
330 458845 PowerPC: The L field for the dcbf and sync instruction should be
332 458915 Remove register cache to fix 458915 gdbserver causes wrong syscall return
333 459031 Documentation on --error-exitcode incomplete
334 459477 XERROR messages lacks ending '\n' in vgdb
335 462007 Implicit int in none/tests/faultstatus.c
337 To see details of a given bug, visit
338 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
339 where XXXXXX is the bug number as listed above.
341 (3.20.0.RC1: 20 Oct 2022)
344 Release 3.19.0 (11 Apr 2022)
345 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
347 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
348 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
349 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
350 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
351 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
352 AMD64/macOS 10.13 and nanoMIPS/Linux.
354 * ==================== CORE CHANGES ===================
356 * Fix Rust v0 name demangling.
357 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
358 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
359 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with HardenedBSD
361 * ================== PLATFORM CHANGES =================
364 - ignore the "v8.x" architecture levels, only look at actual CPU features
365 present. Fixes mismatch detected between RDMA and atomics features
366 preventing startup on some QEMU configurations.
367 - Implement LD{,A}XP and ST{,L}XP
368 - Fix incorrect code emitted for doubleword CAS.
371 - Fix sys_ipc semtimedop syscall
372 - Fix VFLRX and WFLRX instructions
373 - Fix EXRL instruction with negative offset
376 - Reimplement the vbpermq instruction support to generate less Iops and
377 avoid overflowing internal buffers.
378 - Fix checking for scv support to avoid "Facility 'SCV' unavailable (12),
379 exception" messages in dmsg.
380 - Fix setting condition code for Vector Compare quad word instructions.
381 - Fix fix lxsibzx, lxsihzx and lxsihzx instructions so they only load
382 their respective sized data.
383 - Fix the prefixed stq instruction in PC relative mode.
385 * ==================== TOOL CHANGES ===================
388 - Speed up --track-origins=yes for large (in the range of hundreds to
389 thousands of megabytes) mmap/munmaps.
391 - Several fixes for new versions of libstd++ using new posix try_lock
394 * ==================== FIXED BUGS ====================
396 The following bugs have been fixed or resolved. Note that "n-i-bz"
397 stands for "not in bugzilla" -- that is, a bug that was reported to us
398 but never got a bugzilla entry. We encourage you to file bugs in
399 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
400 than mailing the developers (or mailing lists) directly -- bugs that
401 are not entered into bugzilla tend to get forgotten about or ignored.
403 403802 leak_cpp_interior fails with some reachable blocks different than expected
404 435732 memcheck/tests/leak_cpp_interior fails with gcc11
405 444242 s390x: Valgrind crashes on EXRL with negative offset
406 444399 arm64: unhandled instruction 0xC87F2D89 (LD{,A}XP and ST{,L}XP).
408 444481 gdb_server test failures on s390x
409 444495 dhat/tests/copy fails on s390x
410 444552 memcheck/tests/sem fails on s390x with glibc 2.34
411 444571 PPC, fix the lxsibzx and lxsihzx so they only load their respective
413 444836 PPC, pstq instruction for R=1 is not storing to the correct address.
414 444925 fexecve syscall wrapper not properly implemented
415 445032 valgrind/memcheck crash with SIGSEGV when SIGVTALRM timer used and
417 445211 Fix out of tree builds
418 445300 [PATCH] Fix building tests with Musl
419 445011 SIGCHLD is sent when valgrind uses debuginfod-find
420 445354 arm64 backend: incorrect code emitted for doubleword CAS
421 445415 arm64 front end: alignment checks missing for atomic instructions
422 445504 Using C++ condition_variable results in bogus "mutex is locked simultaneously by two threads" warning
423 445607 Unhandled amd64-freebsd syscall: 247
424 445668 Inline stack frame generation is broken for Rust binaries
425 445916 Demangle Rust v0 symbols with .llvm suffix
426 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false positives
427 446138 DRD/Helgrind with std::timed_mutex::try_lock_until false positives
428 446281 Add a DRD suppression for fwrite
429 446103 Memcheck: `--track-origins=yes` causes extreme slowdowns for large mmap/munmap
430 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false
431 446251 TARGET_SIGNAL_THR added to enum target_signal
432 446823 FreeBSD - missing syscalls when using libzm4
433 447991 s390x: Valgrind indicates illegal instruction on wflrx
434 447995 Valgrind segfault on power10 due to hwcap checking code
435 449483 Powerpc: vcmpgtsq., vcmpgtuq,, vcmpequq. instructions not setting the
436 condition code correctly.
437 449672 ppc64 --track-origins=yes failures because of bad cmov addHRegUse
438 449838 sigsegv liburing the 'impossible' happened for io_uring_setup
439 450025 Powerc: ACC file not implemented as a logical overlay of the VSR
441 450437 Warn for execve syscall with argv or argv[0] being NULL
442 450536 Powerpc: valgrind throws 'facility scv unavailable exception'
443 451626 Syscall param bpf(attr->raw_tracepoint.name) points to unaddressable byte(s)
444 451827 [ppc64le] VEX temporary storage exhausted with several vbpermq instructions
445 451843 valgrind fails to start on a FreeBSD system which enforces W^X
447 To see details of a given bug, visit
448 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
449 where XXXXXX is the bug number as listed above.
451 (3.19.0.RC1: 02 Apr 2022)
452 (3.19.0.RC2: 08 Apr 2022)
455 Release 3.18.0 (15 Oct 2021)
456 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
458 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
459 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
460 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
461 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
462 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
463 AMD64/macOS 10.13 and nanoMIPS/Linux.
465 * ==================== CORE CHANGES ===================
467 * The libiberty demangler has been updated, which brings support for
468 Rust v0 name demangling. [Update: alas, due to a bug, this support
469 isn't working in 3.18.0.]
471 * __libc_freeres isn't called anymore after the program recieves a
472 fatal signal. Causing some internal glibc resources to hang around,
473 but preventing any crashes after the program has ended.
475 * The DWARF reader is now very much faster at startup when just
476 --read-inline-info=yes (the default in most cases) is given.
478 * glibc 2.34, which moved various functions from libpthread.so into
479 libc.so, is now supported.
481 * ================== PLATFORM CHANGES =================
485 - v8.2 scalar and vector FABD, FACGE, FACGT and FADD.
486 - v8.2 FP compare & conditional compare instructions.
487 - Zero variants of v8.2 FP compare instructions.
491 - Support the miscellaneous-instruction-extensions facility 3 and
492 the vector-enhancements facility 2. This enables programs
493 compiled with "-march=arch13" or "-march=z15" to be executed
498 - ISA 3.1 support is now complete
499 - ISA 3.0 support for the darn instruction added.
500 - ISA 3.0 support for the vector system call instruction scv added.
501 - ISA 3.0 support for the copy, paste and cpabort instructions added.
503 * Support for X86/FreeBSD and AMD64/FreeBSD has been added.
505 * ==================== OTHER CHANGES ====================
507 * Memcheck on amd64: minor fixes to remove some false positive
510 * ==================== FIXED BUGS ====================
512 The following bugs have been fixed or resolved. Note that "n-i-bz"
513 stands for "not in bugzilla" -- that is, a bug that was reported to us
514 but never got a bugzilla entry. We encourage you to file bugs in
515 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
516 than mailing the developers (or mailing lists) directly -- bugs that
517 are not entered into bugzilla tend to get forgotten about or ignored.
519 208531 [PATCH]: FreeBSD support for valgrind
520 368960 WARNING: unhandled amd64-linux syscall: 163 (acct)
521 407589 [Linux] Add support for C11 aligned_alloc() and GNU reallocarray()
522 423963 Error in child thread when CLONE_PIDFD is used
523 426148 crash with "impossible happened" when running BPF CO-RE programs
524 429375 PPC ISA 3.1 support is missing, part 9
525 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
526 431306 Update demangler to support Rust v0 name mangling
527 432387 s390x: z15 instructions support
528 433437 FreeBSD support, part 1
529 433438 FreeBSD support, part 2
530 433439 FreeBSD support, part 3
531 433469 FreeBSD support, part 4
532 433473 FreeBSD support, part 5
533 433477 FreeBSD support, part 6
534 433479 FreeBSD support, part 7
535 433504 FreeBSD support, part 8
536 433506 FreeBSD support, part 9
537 433507 FreeBSD support, part 10
538 433508 FreeBSD support, part 11
539 433510 FreeBSD support, part 12
540 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
541 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
542 434296 s390x: False-positive memcheck diagnostics from vector string
544 434840 PPC64 darn instruction not supported
545 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
546 435908 valgrind tries to fetch from deubginfod for files which already
547 have debug information
548 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2
549 439046 valgrind is unusably large when linked with lld
550 439090 Implement close_range(2)
551 439326 Valgrind 3.17.0 won't compile with Intel 2021 oneAPI compilers
552 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
553 440670 unhandled ppc64le-linux syscall: 252 statfs64 and 253 fstatfs64
554 440906 Fix impossible constraint issue in P10 testcase.
555 441512 Remove a unneeded / unnecessary prefix check.
556 441534 Update the expected output for test_isa_3_1_VRT.
557 442061 very slow execution under Fedora 34 (readdwarf3)
558 443031 Gcc -many change requires explicit .machine directives
559 443033 Add support for the ISA 3.0 mcrxrx instruction
560 443034 Sraw, srawi, srad, sradi, mfs
561 443178 Powerpc, test jm-mfspr expected output needs to be updated.
562 443179 Need new test for the lxvx and stxvx instructions on ISA 2.07 and
564 443180 The subnormal test and the ISA 3.0 test generate compiler warnings
565 443314 In the latest GIT version, Valgrind with "--trace-flags" crashes
567 443605 Don't call final_tidyup (__libc_freeres) on FatalSignal
569 To see details of a given bug, visit
570 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
571 where XXXXXX is the bug number as listed below.
573 (3.18.0.RC1: 12 Oct 2021)
574 (3.18.0: 15 Oct 2021)
578 Release 3.17.0 (19 Mar 2021)
579 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
581 3.17.0 fixes a number of bugs and adds some functional changes: support for
582 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
583 some new instructions for Arm64, S390 and POWER. There are also some tool
586 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
587 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
588 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
589 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
590 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
592 * ==================== CORE CHANGES ===================
594 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
597 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
598 debugging information. When a debuginfo file cannot be found locally,
599 Valgrind is able to query debuginfod servers for the file using its
600 build-id. See the user manual for more information about debuginfod support.
602 * ================== PLATFORM CHANGES =================
606 - Inaccuracies resulting from double-rounding in the simulation of
607 floating-point multiply-add/subtract instructions have been fixed. These
608 should now behave exactly as the hardware does.
610 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
611 ongoing. Support for the half-word variants of at least the following
612 instructions has been added:
614 FABS <Vd>.<T>, <Vn>.<T>
616 FNEG <Vd>.<T>, <Vn>.<T>
618 FSQRT <Vd>.<T>, <Vn>.<T>
623 - Implement the new instructions/features that were added to z/Architecture
624 with the vector-enhancements facility 1. Also cover the instructions from
625 the vector-packed-decimal facility that are defined outside the chapter
626 "Vector Decimal Instructions", but not the ones from that chapter itself.
628 For a detailed list of newly supported instructions see the updates to
629 `docs/internals/s390-opcodes.csv'.
631 Since the miscellaneous instruction extensions facility 2 was already
632 added in Valgrind 3.16.0, this completes the support necessary to run
633 general programs built with `--march=z14' under Valgrind. The
634 vector-packed-decimal facility is currently not exploited by the standard
635 toolchain and libraries.
639 - Various bug fixes. Fix for the sync field to limit setting just two of
640 the two bits in the L-field. Fix the write size for the stxsibx and
641 stxsihx instructions. Fix the modsw and modsd instructions.
643 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
644 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
645 conversion instructions are still missing.
647 * ==================== TOOL CHANGES ====================
649 * General tool changes
651 - All the tools and their vgpreload libraries are now installed under
652 libexec because they cannot be executed directly and should be run through
653 the valgrind executable. This should be an internal, not user visible,
654 change, but might impact valgrind packagers.
656 - The --track-fds option now respects -q, --quiet and won't output anything
657 if no file descriptors are leaked. It also won't report the standard stdin
658 (0), stdout (1) or stderr (2) descriptors as being leaked with
659 --trace-fds=yes anymore. To track whether the standard file descriptors
660 are still open at the end of the program run use --trace-fds=all.
664 - DHAT has been extended, with two new modes of operation. The new
665 --mode=copy flag triggers copy profiling, which records calls to memcpy,
666 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
667 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
668 the new dhat/dhat.h file. This is useful for learning more about hot code
669 paths. See the user manual for more information about the new modes.
671 - Because of these changes, DHAT's file format has changed. DHAT output
672 files produced with earlier versions of DHAT will not work with this
673 version of DHAT's viewer, and DHAT output files produced with this version
674 of DHAT will not work with earlier versions of DHAT's viewer.
676 * ==================== FIXED BUGS ====================
678 The following bugs have been fixed or resolved. Note that "n-i-bz"
679 stands for "not in bugzilla" -- that is, a bug that was reported to us
680 but never got a bugzilla entry. We encourage you to file bugs in
681 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
682 than mailing the developers (or mailing lists) directly -- bugs that
683 are not entered into bugzilla tend to get forgotten about or ignored.
685 To see details of a given bug, visit
686 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
687 where XXXXXX is the bug number as listed below.
689 140178 open("/proc/self/exe", ...); doesn't quite work
690 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
691 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
692 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
693 345077 linux syscall execveat support (linux 3.19)
694 361770 Missing F_ADD_SEALS
695 369029 handle linux syscalls sched_getattr and sched_setattr
696 384729 __libc_freeres inhibits cross-platform valgrind
697 388787 Support for C++17 new/delete
698 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
699 396656 Warnings while reading debug info
700 397605 ioctl FICLONE mishandled
701 401416 Compile failure with openmpi 4.0
702 408663 Suppression file for musl libc
703 404076 s390x: z14 vector instructions not implemented
704 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
705 (actually affected all x86 and nanomips regardless of host bitness)
706 413547 regression test does not check for Arm 64 features.
707 414268 Enable AArch64 feature detection and decoding for v8.x instructions
708 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
709 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
710 422261 platform selection fails for unqualified client name
711 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
712 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
713 423195 PPC ISA 3.1 support is missing, part 1
714 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
715 424012 crash with readv/writev having invalid but not NULL arg2 iovec
716 424298 amd64: Implement RDSEED
717 425232 PPC ISA 3.1 support is missing, part 2
718 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
719 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
720 426123 PPC ISA 3.1 support is missing, part 3
721 426144 Fix "condition variable has not been initialized" on Fedora 33.
722 427400 PPC ISA 3.1 support is missing, part 4
723 427401 PPC ISA 3.1 support is missing, part 5
724 427404 PPC ISA 3.1 support is missing, part 6
725 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
726 427787 Support new faccessat2 linux syscall (439)
727 427969 debuginfo section duplicates a section in the main ELF file
728 428035 drd: Unbreak the musl build
729 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
730 428716 cppcheck detects potential leak in VEX/useful/smchash.c
731 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
732 429352 PPC ISA 3.1 support is missing, part 7
733 429354 PPC ISA 3.1 support is missing, part 8
734 429692 unhandled ppc64le-linux syscall: 147 (getsid)
735 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
737 429952 Errors when building regtest with clang
738 430354 ppc stxsibx and stxsihx instructions write too much data
739 430429 valgrind.h doesn't compile on s390x with clang
740 430485 expr_is_guardable doesn't handle Iex_Qop
741 431556 Complete arm64 FADDP v8.2 instruction support
742 432102 Add support for DWARF5 as produced by GCC11
743 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
744 432381 drd: Process STACK_REGISTER client requests
745 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
746 432672 vg_regtest: test-specific environment variables not reset between tests
747 432809 VEX should support REX.W + POPF
748 432861 PPC modsw and modsd give incorrect results for 1 mod 12
749 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
750 432215 Add debuginfod functionality
751 433323 Use pkglibexecdir as vglibdir
752 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
753 433629 valgrind/README has type "abd" instead of "and"
754 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
755 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
756 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
757 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
758 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
760 (3.17.0.RC1: 13 Mar 2021)
761 (3.17.0.RC2: 17 Mar 2021)
762 (3.17.0: 19 Mar 2021)
766 Release 3.16.1 (22 June 2020)
767 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
769 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
770 fixes character encoding problems in the documentation HTML.
772 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
773 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
775 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
779 Release 3.16.0 (27 May 2020)
780 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
782 3.16.0 is a feature release with many improvements and the usual collection of
785 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
786 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
787 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
788 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
789 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
791 * ==================== CORE CHANGES ===================
793 * It is now possible to dynamically change the value of many command line
794 options while your program (or its children) are running under Valgrind.
796 To see the list of dynamically changeable options, run
797 "valgrind --help-dyn-options".
799 You can change the options from the shell by using vgdb to launch
800 the monitor command "v.clo <clo option>...".
801 The same monitor command can be used from a gdb connected
802 to the valgrind gdbserver.
803 Your program can also change the dynamically changeable options using
804 the client request VALGRIND_CLO_CHANGE(option).
806 * ================== PLATFORM CHANGES =================
808 * MIPS: preliminary support for nanoMIPS instruction set has been added.
810 * ==================== TOOL CHANGES ====================
814 - The implicit memcpy done by each call to realloc now counts towards the
815 read and write counts of resized heap blocks, making those counts higher
820 - cg_annotate's --auto and --show-percs options now default to 'yes', because
821 they are usually wanted.
825 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
826 because they are usually wanted.
828 - The command option --collect-systime has been enhanced to specify
829 the unit used to record the elapsed time spent during system calls.
830 The command option now accepts the values no|yes|msec|usec|nsec,
831 where yes is a synonym of msec. When giving the value nsec, the
832 system cpu time of system calls is also recorded.
836 - Several memcheck options are now dynamically changeable.
837 Use valgrind --help-dyn-options to list them.
839 - The release 3.15 introduced a backward incompatible change for
840 some suppression entries related to preadv and pwritev syscalls.
841 When reading a suppression entry using the unsupported 3.14 format,
842 valgrind will now produce a warning to say the suppression entry will not
843 work, and suggest the needed change.
845 - Significantly fewer false positive errors on optimised code generated by
846 Clang and GCC. In particular, Memcheck now deals better with the
847 situation where the compiler will transform C-level "A && B" into "B && A"
848 under certain circumstances (in which the transformation is valid).
849 Handling of integer equality/non-equality checks on partially defined
850 values is also improved on some architectures.
854 - The exprimental Stack and Global Array Checking tool has been removed.
855 It only ever worked on x86 and amd64, and even on those it had a
856 high false positive rate and was slow. An alternative for detecting
857 stack and global array overruns is using the AddressSanitizer (ASAN)
858 facility of the GCC and Clang compilers, which require you to rebuild
859 your code with -fsanitize=address.
861 * ==================== OTHER CHANGES ====================
863 * New and modified GDB server monitor features:
865 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
867 - The gdbserver monitor commands that require an address and an optional
868 length argument now accepts the alternate 'C like' syntax "address[length]".
869 For example, the memcheck command "monitor who_points_at 0x12345678 120"
870 can now also be given as "monitor who_points_at 0x12345678[120]".
872 * ==================== FIXED BUGS ====================
874 The following bugs have been fixed or resolved. Note that "n-i-bz"
875 stands for "not in bugzilla" -- that is, a bug that was reported to us
876 but never got a bugzilla entry. We encourage you to file bugs in
877 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
878 than mailing the developers (or mailing lists) directly -- bugs that
879 are not entered into bugzilla tend to get forgotten about or ignored.
881 To see details of a given bug, visit
882 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
883 where XXXXXX is the bug number as listed below.
