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.
24 * Add support for the FreeBSD close_range system call.
26 * ==================== TOOL CHANGES ===================
28 * Memcheck now tests and warns about the values used for
29 alignment and size. These apply to various functions: memalign,
30 posix_memalign and aligned_alloc in C and various overloads
31 of operators new and delete in C++. The kinds of error that can
33 - invalid alignment, for instance the alignment is usually required
35 - mismatched alignment between aligned allocation and aligned
37 - mismatched size when sized delete is used
38 - bad size for functions that have implementation defined behaviour
39 when the requested size is zero
42 - You can now profile part of a program's execution using the new
43 `CACHEGRIND_START_INSTRUMENTATION` and `CACHEGRIND_STOP_INSTRUMENTATION`
44 client requests, along with the new `--instr-at-start` option. The
45 behaviour is the same as Callgrind's equivalent functionality.
47 * ==================== FIXED BUGS ====================
49 The following bugs have been fixed or resolved. Note that "n-i-bz"
50 stands for "not in bugzilla" -- that is, a bug that was reported to us
51 but never got a bugzilla entry. We encourage you to file bugs in
52 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
53 than mailing the developers (or mailing lists) directly -- bugs that
54 are not entered into bugzilla tend to get forgotten about or ignored.
56 426751 Valgrind reports "still reachable" memory using musl (alpine running inside docker)
57 433857 Add validation to C++17 aligned new/delete alignment size
58 433859 Add mismatched detection to C++ 17 aligned new/delete
59 460192 Add epoll_pwait2
60 466105 aligned_alloc problems, part 2
61 467441 Add mismatched detection to C++ 14 sized delete
62 469049 link failure on ppc64 (big endian) valgrind 3.20
63 469146 massif --ignore-fn does not ignore inlined functions
64 469768 Make it possible to install gdb scripts in a different location
65 470121 Can't run callgrind_control with valgrind 3.21.0 because of perl errors
66 470132 s390x: Assertion failure on VGM instruction
67 470520 Multiple realloc zero errors crash in MC_(eq_Error)
68 470713 Failure on the Yosys project: valgrind: m_libcfile.c:1802
69 (Bool vgPlain_realpath(const HChar *, HChar *)):
70 Assertion 'resolved' failed
71 470830 Don't print actions vgdb me ... continue for vgdb --multi mode
72 470978 s390x: Valgrind cannot start qemu-kvm when "sysctl vm.allocate_pgste=0"
73 471311 gdb --multi mode stdout redirecting to stderr
74 471807 Add support for lazy reading and downloading of DWARF debuginfo
75 472219 Syscall param ppoll(ufds.events) points to uninitialised byte(s)
76 472875 none/tests/s390x/dfp-1 failure
77 472963 Broken regular expression in configure.ac
78 473604 Fix bug472219.c compile failure with Clang 16
79 473677 make check compile failure with Clang 16 based on GCC 13.x
80 473745 must-be-redirected function - strlen
81 473870 FreeBSD 14 applications fail early at startup
82 473944 Handle mold linker split RW PT_LOAD segments correctly
83 474332 aligned_alloc under Valgrind returns nullptr when alignment is not a multiple of sizeof(void *)
84 n-i-bz Allow arguments with spaces in .valgrindrc files
86 To see details of a given bug, visit
87 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
88 where XXXXXX is the bug number as listed above.
90 Release 3.21.0 (28 Apr 2023)
91 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
93 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
94 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
95 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
96 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
97 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
98 AMD64/macOS 10.13 and nanoMIPS/Linux.
100 * ==================== CORE CHANGES ===================
102 * When GDB is used to debug a program running under valgrind using
103 the valgrind gdbserver, GDB will automatically load some
104 python code provided in valgrind defining GDB front end commands
105 corresponding to the valgrind monitor commands.
106 These GDB front end commands accept the same format as
107 the monitor commands directly sent to the Valgrind gdbserver.
108 These GDB front end commands provide a better integration
109 in the GDB command line interface, so as to use for example
110 GDB auto-completion, command specific help, searching for
111 a command or command help matching a regexp, ...
112 For relevant monitor commands, GDB will evaluate arguments
113 to make the use of monitor commands easier.
114 For example, instead of having to print the address of a variable
115