885 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
886 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
887 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
888 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
889 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
890 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
891 400872 Add nanoMIPS support to Valgrind
892 403212 drd/tests/trylock hangs on FreeBSD
893 404406 s390x: z14 miscellaneous instructions not implemented
894 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
895 406561 mcinfcallWSRU gdbserver_test fails on ppc64
896 406824 Unsupported baseline
897 407218 Add support for the copy_file_range syscall
898 407307 Intercept stpcpy also in ld.so for arm64
899 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
901 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
902 408009 Expose rdrand and f16c even on avx if host cpu supports them
903 408091 Missing pkey syscalls
904 408414 Add support for missing for preadv2 and pwritev2 syscalls
905 409141 Valgrind hangs when SIGKILLed
906 409206 Support for Linux PPS and PTP ioctls
907 409367 exit_group() after signal to thread waiting in futex() causes hangs
908 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
909 409780 References to non-existent configure.in
910 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
911 410599 Non-deterministic behaviour of pth_self_kill_15_other test
912 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
913 411134 Allow the user to change a set of command line options during execution
914 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
915 412344 Problem setting mips flags with specific paths
916 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
917 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
918 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
919 413603 callgrind_annotate/cg_annotate truncate function names at '#'
920 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
921 415136 ARMv8.1 Compare-and-Swap instructions are not supported
922 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
923 416239 valgrind crashes when handling clock_adjtime
924 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
925 416286 DRD reports "conflicting load" error on std::mutex::lock()
926 416301 s390x: "compare and signal" not supported
927 416387 finit_module and bpf syscalls are unhandled on arm64
928 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
929 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
930 416753 new 32bit time syscalls for 2038+
931 417075 pwritev(vector[...]) suppression ignored
932 417075 is not fixed, but incompatible supp entries are detected
933 and a warning is produced for these.
934 417187 [MIPS] Conditional branch problem since 'grail' changes
935 417238 Test memcheck/tests/vbit-test fails on mips64 BE
936 417266 Make memcheck/tests/linux/sigqueue usable with musl
937 417281 s390x: /bin/true segfaults with "grail" enabled
938 417427 commit to fix vki_siginfo_t definition created numerous regression
940 417452 s390_insn_store_emit: dst->tag for HRcVec128
941 417578 Add suppressions for glibc DTV leaks
942 417906 clone with CLONE_VFORK and no CLONE_VM fails
943 418004 Grail code additions break ppc64.
944 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
945 418997 s390x: Support Iex_ITE for float and vector types
946 419503 s390x: Avoid modifying registers returned from isel functions
947 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
948 421570 std_mutex fails on Arm v8.1 h/w
949 434035 vgdb might crash if valgrind is killed
950 n-i-bz Fix minor one time leaks in dhat.
951 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
952 n-i-bz Add support for the Linux io_uring system calls
953 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
954 n-i-bz Fix non-glibc build of test suite with s390x_features
955 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
956 423195 PPC ISA 3.1 support is missing, part 1
958 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
959 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
960 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
964 Release 3.15.0 (12 April 2019)
965 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
967 3.15.0 is a feature release with many improvements and the usual collection of
970 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
971 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
972 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
973 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
974 support for X86/macOS 10.13 and AMD64/macOS 10.13.
976 * ==================== CORE CHANGES ===================
978 * The XTree Massif output format now makes use of the information obtained
979 when specifying --read-inline-info=yes.
981 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
983 * ==================== TOOL CHANGES ====================
987 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
988 it has been promoted from an experimental tool to a regular tool. Run it
989 with --tool=dhat instead of --tool=exp-dhat.
991 - DHAT now prints only minimal data when the program ends, instead writing
992 the bulk of the profiling data to a file. As a result, the --show-top-n
993 and --sort-by options have been removed.
995 - Profile results can be viewed with the new viewer, dh_view.html. When
996 a run ends, a short message is printed, explaining how to view the result.
998 - See the documentation for more details.
1002 - cg_annotate has a new option, --show-percs, which prints percentages next
1003 to all event counts.
1007 - callgrind_annotate has a new option, --show-percs, which prints percentages
1008 next to all event counts.
1010 - callgrind_annotate now inserts commas in call counts, and
1011 sort the caller/callee lists in the call tree.
1015 - The default value for --read-inline-info is now "yes" on
1016 Linux/Android/Solaris. It is still "no" on other OS.
1020 - The option --xtree-leak=yes (to output leak result in xtree format)
1021 automatically activates the option --show-leak-kinds=all, as xtree
1022 visualisation tools such as kcachegrind can in any case select what kind
1023 of leak to visualise.
1025 - There has been further work to avoid false positives. In particular,
1026 integer equality on partially defined inputs (C == and !=) is now handled
1029 * ==================== OTHER CHANGES ====================
1031 * The new option --show-error-list=no|yes displays, at the end of the run, the
1032 list of detected errors and the used suppressions. Prior to this change,
1033 showing this information could only be done by specifying "-v -v", but that
1034 also produced a lot of other possibly-non-useful messages. The option -s is
1035 equivalent to --show-error-list=yes.
1037 * ==================== FIXED BUGS ====================
1039 The following bugs have been fixed or resolved. Note that "n-i-bz"
1040 stands for "not in bugzilla" -- that is, a bug that was reported to us
1041 but never got a bugzilla entry. We encourage you to file bugs in
1042 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1043 than mailing the developers (or mailing lists) directly -- bugs that
1044 are not entered into bugzilla tend to get forgotten about or ignored.
1046 To see details of a given bug, visit
1047 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1048 where XXXXXX is the bug number as listed below.
1050 385411 s390x: z13 vector floating-point instructions not implemented
1051 397187 z13 vector register support for vgdb gdbserver
1052 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
1053 398870 Please add support for instruction vcvtps2ph
1054 399287 amd64 front end: Illegal Instruction vcmptrueps
1055 399301 Use inlined frames in Massif XTree output.
1056 399322 Improve callgrind_annotate output
1057 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
1058 400164 helgrind test encounters mips x-compiler warnings and assembler error
1059 400490 s390x: VRs allocated as if separate from FPRs
1060 400491 s390x: Operand of LOCH treated as unsigned integer
1061 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
1062 options, which specify a mips64 processor
1063 401112 LLVM 5.0 generates comparison against partially initialized data
1064 401277 More bugs in z13 support
1065 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
1066 401578 drd: crashes sometimes on fork()
1067 401627 memcheck errors with glibc avx2 optimized wcsncmp
1068 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
1069 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
1070 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
1072 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
1073 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
1074 402123 invalid assembler opcodes for mips32r2
1075 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
1076 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
1077 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
1078 402351 mips64 libvexmultiarch_test fails on s390x
1079 402369 Overhaul DHAT
1080 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
1081 402480 Do not use %rsp in clobber list
1082 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
1083 402515 Implement new option --show-error-list=no|yes / -s
1084 402519 POWER 3.0 addex instruction incorrectly implemented
1085 402781 Redo the cache used to process indirect branch targets
1086 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
1087 403552 s390x: wrong facility bit checked for vector facility
1088 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
1089 404638 Add VG_(replaceIndexXA)
1090 404843 s390x: backtrace sometimes ends prematurely
1091 404888 autotools cleanup series
1092 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
1093 405182 Valgrind fails to build with Clang
1094 405205 filter_libc: remove the line holding the futex syscall error entirely
1095 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
1096 the upper and lower 32-bits of the 64-bit result
1097 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
1098 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
1099 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
1101 405403 s390x disassembler cannot be used on x86
1102 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
1103 405458 MIPS mkFormVEC arguments swapped?
1104 405716 drd: Fix an integer overflow in the stack margin calculation
1105 405722 Support arm64 core dump
1106 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
1107 of the 64-bit destination field.
1108 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
1109 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
1110 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
1112 406256 PPC64, vector floating point instructions don't handle subnormal
1113 according to VSCR[NJ] bit setting.
1114 406352 cachegrind/callgrind fails ann tests because of missing a.c
1115 406354 dhat is broken on x86 (32bit)
1116 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
1117 406357 gdbserver_tests fails because of gdb output change
1118 406360 memcheck/tests/libstdc++.supp needs more supression variants
1119 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
1120 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
1121 407340 PPC64, does not support the vlogefp, vexptefp instructions.
1122 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
1123 n-i-bz Fix callgrind_annotate non deterministic order for equal total
1124 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
1125 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
1126 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
1128 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
1129 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
1130 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
1134 Release 3.14.0 (9 October 2018)
1135 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1137 3.14.0 is a feature release with many improvements and the usual collection of
1140 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1141 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1142 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1143 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
1144 support for X86/macOS 10.13, AMD64/macOS 10.13.
1146 * ==================== CORE CHANGES ===================
1148 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
1149 debug info for unloaded code. This allows saved stack traces (e.g. for
1150 memory leaks) to include file/line info for code that has been dlclose'd (or
1151 similar). See the user manual for more information and known limitations.
1153 * Ability to specify suppressions based on source file name and line number.
1155 * Majorly overhauled register allocator. No end-user changes, but the JIT
1156 generates code a bit more quickly now.
1158 * ================== PLATFORM CHANGES =================
1160 * Preliminary support for macOS 10.13 has been added.
1162 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
1164 * mips: support for MIPS SIMD architecture (MSA) has been added.
1166 * mips: support for MIPS N32 ABI has been added.
1168 * s390: partial support for vector instructions (integer and string) has been
1171 * ==================== TOOL CHANGES ====================
1173 * Helgrind: Addition of a flag
1174 --delta-stacktrace=no|yes [yes on linux amd64/x86]
1175 which specifies how full history stack traces should be computed.
1176 Setting this to =yes can speed up Helgrind by 25% when using
1177 --history-level=full.
1179 * Memcheck: reduced false positive rate for optimised code created by Clang 6
1180 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
1181 blocks more carefully to determine where it can avoid expensive definedness
1182 checks without loss of precision. This is controlled by the flag
1183 --expensive-definedness-checks=no|auto|yes [auto].
1185 * ==================== OTHER CHANGES ====================
1187 * Valgrind is now buildable with link-time optimisation (LTO). A new
1188 configure option --enable-lto=yes allows building Valgrind with LTO. If the
1189 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
1190 Note that if you are doing Valgrind development, --enable-lto=yes massively
1191 slows down the build process.
1193 * ==================== FIXED BUGS ====================
1195 The following bugs have been fixed or resolved. Note that "n-i-bz"
1196 stands for "not in bugzilla" -- that is, a bug that was reported to us
1197 but never got a bugzilla entry. We encourage you to file bugs in
1198 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1199 than mailing the developers (or mailing lists) directly -- bugs that
1200 are not entered into bugzilla tend to get forgotten about or ignored.
1202 To see details of a given bug, visit
1203 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1204 where XXXXXX is the bug number as listed below.
1206 79362 Debug info is lost for .so files when they are dlclose'd
1207 208052 strlcpy error when n = 0
1208 255603 exp-sgcheck Assertion '!already_present' failed
1209 338252 building valgrind with -flto (link time optimisation) fails
1210 345763 MIPS N32 ABI support
1211 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
1212 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
1213 372347 Replacement problem of the additional c++14/c++17 new/delete operators
1214 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1215 376257 helgrind history full speed up using a cached stack
1216 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
1218 379748 Fix missing pselect syscall (OS X 10.11)
1219 379754 Fix missing syscall ulock_wait (OS X 10.12)
1220 380397 s390x: __GI_strcspn() replacemenet needed
1221 381162 possible array overrun in VEX register allocator
1222 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
1223 381274 powerpc too chatty even with --sigill-diagnostics=no
1224 381289 epoll_pwait can have a NULL sigmask
1225 381553 VEX register allocator v3
1226 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
1227 381769 Use ucontext_t instead of struct ucontext
1228 381805 arm32 needs ld.so index hardwire for new glibc security fixes
1229 382256 gz compiler flag test doesn't work for gold
1230 382407 vg_perf needs "--terse" command line option
1231 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
1232 382563 MIPS MSA ASE support
1233 382998 xml-socket doesn't work
1234 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
1235 383723 Fix missing kevent_qos syscall (macOS 10.11)
1236 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
1237 384096 Mention AddrCheck at Memcheck's command line option [..]
1238 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
1239 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
1240 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
1241 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
1242 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
1243 384337 performance improvements to VEX register allocator v2 and v3
1244 384526 reduce number of spill insns generated by VEX register allocator v3
1245 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
1246 384631 Sanitise client args as printed with -v
1247 384633 Add a simple progress-reporting facility
1248 384987 VEX regalloc: allocate caller-save registers for short lived vregs
1249 385055 PPC VEX temporary storage exhausted
1250 385182 PPC64 is missing support for the DSCR
1251 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
1252 385207 PPC64, generate_store_FPRF() generates too many Iops
1253 385208 PPC64, xxperm instruction exhausts temporary memory
1254 385210 PPC64, vpermr instruction could exhaust temporary memory
1255 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
1256 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
1257 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
1258 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
1259 385408 s390x: z13 vector "support" instructions not implemented
1260 385409 s390x: z13 vector integer instructions not implemented
1261 385410 s390x: z13 vector string instructions not implemented
1262 385412 s390x: new non-vector z13 instructions not implemented
1263 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
1264 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
1265 385939 Optionally exit on the first error
1266 386318 valgrind.org/info/tools.html is missing SGCheck
1267 386425 running valgrind + wine on armv7l gives illegal opcode
1268 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
1269 387410 MIPSr6 support
1270 387664 Memcheck: make expensive-definedness-checks be the default
1271 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
1272 387766 asm shifts cause false positive "Conditional jump or move depends
1273 on uninitialised value"
1274 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
1275 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
1276 388786 Support bpf syscall in amd64 Linux
1277 388862 Add replacements for wmemchr and wcsnlen on Linux
1278 389065 valgrind meets gcc flag -Wlogical-op
1279 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
1280 390471 suppression by specification of source-file line number
1281 390723 make xtree dump files world wide readable, similar to log files
1282 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
1283 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
1284 392118 unhandled amd64-linux syscall: 332 (statx)
1285 392449 callgrind not clearing the number of calls properly
1286 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
1287 lxssp, stxssp and stxvl instructions.
1288 393023 callgrind_control risks using the wrong vgdb
1289 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
1290 393099 posix_memalign() invalid write if alignment == 0
1291 393146 failing assert "is_DebugInfo_active(di)"
1292 395709 PPC64 is missing support for the xvnegsp instruction
1293 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
1295 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
1296 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
1297 396839 s390x: Trap instructions not implemented
1298 396887 arch_prctl should return EINVAL on unknown option
1299 == 397286 crash before launching binary (Unsupported arch_prctl option)
1300 == 397393 valgrind: the 'impossible' happened: (Archlinux)
1301 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
1302 396906 compile tests failure on mips32-linux: broken inline asm in tests on
1304 397012 glibc ld.so uses arch_prctl on i386
1305 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
1306 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
1307 397424 glibc 2.27 and gdb_server tests
1308 398028 Assertion `cfsi_fits` failing in simple C program
1309 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
1311 n-i-bz Fix missing workq_ops operations (macOS)
1312 n-i-bz fix bug in strspn replacement
1313 n-i-bz Add support for the Linux BLKFLSBUF ioctl
1314 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
1315 n-i-bz Fix possible stack trashing by semctl syscall wrapping
1316 n-i-bz Add support for the Linux membarrier() system call
1317 n-i-bz x86 front end: recognise and handle UD2 correctly
1318 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
1319 correctly aligned before entering the handler.
1321 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
1322 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
1323 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
1327 Release 3.13.0 (15 June 2017)
1328 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1330 3.13.0 is a feature release with many improvements and the usual collection of
1333 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1334 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1335 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1336 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
1338 * ==================== CORE CHANGES ===================
1340 * The translation cache size has been increased to keep up with the demands of
1341 large applications. The maximum number of sectors has increased from 24 to
1342 48. The default number of sectors has increased from 16 to 32 on all
1343 targets except Android, where the increase is from 6 to 12.
1345 * The amount of memory that Valgrind can use has been increased from 64GB to
1346 128GB. In particular this means your application can allocate up to about
1347 60GB when running on Memcheck.
1349 * Valgrind's default load address has been changed from 0x3800'0000 to
1350 0x5800'0000, so as to make it possible to load larger executables. This
1351 should make it possible to load executables of size at least 1200MB.
1353 * A massive spaceleak caused by reading compressed debuginfo files has been
1354 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
1357 * The C++ demangler has been updated.
1359 * Support for demangling Rust symbols has been added.
1361 * A new representation of stack traces, the "XTree", has been added. An XTree
1362 is a tree of stacktraces with data associated with the stacktraces. This is
1363 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
1364 consumption of your program. Reporting is controlled by the new options
1365 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
1367 A report can also be produced on demand using the gdbserver monitor command
1368 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
1369 format' and 'massif format. The existing visualisers for these formats (e.g.
1370 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
1371 analyse these reports.
1373 Memcheck can also produce XTree leak reports using the Callgrind file
1374 format. For more details, see the user manual.
1376 * ================== PLATFORM CHANGES =================
1378 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
1380 * amd64: fixes for JIT failure problems on long AVX2 code blocks
1382 * amd64 and x86: support for CET prefixes has been added
1384 * arm32: a few missing ARMv8 instructions have been implemented
1386 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
1387 Store-Conditional instructions has been added. This is to deal with
1388 processor implementations that implement the LL/SC specifications strictly
1389 and as a result cause Valgrind to hang in certain situations. The
1390 alternative implementation is automatically enabled at startup, as required.
1391 You can use the option --sim-hints=fallback-llsc to force-enable it if you
1394 * Support for OSX 10.12 has been improved.
1396 * On Linux, clone handling has been improved to honour CLONE_VFORK that
1397 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
1398 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
1399 CLONE_VM exact semantics will (still) not work.
1401 * The TileGX/Linux port has been removed because it appears to be both unused
1404 * ==================== TOOL CHANGES ====================
1408 - Memcheck should give fewer false positives when running optimised
1409 Clang/LLVM generated code.
1411 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1413 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
1414 to produce the end of execution leak report in a xtree callgrind format
1417 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
1418 the leak report in an xtree file.
1422 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1424 - For some workloads (typically, for big applications), Massif memory
1425 consumption and CPU consumption has decreased significantly.
1429 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1431 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
1432 for Ada gnat compiled applications.
1434 * ==================== OTHER CHANGES ====================
1436 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
1437 append the inner guest stacktrace to the inner host stacktrace. This helps
1438 to investigate the errors reported by the outer, when they are caused by the
1439 inner guest program (such as an inner regtest). See README_DEVELOPERS for
1442 * To allow fast detection of callgrind files by desktop environments and file
1443 managers, the format was extended to have an optional first line that
1444 uniquely identifies the format ("# callgrind format"). Callgrind creates
1445 this line now, as does the new xtree functionality.
1447 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
1448 have a new %n format letter that is replaced by a sequence number.
1450 * "--version -v" now shows the SVN revision numbers from which Valgrind was
1453 * ==================== FIXED BUGS ====================
1455 The following bugs have been fixed or resolved. Note that "n-i-bz"
1456 stands for "not in bugzilla" -- that is, a bug that was reported to us
1457 but never got a bugzilla entry. We encourage you to file bugs in
1458 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1459 than mailing the developers (or mailing lists) directly -- bugs that
1460 are not entered into bugzilla tend to get forgotten about or ignored.
1462 To see details of a given bug, visit
1463 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1464 where XXXXXX is the bug number as listed below.
1466 162848 --log-file output isn't split when a program forks
1467 340777 Illegal instruction on mips (ar71xx)
1468 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
1469 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
1470 to a different stack.
1471 344139 x86 stack-seg overrides, needed by the Wine people
1472 344524 store conditional of guest applications always fail - observed on
1474 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
1475 352395 Please provide SVN revision info in --version -v
1476 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
1477 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
1478 INVALID_POSIX_THREADID' failed
1479 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
1480 barrier implementation
1481 358697 valgrind.h: Some code remains even when defining NVALGRIND
1482 359202 Add musl libc configure/compile
1483 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
1484 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
1485 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
1486 362223 assertion failed when .valgrindrc is a directory instead of a file
1487 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
1488 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
1489 368507 can't malloc chunks larger than about 34GB
1490 368529 Android arm target link error, missing atexit and pthread_atfork
1491 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
1492 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
1493 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
1494 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
1495 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
1496 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
1497 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
1498 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
1499 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
1500 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
1501 370028 Reduce the number of compiler warnings on MIPS platforms
1502 370635 arm64 missing syscall getcpu
1503 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
1504 371227 Clean AArch64 syscall table
1505 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
1506 371471 Valgrind complains about non legit memory leaks on placement new (C++)
1507 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
1508 371503 disInstr(arm64): unhandled instruction 0xF89F0000
1509 371869 support '%' in symbol Z-encoding
1510 371916 execution tree xtree concept
1511 372120 c++ demangler demangles symbols which are not c++
1512 372185 Support of valgrind on ARMv8 with 32 bit executable
1513 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1514 372195 Power PC, xxsel instruction is not always recognized.
1515 372504 Hanging on exit_group
1516 372600 process loops forever when fatal signals are arriving quickly
1517 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1518 373046 Stacks registered by core are never deregistered
1519 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1520 373086 Implement additional Xen hypercalls
1521 373192 Calling posix_spawn in glibc 2.24 completely broken
1522 373488 Support for fanotify API on ARM64 architecture
1523 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1524 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1525 373938 const IRExpr arguments for matchIRExpr()
1526 374719 some spelling fixes
1527 374963 increase valgrind's load address to prevent mmap failure
1528 375514 valgrind_get_tls_addr() does not work in case of static TLS
1529 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1530 for ML_(find_rx_mapping)()
1531 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1532 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1533 == 377159 "vex: the `impossible' happened" still present
1534 == 375150 Assertion 'tres.status == VexTransOK' failed
1535 == 378068 valgrind crashes on AVX2 function in FFmpeg
1536 376142 Segfaults on MIPS Cavium Octeon boards
1537 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1538 376455 Solaris: unhandled syscall lgrpsys(180)
1539 376518 Solaris: unhandled fast trap getlgrp(6)
1540 376611 ppc64 and arm64 don't know about prlimit64 syscall
1541 376729 PPC64, remove R2 from the clobber list
1543 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1544 to be wrongly marked as addressable
1545 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1546 PIE enabled by default
1547 377376 memcheck/tests/linux/getregset fails with glibc2.24
1548 377427 PPC64, lxv instruction failing on odd destination register
1549 377478 PPC64: ISA 3.0 setup fixes
1550 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1551 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1552 and 2 args for FUTEX_TRYLOCK_PI
1553 377717 Fix massive space leak when reading compressed debuginfo sections
1554 377891 Update Xen 4.6 domctl wrappers
1555 377930 fcntl syscall wrapper is missing flock structure check
1556 378524 libvexmultiarch_test regression on s390x and ppc64
1557 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1558 378673 Update libiberty demangler
1559 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1560 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1561 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1562 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1563 (task_register_dyld_image_infos)
1564 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1565 (task_register_dyld_shared_cache_image_info)
1566 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1567 379473 MIPS: add support for rdhwr cycle counter register
1568 379504 remove TileGX/Linux port
1569 379525 Support more x86 nop opcodes
1570 379838 disAMode(x86): not an addr!
1571 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1572 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1573 379895 clock_gettime does not execute POST syscall wrapper
1574 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1575 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1576 380200 xtree generated callgrind files refer to files without directory name
1577 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1578 380397 s390x: __GI_strcspn() replacement needed
1579 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1581 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1582 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1583 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1587 Release 3.12.0 (20 October 2016)
1588 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1590 3.12.0 is a feature release with many improvements and the usual
1591 collection of bug fixes.
1593 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1594 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1595 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1596 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1597 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1598 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1600 * ================== PLATFORM CHANGES =================
1602 * POWER: Support for ISA 3.0 has been added
1604 * mips: support for O32 FPXX ABI has been added.
1605 * mips: improved recognition of different processors
1606 * mips: determination of page size now done at run time
1608 * amd64: Partial support for AMD FMA4 instructions.
1610 * arm, arm64: Support for v8 crypto and CRC instructions.
1612 * Improvements and robustification of the Solaris port.
1614 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1616 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1617 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1618 where possible. Valgrind's support for 32-bit x86 has stagnated in
1619 recent years and has fallen far behind that for 64-bit x86
1620 instructions. By contrast 64-bit x86 is well supported, up to and
1623 * ==================== TOOL CHANGES ====================
1627 - Added meta mempool support for describing a custom allocator which:
1628 - Auto-frees all chunks assuming that destroying a pool destroys all
1630 - Uses itself to allocate other memory blocks
1632 - New flag --ignore-range-below-sp to ignore memory accesses below
1633 the stack pointer, if you really have to. The related flag
1634 --workaround-gcc296-bugs=yes is now deprecated. Use
1635 --ignore-range-below-sp=1024-1 as a replacement.
1639 - Improved thread startup time significantly on non-Linux platforms.
1643 - Added collection of the metric "tot-blocks-allocd"
1645 * ==================== OTHER CHANGES ====================
1647 * Replacement/wrapping of malloc/new related functions is now done not just
1648 for system libraries by default, but for any globally defined malloc/new
1649 related function (both in shared libraries and statically linked alternative
1650 malloc implementations). The dynamic (runtime) linker is excluded, though.
1651 To only intercept malloc/new related functions in
1652 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1653 "nouserintercepts" can be any non-existing library name).
1654 This new functionality is not implemented for MacOS X.
1656 * The maximum number of callers in a suppression entry is now equal to
1657 the maximum size for --num-callers (500).
1658 Note that --gen-suppressions=yes|all similarly generates suppressions
1659 containing up to --num-callers frames.
1661 * New and modified GDB server monitor features:
1663 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1664 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1667 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1668 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1671 * Valgrind is able to read compressed debuginfo sections in two formats:
1672 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1673 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1675 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1676 for the most common use case (x86_64-linux, Memcheck) has been
1679 * Improved performance for programs that do a lot of discarding of
1680 instruction address ranges of 8KB or less.
1682 * The C++ symbol demangler has been updated.
1684 * More robustness against invalid syscall parameters on Linux.
1686 * ==================== FIXED BUGS ====================
1688 The following bugs have been fixed or resolved. Note that "n-i-bz"
1689 stands for "not in bugzilla" -- that is, a bug that was reported to us
1690 but never got a bugzilla entry. We encourage you to file bugs in
1691 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1692 than mailing the developers (or mailing lists) directly -- bugs that
1693 are not entered into bugzilla tend to get forgotten about or ignored.
1695 To see details of a given bug, visit
1696 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1697 where XXXXXX is the bug number as listed below.
1699 191069 Exiting due to signal not reported in XML output
1700 199468 Suppressions: stack size limited to 25
1701 while --num-callers allows more frames
1702 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1703 278744 cvtps2pd with redundant RexW
1704 303877 valgrind doesn't support compressed debuginfo sections.
1705 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1706 348345 Assertion fails for negative lineno
1707 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1708 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1709 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1710 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1711 352197 mips: mmap2() not wrapped correctly for page size > 4096
1712 353083 arm64 doesn't implement various xattr system calls
1713 353084 arm64 doesn't support sigpending system call
1714 353137 www: update info for Supported Platforms
1715 353138 www: update "The Valgrind Developers" page
1716 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1719 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1720 353398 WARNING: unhandled amd64-solaris syscall: 207
1721 353660 XML in auxwhat tag not escaping reserved symbols properly
1722 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1723 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1724 353802 ELF debug info reader confused with multiple .rodata sections
1725 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1726 353917 unhandled amd64-solaris syscall fchdir(120)
1727 353920 unhandled amd64-solaris syscall: 170
1728 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1729 354392 unhandled amd64-solaris syscall: 171
1730 354797 Vbit test does not include Iops for Power 8 instruction support
1731 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1735 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1736 355188 valgrind should intercept all malloc related global functions
1737 355454 do not intercept malloc related symbols from the runtime linker
1738 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1739 356044 Dwarf line info reader misinterprets is_stmt register
1740 356112 mips: replace addi with addiu
1741 356393 valgrind (vex) crashes because isZeroU happened
1744 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1745 356678 arm64-linux: unhandled syscall 232 (mincore)
1746 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1747 356823 Unsupported ARM instruction: stlex
1748 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1749 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1750 357673 crash if I try to run valgrind with a binary link with libcurl
1751 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1752 357871 pthread_spin_destroy not properly wrapped
1753 357887 Calls to VG_(fclose) do not close the file descriptor
1754 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1755 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1756 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1757 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1758 359181 Buffer Overflow during Demangling
1759 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1760 359289 s390x: popcnt (B9E1) not implemented
1761 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1762 359503 Add missing syscalls for aarch64 (arm64)
1763 359645 "You need libc6-dbg" help message could be more helpful
1764 359703 s390: wire up separate socketcalls system calls
1765 359724 getsockname might crash - deref_UInt should call safe_to_deref
1766 359733 amd64 implement ld.so strchr/index override like x86
1767 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1768 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1770 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1771 359871 Incorrect mask handling in ppoll
1772 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1773 360008 Contents of Power vr registers contents is not printed correctly when
1774 the --vgdb-shadow-registers=yes option is used
1775 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1776 non-zero shadow bits
1777 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1778 360425 arm64 unsupported instruction ldpsw
1780 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1781 360571 Error about the Android Runtime reading below the stack pointer on ARM
1782 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1783 360749 kludge for multiple .rodata sections on Solaris no longer needed
1784 360752 raise the number of reserved fds in m_main.c from 10 to 12
1785 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1786 361226 s390x: risbgn (EC59) not implemented
1787 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1788 361354 ppc64[le]: wire up separate socketcalls system calls
1789 361615 Inconsistent termination for multithreaded process terminated by signal
1790 361926 Unhandled Solaris syscall: sysfs(84)
1791 362009 V dumps core on unimplemented functionality before threads are created
1792 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1793 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1794 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1795 362953 Request for an update to the Valgrind Developers page
1796 363680 add renameat2() support
1797 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1798 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1799 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1800 364058 clarify in manual limitations of array overruns detections
1801 364413 pselect sycallwrapper mishandles NULL sigmask
1802 364728 Power PC, missing support for several HW registers in
1803 get_otrack_shadow_offset_wrk()
1804 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1805 365273 Invalid write to stack location reported after signal handler runs
1806 365912 ppc64BE segfault during jm-insns test (RELRO)
1807 366079 FPXX Support for MIPS32 Valgrind
1808 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1809 366344 Multiple unhandled instruction for Aarch64
1810 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1811 367995 Integration of memcheck with custom memory allocator
1812 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1813 368412 False positive result for altivec capability check
1814 368416 Add tc06_two_races_xml.exp output for ppc64
1815 368419 Perf Events ioctls not implemented
1816 368461 mmapunmap test fails on ppc64
1817 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1818 369000 AMD64 fma4 instructions unsupported.
1819 369169 ppc64 fails jm_int_isa_2_07 test
1820 369175 jm_vec_isa_2_07 test crashes on ppc64
1821 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1822 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1823 369359 msghdr_foreachfield can crash when handling bad iovec
1824 369360 Bad sigprocmask old or new sets can crash valgrind
1825 369361 vmsplice syscall wrapper crashes on bad iovec
1826 369362 Bad sigaction arguments crash valgrind
1827 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1828 369402 Bad set/get_thread_area pointer crashes valgrind
1829 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1830 369446 valgrind crashes on unknown fcntl command
1831 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1832 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1833 370265 ISA 3.0 HW cap stuff needs updating
1834 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1835 372195 Power PC, xxsel instruction is not always recognized
1837 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1838 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1839 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1840 n-i-bz Fix ppoll_alarm exclusion on OS X
1841 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1842 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1843 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1844 n-i-bz amd64: memcheck false positive with shr %edx
1845 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1846 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1847 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1848 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1850 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1851 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1852 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1856 Release 3.11.0 (22 September 2015)
1857 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1859 3.11.0 is a feature release with many improvements and the usual
1860 collection of bug fixes.
1862 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1863 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1864 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1865 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1866 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1867 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1869 * ================== PLATFORM CHANGES =================
1871 * Support for Solaris/x86 and Solaris/amd64 has been added.
1873 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1875 * Preliminary support for the Tilera TileGX architecture has been added.
1877 * s390x: It is now required for the host to have the "long displacement"
1878 facility. The oldest supported machine model is z990.
1880 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1881 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1883 * The JIT's register allocator is significantly faster, making the JIT
1884 as a whole somewhat faster, so JIT-intensive activities, for example
1885 program startup, are modestly faster, around 5%.
1887 * There have been changes to the default settings of several command
1888 line flags, as detailed below.
1890 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1891 capable hosts, the simulated CPUID will now indicate AVX2 support.
1893 * ==================== TOOL CHANGES ====================
1897 - The default value for --leak-check-heuristics has been changed from
1898 "none" to "all". This helps to reduce the number of possibly
1899 lost blocks, in particular for C++ applications.
1901 - The default value for --keep-stacktraces has been changed from
1902 "malloc-then-free" to "malloc-and-free". This has a small cost in
1903 memory (one word per malloc-ed block) but allows Memcheck to show the
1904 3 stacktraces of a dangling reference: where the block was allocated,
1905 where it was freed, and where it is acccessed after being freed.
1907 - The default value for --partial-loads-ok has been changed from "no" to
1908 "yes", so as to avoid false positive errors resulting from some kinds
1909 of vectorised loops.
1911 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1912 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1913 than get_vbits when you need to associate byte data value with
1914 their corresponding validity bits.
1916 - The 'block_list' monitor command has been enhanced:
1917 o it can print a range of loss records
1918 o it now accepts an optional argument 'limited <max_blocks>'
1919 to control the number of blocks printed.
1920 o if a block has been found using a heuristic, then
1921 'block_list' now shows the heuristic after the block size.
1922 o the loss records/blocks to print can be limited to the blocks
1923 found via specified heuristics.
1925 - The C helper functions used to instrument loads on
1926 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1927 replaced by handwritten assembly sequences. This gives speedups
1928 in the region of 0% to 7% for those targets only.
1930 - A new command line option, --expensive-definedness-checks=yes|no,
1931 has been added. This is useful for avoiding occasional invalid
1932 uninitialised-value errors in optimised code. Watch out for
1933 runtime degradation, as this can be up to 25%. As always, though,
1934 the slowdown is highly application specific. The default setting
1939 - A new monitor command 'all_snapshots <filename>' dumps all
1940 snapshots taken so far.
1944 - Significant memory reduction and moderate speedups for
1945 --history-level=full for applications accessing a lot of memory
1946 with many different stacktraces.
1948 - The default value for --conflict-cache-size=N has been doubled to
1949 2000000. Users that were not using the default value should
1950 preferably also double the value they give.
1952 The default was changed due to the changes in the "full history"
1953 implementation. Doubling the value gives on average a slightly more
1954 complete history and uses similar memory (or significantly less memory
1955 in the worst case) than the previous implementation.
1957 - The Helgrind monitor command 'info locks' now accepts an optional
1958 argument 'lock_addr', which shows information about the lock at the
1961 - When using --history-level=full, the new Helgrind monitor command
1962 'accesshistory <addr> [<len>]' will show the recorded accesses for
1963 <len> (or 1) bytes at <addr>.
1965 * ==================== OTHER CHANGES ====================
1967 * The default value for the --smc-check option has been changed from
1968 "stack" to "all-non-file" on targets that provide automatic D-I
1969 cache coherence (x86, amd64 and s390x). The result is to provide,
1970 by default, transparent support for JIT generated and self-modifying
1971 code on all targets.
1973 * Mac OS X only: the default value for the --dsymutil option has been
1974 changed from "no" to "yes", since any serious usage on Mac OS X
1975 always required it to be "yes".
1977 * The command line options --db-attach and --db-command have been removed.
1978 They were deprecated in 3.10.0.
1980 * When a process dies due to a signal, Valgrind now shows the signal
1981 and the stacktrace at default verbosity (i.e. verbosity 1).
1983 * The address description logic used by Memcheck and Helgrind now
1984 describes addresses in anonymous segments, file mmap-ed segments,
1985 shared memory segments and the brk data segment.
1987 * The new option --error-markers=<begin>,<end> can be used to mark the
1988 begin/end of errors in textual output mode, to facilitate
1989 searching/extracting errors in output files that mix valgrind errors
1990 with program output.
1992 * The new option --max-threads=<number> can be used to change the number
1993 of threads valgrind can handle. The default is 500 threads which
1994 should be more than enough for most applications.
1996 * The new option --valgrind-stacksize=<number> can be used to change the
1997 size of the private thread stacks used by Valgrind. This is useful
1998 for reducing memory use or increasing the stack size if Valgrind
1999 segfaults due to stack overflow.
2001 * The new option --avg-transtab-entry-size=<number> can be used to specify
2002 the expected instrumented block size, either to reduce memory use or
2003 to avoid excessive retranslation.
2005 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
2007 * New and modified GDB server monitor features:
2009 - When a signal is reported in GDB, you can now use the GDB convenience
2010 variable $_siginfo to examine detailed signal information.
2012 - Valgrind's gdbserver now allows the user to change the signal
2013 to deliver to the process. So, use 'signal SIGNAL' to continue execution
2014 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
2015 continue without passing the signal to the process.
2017 - With GDB >= 7.10, the command 'target remote'
2018 will automatically load the executable file of the process running
2019 under Valgrind. This means you do not need to specify the executable
2020 file yourself, GDB will discover it itself. See GDB documentation about
2021 'qXfer:exec-file:read' packet for more info.
2023 * ==================== FIXED BUGS ====================
2025 The following bugs have been fixed or resolved. Note that "n-i-bz"
2026 stands for "not in bugzilla" -- that is, a bug that was reported to us
2027 but never got a bugzilla entry. We encourage you to file bugs in
2028 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2029 than mailing the developers (or mailing lists) directly -- bugs that
2030 are not entered into bugzilla tend to get forgotten about or ignored.
2032 To see details of a given bug, visit
2033 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2034 where XXXXXX is the bug number as listed below.
2036 116002 VG_(printf): Problems with justification of strings and integers
2037 155125 avoid cutting away file:lineno after long function name
2038 197259 Unsupported arch_prtctl PR_SET_GS option
2039 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
2040 201216 Fix Valgrind does not support pthread_sigmask() on OS X
2041 201435 Fix Darwin: -v does not show kernel version
2042 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
2043 211256 Fixed an outdated comment regarding the default platform.
2044 211529 Incomplete call stacks for code compiled by newer versions of MSVC
2045 211926 Avoid compilation warnings in valgrind.h with -pedantic
2046 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
2048 226609 Crediting upstream authors in man page
2049 231257 Valgrind omits path when executing script from shebang line
2050 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
2051 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
2052 269360 s390x: Fix addressing mode selection for compare-and-swap
2053 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
2055 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
2056 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
2057 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
2058 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
2059 330147 libmpiwrap PMPI_Get_count returns undefined value
2060 333051 mmap of huge pages fails due to incorrect alignment
2062 334802 valgrind does not always explain why a given option is bad
2063 335618 mov.w rN, pc/sp (ARM32)
2064 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
2069 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
2070 338602 AVX2 bit in CPUID missing
2071 338606 Strange message for scripts with invalid interpreter
2072 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
2073 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
2074 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
2076 339156 gdbsrv not called for fatal signal
2077 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
2078 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
2079 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
2080 339442 Fix testsuite build failure on OS X 10.9
2081 339542 Enable compilation with Intel's ICC compiler
2082 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
2083 339688 Mac-specific ASM does not support .version directive (cpuid,
2084 tronical and pushfpopf tests)
2085 339745 Valgrind crash when check Marmalade app (partial fix)
2086 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
2087 339778 Linux/TileGx platform support for Valgrind
2088 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
2089 339789 Fix none/tests/execve test on Mac OS X 10.9
2090 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
2091 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
2092 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
2093 340392 Allow user to select more accurate definedness checking in memcheck
2094 to avoid invalid complaints on optimised code
2095 340430 Fix some grammatical weirdness in the manual.
2096 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
2097 341419 Signal handler ucontext_t not filled out correctly on OS X
2098 341539 VG_(describe_addr) should not describe address as belonging to client
2099 segment if it is past the heap end
2100 341613 Enable building of manythreads and thread-exits tests on Mac OS X
2101 341615 Fix none/tests/darwin/access_extended test on Mac OS X
2102 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
2103 341789 aarch64: shmat fails with valgrind on ARMv8
2104 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
2105 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
2106 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
2107 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
2108 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
2109 342221 socket connect false positive uninit memory for unknown af family
2110 342353 Allow dumping full massif output while valgrind is still running
2111 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
2115 342603 Add I2C_SMBUS ioctl support
2116 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
2117 342683 Mark memory past the initial brk limit as unaddressable
2118 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
2119 342795 Internal glibc __GI_mempcpy call should be intercepted
2120 342841 s390x: Support instructions fiebr(a) and fidbr(a)
2121 343012 Unhandled syscall 319 (memfd_create)
2122 343069 Patch updating v4l2 API support
2123 343173 helgrind crash during stack unwind
2124 343219 fix GET_STARTREGS for arm
2125 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
2126 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
2127 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
2128 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
2129 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
2130 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
2131 343597 ppc64le: incorrect use of offseof macro
2132 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
2133 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
2134 343732 Unhandled syscall 144 (setgid) on aarch64
2135 343733 Unhandled syscall 187 (msgctl and related) on aarch64
2136 343802 s390x: False positive "conditional jump or move depends on [..]
2137 343902 --vgdb=yes doesn't break when --xml=yes is used
2138 343967 Don't warn about setuid/setgid/setcap executable for directories
2139 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
2140 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
2141 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
2142 344054 www - update info for Solaris/illumos
2143 344416 'make regtest' does not work cleanly on OS X
2144 344235 Remove duplicate include of pub_core_aspacemgr.h
2145 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
2146 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
2147 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
2148 344314 callgrind_annotate ... warnings about commands containing newlines
2149 344318 socketcall should wrap recvmmsg and sendmmsg
2150 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
2151 344416 Fix 'make regtest' does not work cleanly on OS X
2152 344499 Fix compilation for Linux kernel >= 4.0.0
2153 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
2154 unix:349 (__pthread_fchdir)
2155 344559 Garbage collection of unused segment names in address space manager
2156 344560 Fix stack traces missing penultimate frame on OS X
2157 344621 Fix memcheck/tests/err_disable4 test on OS X
2158 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
2159 344702 Fix missing libobjc suppressions on OS X 10.10
2161 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
2162 344939 Fix memcheck/tests/xml1 on OS X 10.10
2163 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
2164 345079 Fix build problems in VEX/useful/test_main.c
2165 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
2166 345177 arm64: prfm (reg) not implemented
2167 345215 Performance improvements for the register allocator
2168 345248 add support for Solaris OS in valgrind
2169 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
2170 345394 Fix memcheck/tests/strchr on OS X
2171 345637 Fix memcheck/tests/sendmsg on OS X
2172 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
2173 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
2174 345887 Fix an assertion in the address space manager
2175 345928 amd64: callstack only contains current function for small stacks
2176 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
2177 345987 MIPS64: Implement cavium LHX instruction
2178 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
2179 346185 Fix typo saving altivec register v24
2180 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
2181 and LibVEX_GuestPPC64_get_CR()
2182 346270 Regression tests none/tests/jm_vec/isa_2_07 and
2183 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
2184 346307 fuse filesystem syscall deadlocks
2185 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
2186 346411 MIPS: SysRes::_valEx handling is incorrect
2187 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
2188 346474 PPC64 Power 8, spr TEXASRU register not supported
2189 346487 Compiler generates "note" about a future ABI change for PPC64
2190 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
2191 and causing spurious "invalid read of size 8" warnings
2192 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
2193 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
2194 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
2195 347322 Power PC regression test cleanup
2196 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
2198 347389 unhandled syscall: 373 (Linux ARM syncfs)
2199 347686 Patch set to cleanup PPC64 regtests
2200 347978 Remove bash dependencies where not needed
2201 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
2202 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
2204 348102 Patch updating v4l2 API support
2205 348247 amd64 front end: jno jumps wrongly when overflow is not set
2206 348269 Improve mmap MAP_HUGETLB support.
2207 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
2208 348345 Assertion fails for negative lineno
2209 348377 Unsupported ARM instruction: yield
2210 348565 Fix detection of command line option availability for clang
2211 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
2212 348728 Fix broken check for VIDIOC_G_ENC_INDEX
2213 348748 Fix redundant condition
2214 348890 Fix clang warning about unsupported --param inline-unit-growth=900
2215 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
2216 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
2217 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
2218 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
2219 349626 Implemented additional Xen hypercalls
2220 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
2221 349790 Clean up of the hardware capability checking utilities.
2222 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
2223 349874 Fix typos in source code
2224 349879 memcheck: add handwritten assembly for helperc_LOADV*
2225 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
2226 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
2227 350202 Add limited param to 'monitor block_list'
2228 350290 s390x: Support instructions fixbr(a)
2229 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
2230 350809 Fix none/tests/async-sigs for Solaris
2231 350811 Remove reference to --db-attach which has been removed.
2232 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
2233 350854 hard-to-understand code in VG_(load_ELF)()
2234 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2235 351386 Solaris: Cannot run ld.so.1 under Valgrind
2236 351474 Fix VG_(iseqsigset) as obvious
2237 351531 Typo in /include/vki/vki-xen-physdev.h header guard
2238 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
2239 351858 ldsoexec support on Solaris
2240 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
2241 352130 helgrind reports false races for printfs using mempcpy on FILE* state
2242 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
2243 352320 arm64 crash on none/tests/nestedfs
2244 352765 Vbit test fails on Power 6
2245 352768 The mbar instruction is missing from the Power PC support
2246 352769 Power PC program priority register (PPR) is not supported
2247 n-i-bz Provide implementations of certain compiler builtins to support
2248 compilers that may not provide those
2249 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
2250 n-i-bz Fix compilation on distros with glibc < 2.5
2251 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
2252 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
2253 n-i-bz mremap did not work properly on shared memory
2254 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
2255 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
2257 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
2258 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
2259 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
2263 Release 3.10.1 (25 November 2014)
2264 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2265 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
2266 and backports fixes for all reported missing AArch64 ARMv8 instructions
2267 and syscalls from the trunk. If you package or deliver 3.10.0 for others
2268 to use, you might want to consider upgrading to 3.10.1 instead.
2270 The following bugs have been fixed or resolved. Note that "n-i-bz"
2271 stands for "not in bugzilla" -- that is, a bug that was reported to us
2272 but never got a bugzilla entry. We encourage you to file bugs in
2273 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2274 than mailing the developers (or mailing lists) directly -- bugs that
2275 are not entered into bugzilla tend to get forgotten about or ignored.
2277 To see details of a given bug, visit
2278 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2279 where XXXXXX is the bug number as listed below.
2281 335440 arm64: ld1 (single structure) is not implemented
2282 335713 arm64: unhanded instruction: prfm (immediate)
2283 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
2284 339182 ppc64: AvSplat ought to load destination vector register with [..]
2285 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
2286 339433 ppc64 lxvw4x instruction uses four 32-byte loads
2287 339645 Use correct tag names in sys_getdents/64 wrappers
2288 339706 Fix false positive for ioctl(TIOCSIG) on linux
2289 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
2290 339853 arm64 times syscall unknown
2291 339855 arm64 unhandled getsid/setsid syscalls
2292 339858 arm64 dmb sy not implemented
2293 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
2294 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
2295 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
2297 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
2298 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
2299 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
2300 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
2301 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
2302 340509 arm64: unhandled instruction fcvtas
2303 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
2304 340632 arm64: unhandled instruction fcvtas
2305 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
2306 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
2307 340788 warning: unhandled syscall: 318 (getrandom)
2308 340807 disInstr(arm): unhandled instruction: 0xEE989B20
2309 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
2310 340922 arm64: unhandled getgroups/setgroups syscalls
2311 350251 Fix typo in VEX utility program (test_main.c).
2312 350407 arm64: unhandled instruction ucvtf (vector, integer)
2313 350809 none/tests/async-sigs breaks when run under cron on Solaris
2314 350811 update README.solaris after r15445
2315 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
2316 350854 strange code in VG_(load_ELF)()
2317 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2318 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
2319 n-i-bz Add missing ]] to terminate CDATA.
2320 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
2321 n-i-bz Enable sys_fadvise64_64 on arm32.
2322 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
2323 n-i-bz Add test cases for all known arm64 load/store instructions.
2324 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
2325 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
2326 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
2327 n-i-bz arm64: Implement "fcvtpu w, s".
2328 n-i-bz arm64: implement ADDP and various others
2329 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
2330 n-i-bz arm64: enable FCVT{A,N}S X,S.
2332 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
2336 Release 3.10.0 (10 September 2014)
2337 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2339 3.10.0 is a feature release with many improvements and the usual
2340 collection of bug fixes.
2342 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
2343 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
2344 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
2345 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
2346 significantly improved relative to the 3.9.0 release.
2348 * ================== PLATFORM CHANGES =================
2350 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
2351 is mostly complete, and is usable, but some SIMD instructions are as
2354 * Support for little-endian variant of the 64-bit POWER architecture.
2356 * Support for Android on MIPS32.
2358 * Support for 64bit FPU on MIPS32 platforms.
2360 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
2362 * Configuration for and running on Android targets has changed.
2363 See README.android in the source tree for details.
2365 * ================== DEPRECATED FEATURES =================
2367 * --db-attach is now deprecated and will be removed in the next
2368 valgrind feature release. The built-in GDB server capabilities are
2369 superior and should be used instead. Learn more here:
2370 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
2372 * ==================== TOOL CHANGES ====================
2376 - Client code can now selectively disable and re-enable reporting of
2377 invalid address errors in specific ranges using the new client
2378 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
2379 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
2381 - Leak checker: there is a new leak check heuristic called
2382 "length64". This is used to detect interior pointers pointing 8
2383 bytes inside a block, on the assumption that the first 8 bytes
2384 holds the value "block size - 8". This is used by
2385 sqlite3MemMalloc, for example.
2387 - Checking of system call parameters: if a syscall parameter
2388 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
2389 several fields not initialised, an error is now reported for each
2390 field. Previously, an error was reported only for the first
2391 uninitialised field.
2393 - Mismatched alloc/free checking: a new flag
2394 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
2395 such checks if necessary.
2399 - Improvements to error messages:
2401 o Race condition error message involving heap allocated blocks also
2402 show the thread number that allocated the raced-on block.
2404 o All locks referenced by an error message are now announced.
2405 Previously, some error messages only showed the lock addresses.
2407 o The message indicating where a lock was first observed now also
2408 describes the address/location of the lock.
2410 - Helgrind now understands the Ada task termination rules and
2411 creates a happens-before relationship between a terminated task
2412 and its master. This avoids some false positives and avoids a big
2413 memory leak when a lot of Ada tasks are created and terminated.
2414 The interceptions are only activated with forthcoming releases of
2415 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
2417 - A new GDB server monitor command "info locks" giving the list of
2418 locks, their location, and their status.
2422 - callgrind_control now supports the --vgdb-prefix argument,
2423 which is needed if valgrind was started with this same argument.
2425 * ==================== OTHER CHANGES ====================
2427 * Unwinding through inlined function calls. Stack unwinding can now
2428 make use of Dwarf3 inlined-unwind information if it is available.
2429 The practical effect is that inlined calls become visible in stack
2430 traces. The suppression matching machinery has been adjusted
2431 accordingly. This is controlled by the new option
2432 --read-inline-info=yes|no. Currently this is enabled by default
2433 only on Linux and Android targets and only for the tools Memcheck,
2436 * Valgrind can now read EXIDX unwind information on 32-bit ARM
2437 targets. If an object contains both CFI and EXIDX unwind
2438 information, Valgrind will prefer the CFI over the EXIDX. This
2439 facilitates unwinding through system libraries on arm-android
2442 * Address description logic has been improved and is now common
2443 between Memcheck and Helgrind, resulting in better address
2444 descriptions for some kinds of error messages.
2446 * Error messages about dubious arguments (eg, to malloc or calloc) are
2447 output like other errors. This means that they can be suppressed
2448 and they have a stack trace.
2450 * The C++ demangler has been updated for better C++11 support.
2452 * New and modified GDB server monitor features:
2454 - Thread local variables/storage (__thread) can now be displayed.
2456 - The GDB server monitor command "v.info location <address>"
2457 displays information about an address. The information produced
2458 depends on the tool and on the options given to valgrind.
2459 Possibly, the following are described: global variables, local
2460 (stack) variables, allocated or freed blocks, ...
2462 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
2463 ask the GDB server to stop at the start of program execution, at
2464 the end of the program execution and on Valgrind internal errors.
2466 - A new monitor command "v.info stats" shows various Valgrind core
2467 and tool statistics.
2469 - A new monitor command "v.set hostvisibility" allows the GDB server
2470 to provide access to Valgrind internal host status/memory.
2472 * A new option "--aspace-minaddr=<address>" can in some situations
2473 allow the use of more memory by decreasing the address above which
2474 Valgrind maps memory. It can also be used to solve address
2475 conflicts with system libraries by increasing the default value.
2476 See user manual for details.
2478 * The amount of memory used by Valgrind to store debug info (unwind
2479 info, line number information and symbol data) has been
2480 significantly reduced, even though Valgrind now reads more
2481 information in order to support unwinding of inlined function calls.
2483 * Dwarf3 handling with --read-var-info=yes has been improved:
2485 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
2487 - Code compiled with
2488 -ffunction-sections -fdata-sections -Wl,--gc-sections
2489 no longer causes assertion failures.
2491 * Improved checking for the --sim-hints= and --kernel-variant=
2492 options. Unknown strings are now detected and reported to the user
2495 * The semantics of stack start/end boundaries in the valgrind.h
2496 VALGRIND_STACK_REGISTER client request has been clarified and
2497 documented. The convention is that start and end are respectively
2498 the lowest and highest addressable bytes of the stack.
2500 * ==================== FIXED BUGS ====================
2502 The following bugs have been fixed or resolved. Note that "n-i-bz"
2503 stands for "not in bugzilla" -- that is, a bug that was reported to us
2504 but never got a bugzilla entry. We encourage you to file bugs in
2505 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2506 than mailing the developers (or mailing lists) directly -- bugs that
2507 are not entered into bugzilla tend to get forgotten about or ignored.
2509 To see details of a given bug, visit
2510 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2511 where XXXXXX is the bug number as listed below.
2513 175819 Support for ipv6 socket reporting with --track-fds
2514 232510 make distcheck fails
2515 249435 Analyzing wine programs with callgrind triggers a crash
2516 278972 support for inlined function calls in stacktraces and suppression
2518 291310 FXSAVE instruction marks memory as undefined on amd64
2519 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2520 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2521 315199 vgcore file for threaded app does not show which thread crashed
2522 315952 tun/tap ioctls are not supported
2523 323178 Unhandled instruction: PLDW register (ARM)
2524 323179 Unhandled instruction: PLDW immediate (ARM)
2525 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2526 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2527 325124 [MIPSEL] Compilation error
2528 325477 Phase 4 support for IBM Power ISA 2.07
2529 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2530 325628 Phase 5 support for IBM Power ISA 2.07
2531 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2532 325751 Missing the two privileged Power PC Transactional Memory Instructions
2533 325816 Phase 6 support for IBM Power ISA 2.07
2534 325856 Make SGCheck fail gracefully on unsupported platforms
2535 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2536 326436 DRD: False positive in libstdc++ std::list::push_back
2537 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2538 326462 Refactor vgdb to isolate invoker stuff into separate module
2539 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2540 326623 DRD: false positive conflict report in a field assignment
2541 326724 Valgrind does not compile on OSX 1.9 Mavericks
2542 326816 Intercept for __strncpy_sse2_unaligned missing?
2543 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2544 326983 Clear direction flag after tests on amd64.
2545 327212 Do not prepend the current directory to absolute path names.
2546 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2547 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2548 327284 s390x: Fix translation of the risbg instruction
2549 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2550 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2551 327916 DW_TAG_typedef may have no name
2552 327943 s390x: add a redirection for the 'index' function
2553 328100 XABORT not implemented
2554 328205 Implement additional Xen hypercalls
2555 328454 add support Backtraces with ARM unwind tables (EXIDX)
2556 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2557 328711 valgrind.1 manpage "memcheck options" section is badly generated
2558 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2559 329612 Incorrect handling of AT_BASE for image execution
2560 329694 clang warns about using uninitialized variable
2561 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2562 330228 mmap must align to VKI_SHMLBA on mips32
2563 330257 LLVM does not support `-mno-dynamic-no-pic` option
2564 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2565 330459 --track-fds=yes doesn't track eventfds
2566 330469 Add clock_adjtime syscall support
2567 330594 Missing sysalls on PowerPC / uClibc
2568 330622 Add test to regression suite for POWER instruction: dcbzl
2569 330939 Support for AMD's syscall instruction on x86
2571 330941 Typo in PRE(poll) syscall wrapper
2572 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2573 331254 Fix expected output for memcheck/tests/dw4
2574 331255 Fix race condition in test none/tests/coolo_sigaction
2575 331257 Fix type of jump buffer in test none/tests/faultstatus
2576 331305 configure uses bash specific syntax
2577 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2578 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2579 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2580 331829 Unexpected ioctl opcode sign extension
2581 331830 ppc64: WARNING: unhandled syscall: 96/97
2582 331839 drd/tests/sem_open specifies invalid semaphore name
2583 331847 outcome of drd/tests/thread_name is nondeterministic
2584 332037 Valgrind cannot handle Thumb "add pc, reg"
2585 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2586 consistency checks enabled
2587 332263 intercepts for pthread_rwlock_timedrdlock and
2588 pthread_rwlock_timedwrlock are incorrect
2589 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2591 332276 Implement additional Xen hypercalls
2592 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2593 332765 Fix ms_print to create temporary files in a proper directory
2594 333072 drd: Add semaphore annotations
2595 333145 Tests for missaligned PC+#imm access for arm
2596 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2597 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2598 333248 WARNING: unhandled syscall: unix:443
2599 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2600 333501 cachegrind: assertion: Cache set count is not a power of two.
2603 333666 Recognize MPX instructions and bnd prefix.
2604 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2605 333817 Valgrind reports the memory areas written to by the SG_IO
2607 334049 lzcnt fails silently (x86_32)
2608 334384 Valgrind does not have support Little Endian support for
2610 334585 recvmmsg unhandled (+patch) (arm)
2611 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2612 334727 Build fails with -Werror=format-security
2613 334788 clarify doc about --log-file initial program directory
2614 334834 PPC64 Little Endian support, patch 2
2615 334836 PPC64 Little Endian support, patch 3 testcase fixes
2616 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2617 335034 Unhandled ioctl: HCIGETDEVLIST
2618 335155 vgdb, fix error print statement.
2619 335262 arm64: movi 8bit version is not supported
2620 335263 arm64: dmb instruction is not implemented
2621 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2622 335496 arm64: sbc/abc instructions are not implemented
2623 335554 arm64: unhandled instruction: abs
2624 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2625 335735 arm64: unhandled instruction: cnt
2626 335736 arm64: unhandled instruction: uaddlv
2627 335848 arm64: unhandled instruction: {s,u}cvtf
2628 335902 arm64: unhandled instruction: sli
2629 335903 arm64: unhandled instruction: umull (vector)
2630 336055 arm64: unhandled instruction: mov (element)
2631 336062 arm64: unhandled instruction: shrn{,2}
2632 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2633 336189 arm64: unhandled Instruction: mvn
2634 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2635 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2636 336772 Make moans about unknown ioctls more informative
2637 336957 Add a section about the Solaris/illumos port on the webpage
2638 337094 ifunc wrapper is broken on ppc64
2639 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2640 337528 leak check heuristic for block prefixed by length as 64bit number
2641 337740 Implement additional Xen hypercalls
2642 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2643 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2644 337871 deprecate --db-attach
2645 338023 Add support for all V4L2/media ioctls
2646 338024 inlined functions are not shown if DW_AT_ranges is used
2647 338106 Add support for 'kcmp' syscall
2648 338115 DRD: computed conflict set differs from actual after fork
2649 338160 implement display of thread local storage in gdbsrv
2650 338205 configure.ac and check for -Wno-tautological-compare
2651 338300 coredumps are missing one byte of every segment
2652 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2653 338499 --sim-hints parsing broken due to wrong order in tokens
2654 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2655 338681 Unable to unwind through clone thread created on i386-linux
2656 338698 race condition between gdbsrv and vgdb on startup
2657 338703 helgrind on arm-linux gets false positives in dynamic loader
2658 338791 alt dwz files can be relative of debug/main file
2659 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2660 338932 build V-trunk with gcc-trunk
2661 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2662 345079 Fix build problems in VEX/useful/test_main.c
2663 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2664 n-i-bz s390x: Fix memory corruption for multithreaded applications
2665 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2666 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2667 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2668 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2669 n-i-bz Add command-line option --default-suppressions
2670 n-i-bz Add support for BLKDISCARDZEROES ioctl
2671 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2672 n-i-bz Add support for sys_pivot_root and sys_unshare
2674 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2675 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2676 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2680 Release 3.9.0 (31 October 2013)
2681 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2682 3.9.0 is a feature release with many improvements and the usual
2683 collection of bug fixes.
2685 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2686 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2687 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2688 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2690 * ================== PLATFORM CHANGES =================
2692 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2693 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2695 * Support for MIPS DSP ASE on MIPS32 platforms.
2697 * Support for s390x Decimal Floating Point instructions on hosts that
2698 have the DFP facility installed.
2700 * Support for POWER8 (Power ISA 2.07) instructions
2702 * Support for Intel AVX2 instructions. This is available only on 64
2705 * Initial support for Intel Transactional Synchronization Extensions,
2708 * Initial support for Hardware Transactional Memory on POWER.
2710 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2711 run large GUI apps tolerably well.
2713 * ==================== TOOL CHANGES ====================
2717 - Improvements in handling of vectorised code, leading to
2718 significantly fewer false error reports. You need to use the flag
2719 --partial-loads-ok=yes to get the benefits of these changes.
2721 - Better control over the leak checker. It is now possible to
2722 specify which leak kinds (definite/indirect/possible/reachable)
2723 should be displayed, which should be regarded as errors, and which
2724 should be suppressed by a given leak suppression. This is done
2725 using the options --show-leak-kinds=kind1,kind2,..,
2726 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2727 "match-leak-kinds:" line in suppression entries, respectively.
2729 Note that generated leak suppressions contain this new line and
2730 are therefore more specific than in previous releases. To get the
2731 same behaviour as previous releases, remove the "match-leak-kinds:"
2732 line from generated suppressions before using them.
2734 - Reduced "possible leak" reports from the leak checker by the use
2735 of better heuristics. The available heuristics provide detection
2736 of valid interior pointers to std::stdstring, to new[] allocated
2737 arrays with elements having destructors and to interior pointers
2738 pointing to an inner part of a C++ object using multiple
2739 inheritance. They can be selected individually using the
2740 option --leak-check-heuristics=heur1,heur2,...
2742 - Better control of stacktrace acquisition for heap-allocated
2743 blocks. Using the --keep-stacktraces option, it is possible to
2744 control independently whether a stack trace is acquired for each
2745 allocation and deallocation. This can be used to create better
2746 "use after free" errors or to decrease Valgrind's resource
2747 consumption by recording less information.
2749 - Better reporting of leak suppression usage. The list of used
2750 suppressions (shown when the -v option is given) now shows, for
2751 each leak suppressions, how many blocks and bytes it suppressed
2752 during the last leak search.
2756 - False errors resulting from the use of statically initialised
2757 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2760 - False errors resulting from the use of pthread_cond_waits that
2761 timeout, have been removed.
2763 * ==================== OTHER CHANGES ====================
2765 * Some attempt to tune Valgrind's space requirements to the expected
2766 capabilities of the target:
2768 - The default size of the translation cache has been reduced from 8
2769 sectors to 6 on Android platforms, since each sector occupies
2770 about 40MB when using Memcheck.
2772 - The default size of the translation cache has been increased to 16
2773 sectors on all other platforms, reflecting the fact that large
2774 applications require instrumentation and storage of huge amounts
2775 of code. For similar reasons, the number of memory mapped
2776 segments that can be tracked has been increased by a factor of 6.
2778 - In all cases, the maximum number of sectors in the translation
2779 cache can be controlled by the new flag --num-transtab-sectors.
2781 * Changes in how debug info (line numbers, etc) is read:
2783 - Valgrind no longer temporarily mmaps the entire object to read
2784 from it. Instead, reading is done through a small fixed sized
2785 buffer. This avoids virtual memory usage spikes when Valgrind
2786 reads debuginfo from large shared objects.
2788 - A new experimental remote debug info server. Valgrind can read
2789 debug info from a different machine (typically, a build host)
2790 where debuginfo objects are stored. This can save a lot of time
2791 and hassle when running Valgrind on resource-constrained targets
2792 (phones, tablets) when the full debuginfo objects are stored
2793 somewhere else. This is enabled by the --debuginfo-server=
2796 - Consistency checking between main and debug objects can be
2797 disabled using the --allow-mismatched-debuginfo option.
2799 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2800 scanning can recover stack traces in some cases when the normal
2801 unwind mechanisms fail. Stack scanning is best described as "a
2802 nasty, dangerous and misleading hack" and so is disabled by default.
2803 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2806 * Detection and merging of recursive stack frame cycles. When your
2807 program has recursive algorithms, this limits the memory used by
2808 Valgrind for recorded stack traces and avoids recording
2809 uninteresting repeated calls. This is controlled by the command
2810 line option --merge-recursive-frame and by the monitor command
2811 "v.set merge-recursive-frames".
2813 * File name and line numbers for used suppressions. The list of used
2814 suppressions (shown when the -v option is given) now shows, for each
2815 used suppression, the file name and line number where the suppression
2818 * New and modified GDB server monitor features:
2820 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2821 that can be used to execute gdbserver monitor commands from the
2824 - A new monitor command, "v.info open_fds", that gives the list of
2825 open file descriptors and additional details.
2827 - An optional message in the "v.info n_errs_found" monitor command,
2828 for example "v.info n_errs_found test 1234 finished", allowing a
2829 comment string to be added to the process output, perhaps for the
2830 purpose of separating errors of different tests or test phases.
2832 - A new monitor command "v.info execontext" that shows information
2833 about the stack traces recorded by Valgrind.
2835 - A new monitor command "v.do expensive_sanity_check_general" to run
2836 some internal consistency checks.
2838 * New flag --sigill-diagnostics to control whether a diagnostic
2839 message is printed when the JIT encounters an instruction it can't
2840 translate. The actual behavior -- delivery of SIGILL to the
2841 application -- is unchanged.
2843 * The maximum amount of memory that Valgrind can use on 64 bit targets
2844 has been increased from 32GB to 64GB. This should make it possible
2845 to run applications on Memcheck that natively require up to about 35GB.
2847 * ==================== FIXED BUGS ====================
2849 The following bugs have been fixed or resolved. Note that "n-i-bz"
2850 stands for "not in bugzilla" -- that is, a bug that was reported to us
2851 but never got a bugzilla entry. We encourage you to file bugs in
2852 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2853 than mailing the developers (or mailing lists) directly -- bugs that
2854 are not entered into bugzilla tend to get forgotten about or ignored.
2856 To see details of a given bug, visit
2857 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2858 where XXXXXX is the bug number as listed below.
2860 123837 system call: 4th argument is optional, depending on cmd
2861 135425 memcheck should tell you where Freed blocks were Mallocd
2862 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2863 207815 Adds some of the drm ioctls to syswrap-linux.c
2864 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2865 252955 Impossible to compile with ccache
2866 253519 Memcheck reports auxv pointer accesses as invalid reads.
2867 263034 Crash when loading some PPC64 binaries
2868 269599 Increase deepest backtrace
2869 274695 s390x: Support "compare to/from logical" instructions (z196)
2870 275800 s390x: Autodetect cache info (part 2)
2871 280271 Valgrind reports possible memory leaks on still-reachable std::string
2872 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2873 289578 Backtraces with ARM unwind tables (stack scan flags)
2874 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2875 304832 ppc32: build failure
2876 305431 Use find_buildid shdr fallback for separate .debug files
2877 305728 Add support for AVX2 instructions
2878 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2879 306035 s390x: Fix IR generation for LAAG and friends
2880 306054 s390x: Condition code computation for convert-to-int/logical
2881 306098 s390x: alternate opcode form for convert to/from fixed
2882 306587 Fix cache line detection from auxiliary vector for PPC.
2883 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2884 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2885 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2886 307101 sys_capget second argument can be NULL
2887 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2888 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2889 307113 s390x: DFP support
2890 307141 valgrind does't work in mips-linux system
2891 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2892 307285 x86_amd64 feature test for avx in test suite is wrong
2893 307290 memcheck overlap testcase needs memcpy version filter
2894 307463 Please add "&limit=0" to the "all open bugs" link
2895 307465 --show-possibly-lost=no should reduce the error count / exit code
2896 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2897 307729 pkgconfig support broken valgrind.pc
2898 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2899 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2900 308089 Unhandled syscall on ppc64: prctl
2901 308135 PPC32 MPC8xx has 16 bytes cache size
2902 308321 testsuite memcheck filter interferes with gdb_filter
2904 308341 vgdb should report process exit (or fatal signal)
2905 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2906 308495 Remove build dependency on installed Xen headers
2907 308573 Internal error on 64-bit instruction executed in 32-bit mode
2909 308627 pmovmskb validity bit propagation is imprecise
2910 308644 vgdb command for having the info for the track-fds option
2911 308711 give more info about aspacemgr and arenas in out_of_memory
2912 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2913 308718 ARM implement SMLALBB family of instructions
2914 308886 Missing support for PTRACE_SET/GETREGSET
2915 308930 syscall name_to_handle_at (303 on amd64) not handled
2916 309229 V-bit tester does not report number of tests generated
2917 309323 print unrecognized instuction on MIPS
2918 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2919 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2920 309430 Self hosting ppc64 encounters a vassert error on operand type
2921 309600 valgrind is a bit confused about 0-sized sections
2922 309823 Generate errors for still reachable blocks
2923 309921 PCMPISTRI validity bit propagation is imprecise
2924 309922 none/tests/ppc64/test_dfp5 sometimes fails
2925 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2926 310424 --read-var-info does not properly describe static variables
2927 310792 search additional path for debug symbols
2928 310931 s390x: Message-security assist (MSA) instruction extension [..]
2929 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2930 311318 ARM: "128-bit constant is not implemented" error message
2931 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2932 311690 V crashes because it redirects branches inside of a redirected function
2933 311880 x86_64: make regtest hangs at shell_valid1
2934 311922 WARNING: unhandled syscall: 170
2936 312171 ppc: insn selection for DFP
2937 312571 Rounding mode call wrong for the DFP Iops [..]
2938 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2939 312913 Dangling pointers error should also report the alloc stack trace
2940 312980 Building on Mountain Lion generates some compiler warnings
2941 313267 Adding MIPS64/Linux port to Valgrind
2944 313811 Buffer overflow in assert_fail
2945 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2946 314269 ppc: dead code in insn selection
2947 314718 ARM: implement integer divide instruction (sdiv and udiv)
2948 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2949 315441 sendmsg syscall should ignore unset msghdr msg_flags
2950 315534 msgrcv inside a thread causes valgrind to hang (block)
2951 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2952 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2953 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2954 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2955 316144 valgrind.1 manpage contains unknown ??? strings [..]
2956 316145 callgrind command line options in manpage reference (unknown) [..]
2957 316145 callgrind command line options in manpage reference [..]
2958 316181 drd: Fixed a 4x slowdown for certain applications
2959 316503 Valgrind does not support SSE4 "movntdqa" instruction
2960 316535 Use of |signed int| instead of |size_t| in valgrind messages
2961 316696 fluidanimate program of parsec 2.1 stuck
2962 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2963 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2964 317186 "Impossible happens" when occurs VCVT instruction on ARM
2965 317318 Support for Threading Building Blocks "scalable_malloc"
2966 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2967 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2968 317463 bmi testcase IR SANITY CHECK FAILURE
2969 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2970 318050 libmpiwrap fails to compile with out-of-source build
2971 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2972 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2973 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2974 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2975 318932 Add missing PPC64 and PPC32 system call support
2976 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2977 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2978 319494 VEX Makefile-gcc standalone build update after r2702
2979 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2980 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2981 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2982 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2983 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2984 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2985 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2986 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2987 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2988 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2989 320895 add fanotify support (patch included)
2990 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2991 321065 Valgrind updates for Xen 4.3
2992 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2993 321363 Unhandled instruction: SSAX (ARM + Thumb)
2994 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2995 321466 Unhandled instruction: SHASX (ARM + Thumb)
2996 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2997 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2998 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2999 321620 Unhandled instruction: UASX (ARM + Thumb)
3000 321621 Unhandled instruction: USAX (ARM + Thumb)
3001 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
3002 321693 Unhandled instruction: LDRSBT (Thumb)
3003 321694 Unhandled instruction: UQASX (ARM + Thumb)
3004 321696 Unhandled instruction: UQSAX (Thumb + ARM)
3005 321697 Unhandled instruction: UHASX (ARM + Thumb)
3006 321703 Unhandled instruction: UHSAX (ARM + Thumb)
3007 321704 Unhandled instruction: REVSH (ARM + Thumb)
3008 321730 Add cg_diff and cg_merge man pages
3009 321738 Add vgdb and valgrind-listener man pages
3011 321891 Unhandled instruction: LDRHT (Thumb)
3012 321960 pthread_create() then alloca() causing invalid stack write errors
3013 321969 ppc32 and ppc64 don't support [lf]setxattr
3014 322254 Show threadname together with tid if set by application
3015 322294 Add initial support for IBM Power ISA 2.07
3016 322368 Assertion failure in wqthread_hijack under OS X 10.8
3017 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
3018 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
3019 322851 0bXXX binary literal syntax is not standard
3020 323035 Unhandled instruction: LDRSHT(Thumb)
3021 323036 Unhandled instruction: SMMLS (ARM and Thumb)
3022 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
3023 323175 Unhandled instruction: SMLALD (ARM + Thumb)
3024 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
3025 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
3026 323437 Phase 2 support for IBM Power ISA 2.07
3027 323713 Support mmxext (integer sse) subset on i386 (athlon)
3028 323803 Transactional memory instructions are not supported for Power
3029 323893 SSE3 not available on amd cpus in valgrind
3030 323905 Probable false positive from Valgrind/drd on close()
3031 323912 valgrind.h header isn't compatible for mingw64
3032 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
3033 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
3034 324181 mmap does not handle MAP_32BIT
3035 324227 memcheck false positive leak when a thread calls exit+block [..]
3036 324421 Support for fanotify API on ARM architecture
3037 324514 gdbserver monitor cmd output behaviour consistency [..]
3038 324518 ppc64: Emulation of dcbt instructions does not handle [..]
3039 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
3040 324582 When access is made to freed memory, report both allocation [..]
3041 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
3042 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
3043 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
3044 324834 Unhandled instructions in Microsoft C run-time for x86_64
3045 324894 Phase 3 support for IBM Power ISA 2.07
3046 326091 drd: Avoid false race reports from optimized strlen() impls
3047 326113 valgrind libvex hwcaps error on AMD64
3048 n-i-bz Some wrong command line options could be ignored
3049 n-i-bz patch to allow fair-sched on android
3050 n-i-bz report error for vgdb snapshot requested before execution
3051 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
3053 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
3057 Release 3.8.1 (19 September 2012)
3058 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3059 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
3060 that occur moderately frequently in real use cases, adds support for
3061 some missing instructions on ARM, and fixes a deadlock condition on
3062 MacOSX. If you package or deliver 3.8.0 for others to use, you might
3063 want to consider upgrading to 3.8.1 instead.
3065 The following bugs have been fixed or resolved. Note that "n-i-bz"
3066 stands for "not in bugzilla" -- that is, a bug that was reported to us
3067 but never got a bugzilla entry. We encourage you to file bugs in
3068 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3069 than mailing the developers (or mailing lists) directly -- bugs that
3070 are not entered into bugzilla tend to get forgotten about or ignored.
3072 To see details of a given bug, visit
3073 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3074 where XXXXXX is the bug number as listed below.
3077 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
3078 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
3079 298281 wcslen causes false(?) uninitialised value warnings
3080 301281 valgrind hangs on OS X when the process calls system()
3081 304035 disInstr(arm): unhandled instruction 0xE1023053
3082 304867 implement MOVBE instruction in x86 mode
3083 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
3084 305042 amd64: implement 0F 7F encoding of movq between two registers
3085 305199 ARM: implement QDADD and QDSUB
3086 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
3087 305513 killed by fatal signal: SIGSEGV
3088 305690 DRD reporting invalid semaphore when sem_trywait fails
3089 305926 Invalid alignment checks for some AVX instructions
3090 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
3091 306310 3.8.0 release tarball missing some files
3092 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
3093 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
3094 n-i-bz shmat of a segment > 4Gb does not work
3095 n-i-bz simulate_control_c script wrong USR1 signal number on mips
3096 n-i-bz vgdb ptrace calls wrong on mips [...]
3097 n-i-bz Fixes for more MPI false positives
3098 n-i-bz exp-sgcheck's memcpy causes programs to segfault
3099 n-i-bz OSX build w/ clang: asserts at startup
3100 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
3101 n-i-bz fix a couple of union tag-vs-field mixups
3102 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
3104 The following bugs were fixed in 3.8.0 but not listed in this NEWS
3107 254088 Valgrind should know about UD2 instruction
3110 304754 NEWS blows TeX's little mind
3112 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
3116 Release 3.8.0 (10 August 2012)
3117 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3118 3.8.0 is a feature release with many improvements and the usual
3119 collection of bug fixes.
3121 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3122 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
3123 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
3124 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
3125 There is initial support for MacOSX 10.8, but it is not usable for
3126 serious work at present.
3128 * ================== PLATFORM CHANGES =================
3130 * Support for MIPS32 platforms running Linux. Valgrind has been
3131 tested on MIPS32 and MIPS32r2 platforms running different Debian
3132 Squeeze and MeeGo distributions. Both little-endian and big-endian
3133 cores are supported. The tools Memcheck, Massif and Lackey have
3134 been tested and are known to work. See README.mips for more details.
3136 * Preliminary support for Android running on x86.
3138 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
3140 * Support for Intel AVX instructions and for AES instructions. This
3141 support is available only for 64 bit code.
3143 * Support for POWER Decimal Floating Point instructions.
3145 * ==================== TOOL CHANGES ====================
3147 * Non-libc malloc implementations are now supported. This is useful
3148 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
3149 Using the new option --soname-synonyms, such tools can be informed
3150 that the malloc implementation is either linked statically into the
3151 executable, or is present in some other shared library different
3152 from libc.so. This makes it possible to process statically linked
3153 programs, and programs using other malloc libraries, for example
3154 TCMalloc or JEMalloc.
3156 * For tools that provide their own replacement for malloc et al, the
3157 option --redzone-size=<number> allows users to specify the size of
3158 the padding blocks (redzones) added before and after each client
3159 allocated block. Smaller redzones decrease the memory needed by
3160 Valgrind. Bigger redzones increase the chance to detect blocks
3161 overrun or underrun. Prior to this change, the redzone size was
3162 hardwired to 16 bytes in Memcheck.
3166 - The leak_check GDB server monitor command now can
3167 control the maximum nr of loss records to output.
3169 - Reduction of memory use for applications allocating
3170 many blocks and/or having many partially defined bytes.
3172 - Addition of GDB server monitor command 'block_list' that lists
3173 the addresses/sizes of the blocks of a leak search loss record.
3175 - Addition of GDB server monitor command 'who_points_at' that lists
3176 the locations pointing at a block.
3178 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
3179 detect an invalid access of these redzones, by marking them
3180 noaccess. Similarly, if a redzone size is given for a memory
3181 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
3182 This still allows to find some bugs if the user has forgotten to
3183 mark the pool superblock noaccess.
3185 - Performance of memory leak check has been improved, especially in
3186 cases where there are many leaked blocks and/or many suppression
3187 rules used to suppress leak reports.
3189 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
3190 more precise analysis, which is important for LLVM/Clang
3191 generated code. This is at the cost of somewhat reduced
3192 performance. Note there is no change to analysis precision or
3193 costs on Linux targets.
3197 - Added even more facilities that can help finding the cause of a data
3198 race, namely the command-line option --ptrace-addr and the macro
3199 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
3201 - Fixed a subtle bug that could cause false positive data race reports.
3203 * ==================== OTHER CHANGES ====================
3205 * The C++ demangler has been updated so as to work well with C++
3206 compiled by up to at least g++ 4.6.
3208 * Tool developers can make replacement/wrapping more flexible thanks
3209 to the new option --soname-synonyms. This was reported above, but
3210 in fact is very general and applies to all function
3211 replacement/wrapping, not just to malloc-family functions.
3213 * Round-robin scheduling of threads can be selected, using the new
3214 option --fair-sched= yes. Prior to this change, the pipe-based
3215 thread serialisation mechanism (which is still the default) could
3216 give very unfair scheduling. --fair-sched=yes improves
3217 responsiveness of interactive multithreaded applications, and
3218 improves repeatability of results from the thread checkers Helgrind
3221 * For tool developers: support to run Valgrind on Valgrind has been
3222 improved. We can now routinely Valgrind on Helgrind or Memcheck.
3224 * gdbserver now shows the float shadow registers as integer
3225 rather than float values, as the shadow values are mostly
3226 used as bit patterns.
3228 * Increased limit for the --num-callers command line flag to 500.
3230 * Performance improvements for error matching when there are many
3231 suppression records in use.
3233 * Improved support for DWARF4 debugging information (bug 284184).
3235 * Initial support for DWZ compressed Dwarf debug info.
3237 * Improved control over the IR optimiser's handling of the tradeoff
3238 between performance and precision of exceptions. Specifically,
3239 --vex-iropt-precise-memory-exns has been removed and replaced by
3240 --vex-iropt-register-updates, with extended functionality. This
3241 allows the Valgrind gdbserver to always show up to date register
3244 * Modest performance gains through the use of translation chaining for
3247 * ==================== FIXED BUGS ====================
3249 The following bugs have been fixed or resolved. Note that "n-i-bz"
3250 stands for "not in bugzilla" -- that is, a bug that was reported to us
3251 but never got a bugzilla entry. We encourage you to file bugs in
3252 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3253 than mailing the developers (or mailing lists) directly -- bugs that
3254 are not entered into bugzilla tend to get forgotten about or ignored.
3256 To see details of a given bug, visit
3257 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3258 where XXXXXX is the bug number as listed below.
3260 197914 Building valgrind from svn now requires automake-1.10
3261 203877 increase to 16Mb maximum allowed alignment for memalign et al
3262 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
3263 247386 make perf does not run all performance tests
3264 270006 Valgrind scheduler unfair
3265 270777 Adding MIPS/Linux port to Valgrind
3266 270796 s390x: Removed broken support for the TS insn
3267 271438 Fix configure for proper SSE4.2 detection
3268 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
3269 273475 Add support for AVX instructions
3270 274078 improved configure logic for mpicc
3271 276993 fix mremap 'no thrash checks'
3272 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
3273 281482 memcheck incorrect byte allocation count in realloc() for silly argument
3274 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
3275 283413 Fix wrong sanity check
3276 283671 Robustize alignment computation in LibVEX_Alloc
3277 283961 Adding support for some HCI IOCTLs
3278 284124 parse_type_DIE: confused by: DWARF 4
3279 284864 == 273475 (Add support for AVX instructions)
3280 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
3281 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
3282 285725 == 273475 (Add support for AVX instructions)
3283 286261 add wrapper for linux I2C_RDWR ioctl
3284 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
3285 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
3286 286384 configure fails "checking for a supported version of gcc"
3287 286497 == 273475 (Add support for AVX instructions)
3288 286596 == 273475 (Add support for AVX instructions)
3289 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
3290 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
3291 287260 Incorrect conditional jump or move depends on uninitialised value(s)
3292 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
3293 287307 == 273475 (Add support for AVX instructions)
3294 287858 VG_(strerror): unknown error
3295 288298 (MacOSX) unhandled syscall shm_unlink
3296 288995 == 273475 (Add support for AVX instructions)
3297 289470 Loading of large Mach-O thin binaries fails.
3298 289656 == 273475 (Add support for AVX instructions)
3299 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
3300 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
3301 289839 s390x: Provide support for unicode conversion instructions
3302 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
3303 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
3304 290655 Add support for AESKEYGENASSIST instruction
3305 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
3306 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
3307 291253 ES register not initialised in valgrind simulation
3308 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
3309 291865 s390x: Support the "Compare Double and Swap" family of instructions
3310 292300 == 273475 (Add support for AVX instructions)
3311 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
3312 292493 == 273475 (Add support for AVX instructions)
3313 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
3314 292627 Missing support for some SCSI ioctls
3315 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
3316 292841 == 273475 (Add support for AVX instructions)
3317 292993 implement the getcpu syscall on amd64-linux
3318 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
3319 293088 Add some VEX sanity checks for ppc64 unhandled instructions
3320 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
3321 293754 PCMPxSTRx not implemented for 16-bit characters
3322 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
3323 293808 CLFLUSH not supported by latest VEX for amd64
3324 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
3325 294048 MPSADBW instruction not implemented
3326 294055 regtest none/tests/shell fails when locale is not set to C
3327 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
3328 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
3329 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
3330 294260 disInstr_AMD64: disInstr miscalculated next %rip
3331 294523 --partial-loads-ok=yes causes false negatives
3332 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
3333 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
3334 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
3335 295089 can not annotate source for both helgrind and drd
3336 295221 POWER Processor decimal floating point instruction support missing
3337 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
3338 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
3339 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
3340 295617 ARM - Add some missing syscalls
3341 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
3342 296229 Linux user input device ioctls missing wrappers
3343 296318 ELF Debug info improvements (more than one rx/rw mapping)
3344 296422 Add translation chaining support
3345 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
3346 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
3347 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
3348 297078 gdbserver signal handling problems [..]
3349 297147 drd false positives on newly allocated memory
3350 297329 disallow decoding of IBM Power DFP insns on some machines
3351 297497 POWER Processor decimal floating point instruction support missing
3352 297701 Another alias for strncasecmp_l in libc-2.13.so
3353 297911 'invalid write' not reported when using APIs for custom mem allocators.
3354 297976 s390x: revisit EX implementation
3355 297991 Valgrind interferes with mmap()+ftell()
3356 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
3357 297993 Fix compilation of valgrind with gcc -g3.
3358 298080 POWER Processor DFP support missing, part 3
3359 298227 == 273475 (Add support for AVX instructions)
3360 298335 == 273475 (Add support for AVX instructions)
3361 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
3362 298394 s390x: Don't bail out on an unknown machine model. [..]
3363 298421 accept4() syscall (366) support is missing for ARM
3364 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
3365 298732 valgrind installation problem in ubuntu with kernel version 3.x
3366 298862 POWER Processor DFP instruction support missing, part 4
3367 298864 DWARF reader mis-parses DW_FORM_ref_addr
3368 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
3369 299053 Support DWARF4 DW_AT_high_pc constant form
3370 299104 == 273475 (Add support for AVX instructions)
3371 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
3372 299629 dup3() syscall (358) support is missing for ARM
3373 299694 POWER Processor DFP instruction support missing, part 5
3374 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
3375 299803 == 273475 (Add support for AVX instructions)
3376 299804 == 273475 (Add support for AVX instructions)
3377 299805 == 273475 (Add support for AVX instructions)
3378 300140 ARM - Missing (T1) SMMUL
3379 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
3380 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
3381 300414 FCOM and FCOMP unimplemented for amd64 guest
3382 301204 infinite loop in canonicaliseSymtab with ifunc symbol
3383 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
3384 301265 add x86 support to Android build
3385 301984 configure script doesn't detect certain versions of clang
3386 302205 Fix compiler warnings for POWER VEX code and POWER test cases
3387 302287 Unhandled movbe instruction on Atom processors
3388 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
3389 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
3390 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
3391 302656 == 273475 (Add support for AVX instructions)
3392 302709 valgrind for ARM needs extra tls support for android emulator [..]
3393 302827 add wrapper for CDROM_GET_CAPABILITY
3394 302901 Valgrind crashes with dwz optimized debuginfo
3395 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
3396 303116 Add support for the POWER instruction popcntb
3397 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
3398 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
3399 303466 == 273475 (Add support for AVX instructions)
3400 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
3401 303963 strstr() function produces wrong results under valgrind callgrind
3402 304054 CALL_FN_xx macros need to enforce stack alignment
3403 304561 tee system call not supported
3404 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
3405 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
3406 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
3407 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
3408 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
3409 n-i-bz s390x: Shadow registers can now be examined using vgdb
3411 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
3412 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
3416 Release 3.7.0 (5 November 2011)
3417 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3418 3.7.0 is a feature release with many significant improvements and the
3419 usual collection of bug fixes.
3421 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3422 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
3423 Support for recent distros and toolchain components (glibc 2.14, gcc
3424 4.6, MacOSX 10.7) has been added.
3426 * ================== PLATFORM CHANGES =================
3428 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
3429 analyse 64-bit programs running on z/Architecture. Most user space
3430 instructions up to and including z10 are supported. Valgrind has
3431 been tested extensively on z9, z10, and z196 machines running SLES
3432 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
3433 known to work well. Callgrind, Helgrind, and DRD work reasonably
3434 well on z9 and later models. See README.s390 for more details.
3436 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
3437 64-bit processes are supported. Some complex threaded applications
3438 (Firefox) are observed to hang when run as 32 bit applications,
3439 whereas 64-bit versions run OK. The cause is unknown. Memcheck
3440 will likely report some false errors. In general, expect some rough
3441 spots. This release also supports MacOSX 10.6, but drops support
3444 * Preliminary support for Android (on ARM). Valgrind can now run
3445 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
3446 README.android for more details, plus instructions on how to get
3449 * Support for the IBM Power ISA 2.06 (Power7 instructions)
3451 * General correctness and performance improvements for ARM/Linux, and,
3452 by extension, ARM/Android.
3454 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
3455 instruction set support is under development but is not available in
3458 * Support for AIX5 has been removed.
3460 * ==================== TOOL CHANGES ====================
3462 * Memcheck: some incremental changes:
3464 - reduction of memory use in some circumstances
3466 - improved handling of freed memory, which in some circumstances
3467 can cause detection of use-after-free that would previously have
3470 - fix of a longstanding bug that could cause false negatives (missed
3471 errors) in programs doing vector saturated narrowing instructions.
3473 * Helgrind: performance improvements and major memory use reductions,
3474 particularly for large, long running applications which perform many
3475 synchronisation (lock, unlock, etc) events. Plus many smaller
3478 - display of locksets for both threads involved in a race
3480 - general improvements in formatting/clarity of error messages
3482 - addition of facilities and documentation regarding annotation
3483 of thread safe reference counted C++ classes
3485 - new flag --check-stack-refs=no|yes [yes], to disable race checking
3486 on thread stacks (a performance hack)
3488 - new flag --free-is-write=no|yes [no], to enable detection of races
3489 where one thread accesses heap memory but another one frees it,
3490 without any coordinating synchronisation event
3492 * DRD: enabled XML output; added support for delayed thread deletion
3493 in order to detect races that occur close to the end of a thread
3494 (--join-list-vol); fixed a memory leak triggered by repeated client
3495 memory allocatation and deallocation; improved Darwin support.
3497 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
3499 * exp-sgcheck: this tool has been reduced in scope so as to improve
3500 performance and remove checking that Memcheck does better.
3501 Specifically, the ability to check for overruns for stack and global
3502 arrays is unchanged, but the ability to check for overruns of heap
3503 blocks has been removed. The tool has accordingly been renamed to
3504 exp-sgcheck ("Stack and Global Array Checking").
3506 * ==================== OTHER CHANGES ====================
3508 * GDB server: Valgrind now has an embedded GDB server. That means it
3509 is possible to control a Valgrind run from GDB, doing all the usual
3510 things that GDB can do (single stepping, breakpoints, examining
3511 data, etc). Tool-specific functionality is also available. For
3512 example, it is possible to query the definedness state of variables
3513 or memory from within GDB when running Memcheck; arbitrarily large
3514 memory watchpoints are supported, etc. To use the GDB server, start
3515 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3518 * Improved support for unfriendly self-modifying code: a new option
3519 --smc-check=all-non-file is available. This adds the relevant
3520 consistency checks only to code that originates in non-file-backed
3521 mappings. In effect this confines the consistency checking only to
3522 code that is or might be JIT generated, and avoids checks on code
3523 that must have been compiled ahead of time. This significantly
3524 improves performance on applications that generate code at run time.
3526 * It is now possible to build a working Valgrind using Clang-2.9 on
3529 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3530 These enable and disable error reporting on a per-thread, and
3531 nestable, basis. This is useful for hiding errors in particularly
3532 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3533 now uses this facility.
3535 * Added the --mod-funcname option to cg_diff.
3537 * ==================== FIXED BUGS ====================
3539 The following bugs have been fixed or resolved. Note that "n-i-bz"
3540 stands for "not in bugzilla" -- that is, a bug that was reported to us
3541 but never got a bugzilla entry. We encourage you to file bugs in
3542 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3543 mailing the developers (or mailing lists) directly -- bugs that are
3544 not entered into bugzilla tend to get forgotten about or ignored.
3546 To see details of a given bug, visit
3547 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3548 where XXXXXX is the bug number as listed below.
3550 79311 malloc silly arg warning does not give stack trace
3551 210935 port valgrind.h (not valgrind) to win32 to support client requests
3552 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3553 243404 Port to zSeries
3554 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3555 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3556 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3557 253206 Some fixes for the faultstatus testcase
3558 255223 capget testcase fails when running as root
3559 256703 xlc_dbl_u32.c testcase broken
3560 256726 Helgrind tests have broken inline asm
3561 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3562 264800 testcase compile failure on zseries
3563 265762 make public VEX headers compilable by G++ 3.x
3564 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3565 266753 configure script does not give the user the option to not use QtCore
3566 266931 gen_insn_test.pl is broken
3567 266961 ld-linux.so.2 i?86-linux strlen issues
3568 266990 setns instruction causes false positive
3569 267020 Make directory for temporary files configurable at run-time.
3570 267342 == 267997 (segmentation fault on Mac OS 10.6)
3571 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3572 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3573 267488 regtest: darwin support for 64-bit build
3574 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3575 267630 Add support for IBM Power ISA 2.06 -- stage 1
3576 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3577 267819 Add client request for informing the core about reallocation
3578 267925 laog data structure quadratic for a single sequence of lock
3579 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3580 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3581 268513 missed optimizations in fold_Expr
3582 268619 s390x: fpr - gpr transfer facility
3583 268620 s390x: reconsider "long displacement" requirement
3584 268621 s390x: improve IR generation for XC
3585 268715 s390x: FLOGR is not universally available
3586 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3587 268930 s390x: MHY is not universally available
3588 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3589 269079 Support ptrace system call on ARM
3590 269144 missing "Bad option" error message
3591 269209 conditional load and store facility (z196)
3592 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3593 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3594 269736 s390x: minor code generation tweaks
3595 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3596 269863 s390x: remove unused function parameters
3597 269864 s390x: tweak s390_emit_load_cc
3598 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3599 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3600 270115 s390x: rewrite some testcases
3601 270309 == 267997 (valgrind crash on startup)
3602 270320 add support for Linux FIOQSIZE ioctl() call
3603 270326 segfault while trying to sanitize the environment passed to execle
3604 270794 IBM POWER7 support patch causes regression in none/tests
3605 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3606 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3607 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3608 270959 s390x: invalid use of R0 as base register
3609 271042 VSX configure check fails when it should not
3610 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3611 271259 s390x: fix code confusion
3612 271337 == 267997 (Valgrind segfaults on MacOS X)
3613 271385 s390x: Implement Ist_MBE
3614 271501 s390x: misc cleanups
3615 271504 s390x: promote likely and unlikely
3616 271579 ppc: using wrong enum type
3617 271615 unhandled instruction "popcnt" (arch=amd10h)
3618 271730 Fix bug when checking ioctls: duplicate check
3619 271776 s390x: provide STFLE instruction support
3620 271779 s390x: provide clock instructions like STCK
3621 271799 Darwin: ioctls without an arg report a memory error
3622 271820 arm: fix type confusion
3623 271917 pthread_cond_timedwait failure leads to not-locked false positive
3624 272067 s390x: fix DISP20 macro
3625 272615 A typo in debug output in mc_leakcheck.c
3626 272661 callgrind_annotate chokes when run from paths containing regex chars
3627 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3628 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3629 272967 make documentation build-system more robust
3630 272986 Fix gcc-4.6 warnings with valgrind.h
3631 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3632 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3633 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3634 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3635 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3636 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3637 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3638 273778 exp-ptrcheck: unhandled sysno == 259
3639 274089 exp-ptrcheck: unhandled sysno == 208
3640 274378 s390x: Various dispatcher tweaks
3641 274447 WARNING: unhandled syscall: 340
3642 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3643 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3644 274926 valgrind does not build against linux-3
3645 275148 configure FAIL with glibc-2.14
3646 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3647 275168 Make Valgrind work for MacOSX 10.7 Lion
3648 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3649 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3650 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3651 275308 Fix implementation for ppc64 fres instruc
3652 275339 s390x: fix testcase compile warnings
3653 275517 s390x: Provide support for CKSM instruction
3654 275710 s390x: get rid of redundant address mode calculation
3655 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3656 275852 == 250101 (valgrind uses all swap space and is killed)
3657 276784 Add support for IBM Power ISA 2.06 -- stage 3
3658 276987 gdbsrv: fix tests following recent commits
3659 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3660 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3661 277471 Unhandled syscall: 340
3662 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3663 277653 ARM: support Thumb2 PLD instruction
3664 277663 ARM: NEON float VMUL by scalar incorrect
3665 277689 ARM: tests for VSTn with register post-index are broken
3666 277694 ARM: BLX LR instruction broken in ARM mode
3667 277780 ARM: VMOV.F32 (immediate) instruction is broken
3668 278057 fuse filesystem syscall deadlocks
3669 278078 Unimplemented syscall 280 on ppc32
3670 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3671 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3672 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3673 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3674 279027 Support for MVCL and CLCL instruction
3675 279027 s390x: Provide support for CLCL and MVCL instructions
3676 279062 Remove a redundant check in the insn selector for ppc.
3677 279071 JDK creates PTEST with redundant REX.W prefix
3678 279212 gdbsrv: add monitor cmd v.info scheduler.
3679 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3680 279698 memcheck discards valid-bits for packuswb
3681 279795 memcheck reports uninitialised values for mincore on amd64
3682 279994 Add support for IBM Power ISA 2.06 -- stage 3
3683 280083 mempolicy syscall check errors
3684 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3685 280710 s390x: config files for nightly builds
3686 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3687 280965 Valgrind breaks fcntl locks when program does mmap
3688 281138 WARNING: unhandled syscall: 340
3689 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3690 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3691 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3692 281468 s390x: handle do_clone and gcc clones in call traces
3693 281488 ARM: VFP register corruption
3694 281828 == 275284 (false memmove warning: "Source and destination overlap")
3695 281883 s390x: Fix system call wrapper for "clone".
3696 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3697 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3698 282238 SLES10: make check fails
3699 282979 strcasestr needs replacement with recent(>=2.12) glibc
3700 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3701 283243 Regression in ppc64 memcheck tests
3702 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3703 283427 re-connect epoll_pwait syscall on ARM linux
3704 283600 gdbsrv: android: port vgdb.c
3705 283709 none/tests/faultstatus needs to account for page size
3706 284305 filter_gdb needs enhancement to work on ppc64
3707 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3708 284472 Thumb2 ROR.W encoding T2 not implemented
3709 284621 XML-escape process command line in XML output
3710 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3711 that have non-power-of-2 sizes (also AMDs)
3712 n-i-bz don't be spooked by libraries mashed by elfhack
3713 n-i-bz don't be spooked by libxul.so linked with gold
3714 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3716 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3717 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3718 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3722 Release 3.6.1 (16 February 2011)
3723 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3724 3.6.1 is a bug fix release. It adds support for some SSE4
3725 instructions that were omitted in 3.6.0 due to lack of time. Initial
3726 support for glibc-2.13 has been added. A number of bugs causing
3727 crashing or assertion failures have been fixed.
3729 The following bugs have been fixed or resolved. Note that "n-i-bz"
3730 stands for "not in bugzilla" -- that is, a bug that was reported to us
3731 but never got a bugzilla entry. We encourage you to file bugs in
3732 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3733 mailing the developers (or mailing lists) directly -- bugs that are
3734 not entered into bugzilla tend to get forgotten about or ignored.
3736 To see details of a given bug, visit
3737 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3738 where XXXXXX is the bug number as listed below.
3740 188572 Valgrind on Mac should suppress setenv() mem leak
3741 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3742 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3743 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3744 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3745 254420 memory pool tracking broken
3746 254957 Test code failing to compile due to changes in memcheck.h
3747 255009 helgrind/drd: crash on chmod with invalid parameter
3748 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3749 255355 helgrind/drd: crash on threaded programs doing fork
3751 255418 (SSE4.x) rint call compiled with ICC
3752 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3753 255888 closing valgrindoutput tag outputted to log-stream on error
3754 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3755 255966 Slowness when using mempool annotations
3756 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3757 256600 super-optimized strcasecmp() false positive
3758 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3759 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3760 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3761 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3762 257276 Missing case in memcheck --track-origins=yes
3763 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3764 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3765 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3766 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3767 263099 callgrind_annotate counts Ir improperly [...]
3768 263877 undefined coprocessor instruction on ARMv7
3769 265964 configure FAIL with glibc-2.13
3770 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3771 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3772 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3773 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3774 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3776 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3780 Release 3.6.0 (21 October 2010)
3781 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3782 3.6.0 is a feature release with many significant improvements and the
3783 usual collection of bug fixes.
3785 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3786 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3787 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3789 -------------------------
3791 Here are some highlights. Details are shown further down:
3793 * Support for ARM/Linux.
3795 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3797 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3799 * Support for the SSE4.2 instruction set.
3801 * Enhancements to the Callgrind profiler, including the ability to
3802 handle CPUs with three levels of cache.
3804 * A new experimental heap profiler, DHAT.
3806 * A huge number of bug fixes and small enhancements.
3808 -------------------------
3810 Here are details of the above changes, together with descriptions of
3811 many other changes, and a list of fixed bugs.
3813 * ================== PLATFORM CHANGES =================
3815 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3816 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3817 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3819 This requires a CPU capable of running the ARMv7-A instruction set
3820 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3821 of the user space instruction set, including ARM and Thumb integer
3822 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3823 Cachegrind and Massif tools work properly; other tools work to
3826 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3827 with support for recent releases of the underlying toolchain
3828 components, notably gcc-4.5 and glibc-2.12.
3830 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3831 support also works much better on OS X 10.5, and is as solid as
3834 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3835 64-bit mode. In 32-bit mode, support is only available up to and
3836 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3837 supported in 64-bit mode, and 32-bit mode does in fact support the
3838 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3839 10.6 on 32-bit targets.
3841 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3842 and including version 2.05 is supported.
3844 * ==================== TOOL CHANGES ====================
3846 * Cachegrind has a new processing script, cg_diff, which finds the
3847 difference between two profiles. It's very useful for evaluating
3848 the performance effects of a change in a program.
3850 Related to this change, the meaning of cg_annotate's (rarely-used)
3851 --threshold option has changed; this is unlikely to affect many
3852 people, if you do use it please see the user manual for details.
3854 * Callgrind now can do branch prediction simulation, similar to
3855 Cachegrind. In addition, it optionally can count the number of
3856 executed global bus events. Both can be used for a better
3857 approximation of a "Cycle Estimation" as derived event (you need to
3858 update the event formula in KCachegrind yourself).
3860 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3861 rather than the L2 cache. This is to accommodate machines with
3862 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3863 cache configuration of such a machine it will run the simulation as
3864 if the L2 cache isn't present. This means the results are less
3865 likely to match the true result for the machine, but
3866 Cachegrind/Callgrind's results are already only approximate, and
3867 should not be considered authoritative. The results are still
3868 useful for giving a general idea about a program's locality.
3870 * Massif has a new option, --pages-as-heap, which is disabled by
3871 default. When enabled, instead of tracking allocations at the level
3872 of heap blocks (as allocated with malloc/new/new[]), it instead
3873 tracks memory allocations at the level of memory pages (as mapped by
3874 mmap, brk, etc). Each mapped page is treated as its own block.
3875 Interpreting the page-level output is harder than the heap-level
3876 output, but this option is useful if you want to account for every
3877 byte of memory used by a program.
3879 * DRD has two new command-line options: --free-is-write and
3880 --trace-alloc. The former allows to detect reading from already freed
3881 memory, and the latter allows tracing of all memory allocations and
3884 * DRD has several new annotations. Custom barrier implementations can
3885 now be annotated, as well as benign races on static variables.
3887 * DRD's happens before / happens after annotations have been made more
3888 powerful, so that they can now also be used to annotate e.g. a smart
3889 pointer implementation.
3891 * Helgrind's annotation set has also been drastically improved, so as
3892 to provide to users a general set of annotations to describe locks,
3893 semaphores, barriers and condition variables. Annotations to
3894 describe thread-safe reference counted heap objects have also been
3897 * Memcheck has a new command-line option, --show-possibly-lost, which
3898 is enabled by default. When disabled, the leak detector will not
3899 show possibly-lost blocks.
3901 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3902 has been added. DHAT keeps track of allocated heap blocks, and also
3903 inspects every memory reference to see which block (if any) is being
3904 accessed. This gives a lot of insight into block lifetimes,
3905 utilisation, turnover, liveness, and the location of hot and cold
3906 fields. You can use DHAT to do hot-field profiling.
3908 * ==================== OTHER CHANGES ====================
3910 * Improved support for unfriendly self-modifying code: the extra
3911 overhead incurred by --smc-check=all has been reduced by
3912 approximately a factor of 5 as compared with 3.5.0.
3914 * Ability to show directory names for source files in error messages.
3915 This is combined with a flexible mechanism for specifying which
3916 parts of the paths should be shown. This is enabled by the new flag
3919 * A new flag, --require-text-symbol, which will stop the run if a
3920 specified symbol is not found it a given shared object when it is
3921 loaded into the process. This makes advanced working with function
3922 intercepting and wrapping safer and more reliable.
3924 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3925 and control of Valgrind is now available for the tools Memcheck and
3926 Helgrind. XML output from Valgrind is available for Memcheck,
3927 Helgrind and exp-Ptrcheck.
3929 * More reliable stack unwinding on amd64-linux, particularly in the
3930 presence of function wrappers, and with gcc-4.5 compiled code.
3932 * Modest scalability (performance improvements) for massive
3933 long-running applications, particularly for those with huge amounts
3936 * Support for analyzing programs running under Wine with has been
3937 improved. The header files <valgrind/valgrind.h>,
3938 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3939 Windows-programs compiled with MinGW or one of the Microsoft Visual
3942 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3943 The 32-bit simulator was not affected. This did not occur often,
3944 but when it did would usually crash the program under test.
3947 * A large number of bugs were fixed. These are shown below.
3949 * A number of bugs were investigated, and were candidates for fixing,
3950 but are not fixed in 3.6.0, due to lack of developer time. They may
3951 get fixed in later releases. They are:
3953 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3954 212419 false positive "lock order violated" (A+B vs A)
3955 213685 Undefined value propagates past dependency breaking instruction
3956 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3957 237920 valgrind segfault on fork failure
3958 242137 support for code compiled by LLVM-2.8
3959 242423 Another unknown Intel cache config value
3960 243232 Inconsistent Lock Orderings report with trylock
3961 243483 ppc: callgrind triggers VEX assertion failure
3962 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3963 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3965 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3966 249435 Analyzing wine programs with callgrind triggers a crash
3967 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3968 250065 Handling large allocations
3969 250101 huge "free" memory usage due to m_mallocfree.c
3970 "superblocks fragmentation"
3971 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3972 252091 Callgrind on ARM does not detect function returns correctly
3973 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3974 254420 memory pool tracking broken
3975 n-i-bz support for adding symbols for JIT generated code
3978 The following bugs have been fixed or resolved. Note that "n-i-bz"
3979 stands for "not in bugzilla" -- that is, a bug that was reported to us
3980 but never got a bugzilla entry. We encourage you to file bugs in
3981 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3982 mailing the developers (or mailing lists) directly -- bugs that are
3983 not entered into bugzilla tend to get forgotten about or ignored.
3985 To see details of a given bug, visit
3986 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3987 where XXXXXX is the bug number as listed below.
3989 135264 dcbzl instruction missing
3991 153699 Valgrind should report unaligned reads with movdqa
3993 190429 Valgrind reports lost of errors in ld.so
3994 with x86_64 2.9.90 glibc
3995 197266 valgrind appears to choke on the xmms instruction
3997 197988 Crash when demangling very large symbol names
3998 202315 unhandled syscall: 332 (inotify_init1)
3999 203256 Add page-level profiling to Massif
4000 205093 dsymutil=yes needs quotes, locking (partial fix)
4001 205241 Snow Leopard 10.6 support (partial fix)
4002 206600 Leak checker fails to upgrade indirect blocks when their
4003 parent becomes reachable
4004 210935 port valgrind.h (not valgrind) to win32 so apps run under
4005 wine can make client requests
4006 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
4007 within Linux ip-stack checksum functions
4008 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
4010 213685 Undefined value propagates past dependency breaking instruction
4012 215914 Valgrind inserts bogus empty environment variable
4014 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
4015 222545 shmat fails under valgind on some arm targets
4016 222560 ARM NEON support
4019 232509 Docs build fails with formatting inside <title></title> elements
4021 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
4022 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
4023 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
4024 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
4025 237485 symlink (syscall 57) is not supported on Mac OS
4026 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
4028 238208 is_just_below_ESP doesn't take into account red-zone
4029 238345 valgrind passes wrong $0 when executing a shell script
4030 238679 mq_timedreceive syscall doesn't flag the reception buffer
4032 238696 fcntl command F_DUPFD_CLOEXEC not supported
4033 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4034 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4035 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
4037 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
4043 242606 unhandled syscall: setegid (in Ptrcheck)
4044 242814 Helgrind "Impossible has happened" during
4045 QApplication::initInstance();
4046 243064 Valgrind attempting to read debug information from iso
4047 243270 Make stack unwinding in Valgrind wrappers more reliable
4048 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
4049 sysno = 277 (mq_open)
4050 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
4051 244493 ARM VFP d16-d31 registers support
4052 244670 add support for audit_session_self syscall on Mac OS 10.6
4053 244921 The xml report of helgrind tool is not well format
4054 244923 In the xml report file, the <preamble> not escape the
4055 xml char, eg '<','&','>'
4056 245535 print full path names in plain text reports
4057 245925 x86-64 red zone handling problem
4058 246258 Valgrind not catching integer underruns + new [] s
4059 246311 reg/reg cmpxchg doesn't work on amd64
4060 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
4061 246888 Improve Makefile.vex.am
4062 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
4063 to [f]chmod_extended
4064 247526 IBM POWER6 (ISA 2.05) support is incomplete
4065 247561 Some leak testcases fails due to reachable addresses in
4067 247875 sizeofIRType to handle Ity_I128
4068 247894 [PATCH] unhandled syscall sys_readahead
4069 247980 Doesn't honor CFLAGS passed to configure
4070 248373 darwin10.supp is empty in the trunk
4071 248822 Linux FIBMAP ioctl has int parameter instead of long
4072 248893 [PATCH] make readdwarf.c big endianess safe to enable
4073 unwinding on big endian systems
4074 249224 Syscall 336 not supported (SYS_proc_info)
4076 249775 Incorrect scheme for detecting NEON capabilities of host CPU
4077 249943 jni JVM init fails when using valgrind
4078 249991 Valgrind incorrectly declares AESKEYGENASSIST support
4080 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
4081 250799 frexp$fenv_access_off function generates SIGILL
4082 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
4083 251251 support pclmulqdq insn
4084 251362 valgrind: ARM: attach to debugger either fails or provokes
4086 251674 Unhandled syscall 294
4089 254257 Add support for debugfiles found by build-id
4090 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
4091 254646 Wrapped functions cause stack misalignment on OS X
4092 (and possibly Linux)
4093 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
4095 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
4099 Release 3.5.0 (19 August 2009)
4100 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4101 3.5.0 is a feature release with many significant improvements and the
4102 usual collection of bug fixes. The main improvement is that Valgrind
4103 now works on Mac OS X.
4105 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
4106 and X86/Darwin. Support for recent distros and toolchain components
4107 (glibc 2.10, gcc 4.5) has been added.
4109 -------------------------
4111 Here is a short summary of the changes. Details are shown further
4114 * Support for Mac OS X (10.5.x).
4116 * Improvements and simplifications to Memcheck's leak checker.
4118 * Clarification and simplifications in various aspects of Valgrind's
4121 * XML output for Helgrind and Ptrcheck.
4123 * Performance and stability improvements for Helgrind and DRD.
4125 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
4127 * A new experimental tool, BBV, useful for computer architecture
4130 * Improved Wine support, including ability to read Windows PDB
4133 -------------------------
4135 Here are details of the above changes, followed by descriptions of
4136 many other minor changes, and a list of fixed bugs.
4139 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
4140 called "Darwin" because that is the name of the OS core, which is the
4141 level that Valgrind works at.)
4145 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
4146 because it would require work and 10.4 is only becoming less common.
4148 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
4149 fairly well. For 10.5.x, 32-bit programs are the default even on
4150 64-bit machines, so it handles most current programs.
4152 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
4153 officially supported, but simple programs at least will probably work.
4154 However, start-up is slow.
4156 - PowerPC machines are not supported.
4158 Things that don't work:
4160 - The Ptrcheck tool.
4162 - Objective-C garbage collection.
4166 - If you have Rogue Amoeba's "Instant Hijack" program installed,
4167 Valgrind will fail with a SIGTRAP at start-up. See
4168 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
4173 - You will likely find --dsymutil=yes a useful option, as error
4174 messages may be imprecise without it.
4176 - Mac OS X support is new and therefore will be less robust than the
4177 Linux support. Please report any bugs you find.
4179 - Threaded programs may run more slowly than on Linux.
4181 Many thanks to Greg Parker for developing this port over several years.
4184 * Memcheck's leak checker has been improved.
4186 - The results for --leak-check=summary now match the summary results
4187 for --leak-check=full. Previously they could differ because
4188 --leak-check=summary counted "indirectly lost" blocks and
4189 "suppressed" blocks as "definitely lost".
4191 - Blocks that are only reachable via at least one interior-pointer,
4192 but are directly pointed to by a start-pointer, were previously
4193 marked as "still reachable". They are now correctly marked as
4196 - The default value for the --leak-resolution option has been
4197 changed from "low" to "high". In general, this means that more
4198 leak reports will be produced, but each leak report will describe
4199 fewer leaked blocks.
4201 - With --leak-check=full, "definitely lost" and "possibly lost"
4202 leaks are now considered as proper errors, ie. they are counted
4203 for the "ERROR SUMMARY" and affect the behaviour of
4204 --error-exitcode. These leaks are not counted as errors if
4205 --leak-check=summary is specified, however.
4207 - Documentation for the leak checker has been improved.
4210 * Various aspects of Valgrind's text output have changed.
4212 - Valgrind's start-up message has changed. It is shorter but also
4213 includes the command being run, which makes it easier to use
4214 --trace-children=yes. An example:
4216 - Valgrind's shut-down messages have also changed. This is most
4217 noticeable with Memcheck, where the leak summary now occurs before
4218 the error summary. This change was necessary to allow leaks to be
4219 counted as proper errors (see the description of the leak checker
4220 changes above for more details). This was also necessary to fix a
4221 longstanding bug in which uses of suppressions against leaks were
4222 not "counted", leading to difficulties in maintaining suppression
4223 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
4225 - Behavior of -v has changed. In previous versions, -v printed out
4226 a mixture of marginally-user-useful information, and tool/core
4227 statistics. The statistics printing has now been moved to its own
4228 flag, --stats=yes. This means -v is less verbose and more likely
4229 to convey useful end-user information.
4231 - The format of some (non-XML) stack trace entries has changed a
4232 little. Previously there were six possible forms:
4234 0x80483BF: really (a.c:20)
4235 0x80483BF: really (in /foo/a.out)
4237 0x80483BF: (within /foo/a.out)
4238 0x80483BF: ??? (a.c:20)
4241 The third and fourth of these forms have been made more consistent
4242 with the others. The six possible forms are now:
4244 0x80483BF: really (a.c:20)
4245 0x80483BF: really (in /foo/a.out)
4246 0x80483BF: really (in ???)
4247 0x80483BF: ??? (in /foo/a.out)
4248 0x80483BF: ??? (a.c:20)
4251 Stack traces produced when --xml=yes is specified are different
4255 * Helgrind and Ptrcheck now support XML output, so they can be used
4256 from GUI tools. Also, the XML output mechanism has been
4259 - The XML format has been overhauled and generalised, so it is more
4260 suitable for error reporting tools in general. The Memcheck
4261 specific aspects of it have been removed. The new format, which
4262 is an evolution of the old format, is described in
4263 docs/internals/xml-output-protocol4.txt.
4265 - Memcheck has been updated to use the new format.
4267 - Helgrind and Ptrcheck are now able to emit output in this format.
4269 - The XML output mechanism has been overhauled. XML is now output
4270 to its own file descriptor, which means that:
4272 * Valgrind can output text and XML independently.
4274 * The longstanding problem of XML output being corrupted by
4275 unexpected un-tagged text messages is solved.
4277 As before, the destination for text output is specified using
4278 --log-file=, --log-fd= or --log-socket=.
4280 As before, XML output for a tool is enabled using --xml=yes.
4282 Because there's a new XML output channel, the XML output
4283 destination is now specified by --xml-file=, --xml-fd= or
4286 Initial feedback has shown this causes some confusion. To
4287 clarify, the two envisaged usage scenarios are:
4289 (1) Normal text output. In this case, do not specify --xml=yes
4290 nor any of --xml-file=, --xml-fd= or --xml-socket=.
4292 (2) XML output. In this case, specify --xml=yes, and one of
4293 --xml-file=, --xml-fd= or --xml-socket= to select the XML
4294 destination, one of --log-file=, --log-fd= or --log-socket=
4295 to select the destination for any remaining text messages,
4296 and, importantly, -q.
4298 -q makes Valgrind completely silent on the text channel,
4299 except in the case of critical failures, such as Valgrind
4300 itself segfaulting, or failing to read debugging information.
4301 Hence, in this scenario, it suffices to check whether or not
4302 any output appeared on the text channel. If yes, then it is
4303 likely to be a critical error which should be brought to the
4304 attention of the user. If no (the text channel produced no
4305 output) then it can be assumed that the run was successful.
4307 This allows GUIs to make the critical distinction they need to
4308 make (did the run fail or not?) without having to search or
4309 filter the text output channel in any way.
4311 It is also recommended to use --child-silent-after-fork=yes in
4315 * Improvements and changes in Helgrind:
4317 - XML output, as described above
4319 - Checks for consistent association between pthread condition
4320 variables and their associated mutexes are now performed.
4322 - pthread_spinlock functions are supported.
4324 - Modest performance improvements.
4326 - Initial (skeletal) support for describing the behaviour of
4327 non-POSIX synchronisation objects through ThreadSanitizer
4328 compatible ANNOTATE_* macros.
4330 - More controllable tradeoffs between performance and the level of
4331 detail of "previous" accesses in a race. There are now three
4334 * --history-level=full. This is the default, and was also the
4335 default in 3.4.x. It shows both stacks involved in a race, but
4336 requires a lot of memory and can be very slow in programs that
4337 do many inter-thread synchronisation events.
4339 * --history-level=none. This only shows the later stack involved
4340 in a race. This can be much faster than --history-level=full,
4341 but makes it much more difficult to find the other access
4342 involved in the race.
4344 The new intermediate setting is
4346 * --history-level=approx
4348 For the earlier (other) access, two stacks are presented. The
4349 earlier access is guaranteed to be somewhere in between the two
4350 program points denoted by those stacks. This is not as useful
4351 as showing the exact stack for the previous access (as per
4352 --history-level=full), but it is better than nothing, and it's
4353 almost as fast as --history-level=none.
4356 * New features and improvements in DRD:
4358 - The error messages printed by DRD are now easier to interpret.
4359 Instead of using two different numbers to identify each thread
4360 (Valgrind thread ID and DRD thread ID), DRD does now identify
4361 threads via a single number (the DRD thread ID). Furthermore
4362 "first observed at" information is now printed for all error
4363 messages related to synchronization objects.
4365 - Added support for named semaphores (sem_open() and sem_close()).
4367 - Race conditions between pthread_barrier_wait() and
4368 pthread_barrier_destroy() calls are now reported.
4370 - Added support for custom allocators through the macros
4371 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
4372 in <valgrind/valgrind.h>). An alternative for these two macros is
4373 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
4376 - Added support for annotating non-POSIX synchronization objects
4377 through several new ANNOTATE_*() macros.
4379 - OpenMP: added support for the OpenMP runtime (libgomp) included
4380 with gcc versions 4.3.0 and 4.4.0.
4384 - Added two new command-line options (--first-race-only and
4385 --segment-merging-interval).
4388 * Genuinely atomic support for x86/amd64/ppc atomic instructions
4390 Valgrind will now preserve (memory-access) atomicity of LOCK-
4391 prefixed x86/amd64 instructions, and any others implying a global
4392 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
4394 This means that Valgrinded processes will "play nicely" in
4395 situations where communication with other processes, or the kernel,
4396 is done through shared memory and coordinated with such atomic
4397 instructions. Prior to this change, such arrangements usually
4398 resulted in hangs, races or other synchronisation failures, because
4399 Valgrind did not honour atomicity of such instructions.
4402 * A new experimental tool, BBV, has been added. BBV generates basic
4403 block vectors for use with the SimPoint analysis tool, which allows
4404 a program's overall behaviour to be approximated by running only a
4405 fraction of it. This is useful for computer architecture
4406 researchers. You can run BBV by specifying --tool=exp-bbv (the
4407 "exp-" prefix is short for "experimental"). BBV was written by
4411 * Modestly improved support for running Windows applications under
4412 Wine. In particular, initial support for reading Windows .PDB debug
4413 information has been added.
4416 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
4417 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
4421 * The Valgrind client requests VALGRIND_PRINTF and
4422 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
4423 the string was always printed immediately on its own line. Now, the
4424 string will be added to a buffer but not printed until a newline is
4425 encountered, or other Valgrind output is printed (note that for
4426 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
4427 "other Valgrind output"). This allows you to use multiple
4428 VALGRIND_PRINTF calls to build up a single output line, and also to
4429 print multiple output lines with a single request (by embedding
4430 multiple newlines in the string).
4433 * The graphs drawn by Massif's ms_print program have changed slightly:
4435 - The half-height chars '.' and ',' are no longer drawn, because
4436 they are confusing. The --y option can be used if the default
4437 y-resolution is not high enough.
4439 - Horizontal lines are now drawn after the top of a snapshot if
4440 there is a gap until the next snapshot. This makes it clear that
4441 the memory usage has not dropped to zero between snapshots.
4444 * Something that happened in 3.4.0, but wasn't clearly announced: the
4445 option --read-var-info=yes can be used by some tools (Memcheck,
4446 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
4447 variable type and location information. This makes those tools
4448 start up more slowly and increases memory consumption, but
4449 descriptions of data addresses in error messages become more
4453 * exp-Omega, an experimental instantaneous leak-detecting tool, was
4454 disabled in 3.4.0 due to a lack of interest and maintenance,
4455 although the source code was still in the distribution. The source
4456 code has now been removed from the distribution. For anyone
4457 interested, the removal occurred in SVN revision r10247.
4460 * Some changes have been made to the build system.
4462 - VEX/ is now integrated properly into the build system. This means
4463 that dependency tracking within VEX/ now works properly, "make
4464 install" will work without requiring "make" before it, and
4465 parallel builds (ie. 'make -j') now work (previously a
4466 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
4467 was effectively ignored).
4469 - The --with-vex configure option has been removed. It was of
4470 little use and removing it simplified the build system.
4472 - The location of some install files has changed. This should not
4473 affect most users. Those who might be affected:
4475 * For people who use Valgrind with MPI programs, the installed
4476 libmpiwrap.so library has moved from
4477 $(INSTALL)/<platform>/libmpiwrap.so to
4478 $(INSTALL)/libmpiwrap-<platform>.so.
4480 * For people who distribute standalone Valgrind tools, the
4481 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
4482 have moved to $(INSTALL)/libcoregrind-<platform>.a.
4484 These changes simplify the build system.
4486 - Previously, all the distributed suppression (*.supp) files were
4487 installed. Now, only default.supp is installed. This should not
4488 affect users as the other installed suppression files were not
4489 read; the fact that they were installed was a mistake.
4492 * KNOWN LIMITATIONS:
4494 - Memcheck is unusable with the Intel compiler suite version 11.1,
4495 when it generates code for SSE2-and-above capable targets. This
4496 is because of icc's use of highly optimised inlined strlen
4497 implementations. It causes Memcheck to report huge numbers of
4498 false errors even in simple programs. Helgrind and DRD may also
4501 Versions 11.0 and earlier may be OK, but this has not been
4505 The following bugs have been fixed or resolved. Note that "n-i-bz"
4506 stands for "not in bugzilla" -- that is, a bug that was reported to us
4507 but never got a bugzilla entry. We encourage you to file bugs in
4508 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4509 mailing the developers (or mailing lists) directly -- bugs that are
4510 not entered into bugzilla tend to get forgotten about or ignored.
4512 To see details of a given bug, visit
4513 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4514 where XXXXXX is the bug number as listed below.
4516 84303 How about a LockCheck tool?
4517 91633 dereference of null ptr in vgPlain_st_basetype
4518 97452 Valgrind doesn't report any pthreads problems
4519 100628 leak-check gets assertion failure when using
4520 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4521 108528 NPTL pthread cleanup handlers not called
4522 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4523 110128 mallinfo is not implemented...
4524 110770 VEX: Generated files not always updated when making valgrind
4525 111102 Memcheck: problems with large (memory footprint) applications
4526 115673 Vex's decoder should never assert
4527 117564 False positive: Syscall param clone(child_tidptr) contains
4528 uninitialised byte(s)
4529 119404 executing ssh from inside valgrind fails
4530 133679 Callgrind does not write path names to sources with dwarf debug
4532 135847 configure.in problem with non gnu compilers (and possible fix)
4533 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4534 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4535 136230 memcheck reports "possibly lost", should be "still reachable"
4536 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4537 137904 Valgrind reports a memory leak when using POSIX threads,
4539 139076 valgrind VT_GETSTATE error
4540 142228 complaint of elf_dynamic_do_rela in trivial usage
4541 145347 spurious warning with USBDEVFS_REAPURB
4542 148441 (wine) can't find memory leak in Wine, win32 binary
4544 148742 Leak-check fails assert on exit
4545 149878 add (proper) check for calloc integer overflow
4546 150606 Call graph is broken when using callgrind control
4547 152393 leak errors produce an exit code of 0. I need some way to
4548 cause leak errors to result in a nonzero exit code.
4549 157154 documentation (leak-resolution doc speaks about num-callers
4550 def=4) + what is a loss record
4551 159501 incorrect handling of ALSA ioctls
4552 162020 Valgrinding an empty/zero-byte file crashes valgrind
4553 162482 ppc: Valgrind crashes while reading stabs information
4554 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4555 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4556 163560 VEX/test_main.c is missing from valgrind-3.3.1
4557 164353 malloc_usable_size() doesn't return a usable size
4558 165468 Inconsistent formatting in memcheck manual -- please fix
4559 169505 main.c:286 (endOfInstr):
4560 Assertion 'ii->cost_offset == *cost_offset' failed
4561 177206 Generate default.supp during compile instead of configure
4562 177209 Configure valt_load_address based on arch+os
4563 177305 eventfd / syscall 323 patch lost
4564 179731 Tests fail to build because of inlining of non-local asm labels
4565 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4566 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4567 181594 Bogus warning for empty text segment
4568 181707 dwarf doesn't require enumerations to have name
4569 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4570 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4571 Assertion '!already_present' failed.
4572 185359 exp-ptrcheck: unhandled syscall getresuid()
4573 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4574 185816 Valgrind is unable to handle debug info for files with split
4575 debug info that are prelinked afterwards
4576 185980 [darwin] unhandled syscall: sem_open
4577 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4578 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4579 186790 Suppression pattern used for leaks are not reported
4580 186796 Symbols with length>200 in suppression files are ignored
4581 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4582 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4583 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4584 188046 bashisms in the configure script
4585 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4586 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4587 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4588 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4589 assertion fail "!lock->heldBy"
4590 188427 Add support for epoll_create1 (with patch)
4591 188530 Support for SIOCGSTAMPNS
4592 188560 Include valgrind.spec in the tarball
4593 188572 Valgrind on Mac should suppress setenv() mem leak
4594 189054 Valgrind fails to build because of duplicate non-local asm labels
4595 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4596 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4597 189763 drd assertion failure: s_threadinfo[tid].is_recording
4598 190219 unhandled syscall: 328 (x86-linux)
4599 190391 dup of 181394; see above
4600 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4601 190820 No debug information on powerpc-linux
4602 191095 PATCH: Improve usbdevfs ioctl handling
4603 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4605 191189 --xml=yes should obey --gen-suppressions=all
4606 191192 syslog() needs a suppression on macosx
4607 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4608 191761 getrlimit on MacOSX
4609 191992 multiple --fn-skip only works sometimes; dependent on order
4610 192634 V. reports "aspacem sync_check_mapping_callback:
4611 segment mismatch" on Darwin
4612 192954 __extension__ missing on 2 client requests
4613 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4614 194474 "INSTALL" file has different build instructions than "README"
4615 194671 Unhandled syscall (sem_wait?) from mac valgrind
4616 195069 memcheck: reports leak (memory still reachable) for
4618 195169 drd: (vgDrd_barrier_post_wait):
4619 Assertion 'r->sg[p->post_iteration]' failed.
4620 195268 valgrind --log-file doesn't accept ~/...
4621 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4622 195860 WARNING: unhandled syscall: unix:223
4623 196528 need a error suppression for pthread_rwlock_init under os x?
4624 197227 Support aio_* syscalls on Darwin
4625 197456 valgrind should reject --suppressions=(directory)
4626 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4627 197591 unhandled syscall 27 (mincore)
4628 197793 Merge DCAS branch to the trunk == 85756, 142103
4629 197794 Avoid duplicate filenames in Vex
4630 197898 make check fails on current SVN
4631 197901 make check fails also under exp-ptrcheck in current SVN
4632 197929 Make --leak-resolution=high the default
4633 197930 Reduce spacing between leak reports
4634 197933 Print command line of client at start-up, and shorten preamble
4635 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4636 198395 add BBV to the distribution as an experimental tool
4637 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4638 198649 callgrind_annotate doesn't cumulate counters
4639 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4640 199977 Valgrind complains about an unrecognized instruction in the
4641 atomic_incs test program
4642 200029 valgrind isn't able to read Fedora 12 debuginfo
4643 200760 darwin unhandled syscall: unix:284
4644 200827 DRD doesn't work on Mac OS X
4645 200990 VG_(read_millisecond_timer)() does not work correctly
4646 201016 Valgrind does not support pthread_kill() on Mac OS
4647 201169 Document --read-var-info
4648 201323 Pre-3.5.0 performance sanity checking
4649 201384 Review user manual for the 3.5.0 release
4650 201585 mfpvr not implemented on ppc
4651 201708 tests failing because x86 direction flag is left set
4652 201757 Valgrind doesn't handle any recent sys_futex additions
4653 204377 64-bit valgrind can not start a shell script
4654 (with #!/path/to/shell) if the shell is a 32-bit executable
4655 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4656 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4657 about memory allocation events with memory access tracing enabled
4658 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4660 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4664 Release 3.4.1 (28 February 2009)
4665 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4666 3.4.1 is a bug-fix release that fixes some regressions and assertion
4667 failures in debug info reading in 3.4.0, most notably incorrect stack
4668 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4669 other debug info problems are also fixed. A number of bugs in the
4670 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4672 In view of the fact that 3.4.0 contains user-visible regressions
4673 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4674 encouraged to ship 3.4.1 in preference to 3.4.0.
4676 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4677 bugzilla" -- that is, a bug that was reported to us but never got a
4678 bugzilla entry. We encourage you to file bugs in bugzilla
4679 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4680 developers (or mailing lists) directly -- bugs that are not entered
4681 into bugzilla tend to get forgotten about or ignored.
4683 n-i-bz Fix various bugs reading icc-11 generated debug info
4684 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4685 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4686 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4687 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4688 179618 exp-ptrcheck crashed / exit prematurely
4689 179624 helgrind: false positive races with pthread_create and
4690 recv/open/close/read
4691 134207 pkg-config output contains @VG_PLATFORM@
4692 176926 floating point exception at valgrind startup with PPC 440EPX
4693 181594 Bogus warning for empty text segment
4694 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4695 181707 Dwarf3 doesn't require enumerations to have name
4696 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4697 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4698 Assertion '!already_present' failed.
4699 185359 exp-ptrcheck unhandled syscall getresuid()
4701 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4702 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4706 Release 3.4.0 (2 January 2009)
4707 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4708 3.4.0 is a feature release with many significant improvements and the
4709 usual collection of bug fixes. This release supports X86/Linux,
4710 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4711 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4713 3.4.0 brings some significant tool improvements. Memcheck can now
4714 report the origin of uninitialised values, the thread checkers
4715 Helgrind and DRD are much improved, and we have a new experimental
4716 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4717 global arrays. In detail:
4719 * Memcheck is now able to track the origin of uninitialised values.
4720 When it reports an uninitialised value error, it will try to show
4721 the origin of the value, as either a heap or stack allocation.
4722 Origin tracking is expensive and so is not enabled by default. To
4723 use it, specify --track-origins=yes. Memcheck's speed will be
4724 essentially halved, and memory usage will be significantly
4725 increased. Nevertheless it can drastically reduce the effort
4726 required to identify the root cause of uninitialised value errors,
4727 and so is often a programmer productivity win, despite running more
4730 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4731 3.4.0, will be released shortly.
4733 * Helgrind's race detection algorithm has been completely redesigned
4734 and reimplemented, to address usability and scalability concerns:
4736 - The new algorithm has a lower false-error rate: it is much less
4737 likely to report races that do not really exist.
4739 - Helgrind will display full call stacks for both accesses involved
4740 in a race. This makes it easier to identify the root causes of
4743 - Limitations on the size of program that can run have been removed.
4745 - Performance has been modestly improved, although that is very
4748 - Direct support for Qt4 threading has been added.
4750 - pthread_barriers are now directly supported.
4752 - Helgrind works well on all supported Linux targets.
4754 * The DRD thread debugging tool has seen major improvements:
4756 - Greatly improved performance and significantly reduced memory
4759 - Support for several major threading libraries (Boost.Thread, Qt4,
4760 glib, OpenMP) has been added.
4762 - Support for atomic instructions, POSIX semaphores, barriers and
4763 reader-writer locks has been added.
4765 - Works now on PowerPC CPUs too.
4767 - Added support for printing thread stack usage at thread exit time.
4769 - Added support for debugging lock contention.
4771 - Added a manual for Drd.
4773 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4774 checks for misuses of pointers. In that sense it is a bit like
4775 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4776 detect overruns of stack and global arrays, it can detect
4777 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4778 detect accesses heap blocks that have been freed a very long time
4779 ago (millions of blocks in the past).
4781 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4782 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4783 of the main Valgrind documentation. As this is an experimental
4784 tool, we would be particularly interested in hearing about your
4785 experiences with it.
4787 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4788 longer built by default, although the code remains in the repository
4789 and the tarball. This is due to three factors: a perceived lack of
4790 users, a lack of maintenance, and concerns that it may not be
4791 possible to achieve reliable operation using the existing design.
4793 * As usual, support for the latest Linux distros and toolchain
4794 components has been added. It should work well on Fedora Core 10,
4795 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4796 state) is supported, as is glibc-2.9. The C++ demangler has been
4797 updated so as to work well with C++ compiled by even the most recent
4800 * You can now use frame-level wildcards in suppressions. This was a
4801 frequently-requested enhancement. A line "..." in a suppression now
4802 matches zero or more frames. This makes it easier to write
4803 suppressions which are precise yet insensitive to changes in
4806 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4808 * Very basic support for IBM Power6 has been added (64-bit processes only).
4810 * Valgrind is now cross-compilable. For example, it is possible to
4811 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4812 on a ppc32/64-linux target.
4814 * You can set the main thread's stack size at startup using the
4815 new --main-stacksize= flag (subject of course to ulimit settings).
4816 This is useful for running apps that need a lot of stack space.
4818 * The limitation that you can't use --trace-children=yes together
4819 with --db-attach=yes has been removed.
4821 * The following bugs have been fixed. Note that "n-i-bz" stands for
4822 "not in bugzilla" -- that is, a bug that was reported to us but
4823 never got a bugzilla entry. We encourage you to file bugs in
4824 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4825 mailing the developers (or mailing lists) directly.
4827 n-i-bz Make return types for some client requests 64-bit clean
4828 n-i-bz glibc 2.9 support
4829 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4830 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4831 n-i-bz Building in an env without gdb gives bogus gdb attach
4832 92456 Tracing the origin of uninitialised memory
4833 106497 Valgrind does not demangle some C++ template symbols
4835 151612 Suppression with "..." (frame-level wildcards in .supp files)
4836 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4837 159285 unhandled syscall:25 (stime, on x86-linux)
4838 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4839 160954 ppc build of valgrind crashes with illegal instruction (isel)
4840 160956 mallinfo implementation, w/ patch
4841 162092 Valgrind fails to start gnome-system-monitor
4842 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4843 163794 assertion failure with "--track-origins=yes"
4844 163933 sigcontext.err and .trapno must be set together
4845 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4846 164476 Missing kernel module loading system calls
4847 164669 SVN regression: mmap() drops posix file locks
4848 166581 Callgrind output corruption when program forks
4849 167288 Patch file for missing system calls on Cell BE
4850 168943 unsupported scas instruction pentium
4851 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4852 172417 x86->IR: 0x82 ...
4853 172563 amd64->IR: 0xD9 0xF5 - fprem1
4854 173099 .lds linker script generation error
4855 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4856 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4858 174908 --log-file value not expanded correctly for core file
4859 175044 Add lookup_dcookie for amd64
4860 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4862 Developer-visible changes:
4864 * Valgrind's debug-info reading machinery has been majorly overhauled.
4865 It can now correctly establish the addresses for ELF data symbols,
4866 which is something that has never worked properly before now.
4868 Also, Valgrind can now read DWARF3 type and location information for
4869 stack and global variables. This makes it possible to use the
4870 framework to build tools that rely on knowing the type and locations
4871 of stack and global variables, for example exp-Ptrcheck.
4873 Reading of such information is disabled by default, because most
4874 tools don't need it, and because it is expensive in space and time.
4875 However, you can force Valgrind to read it, using the
4876 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4877 make use of such information, if present, to provide source-level
4878 descriptions of data addresses in the error messages they create.
4880 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4881 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).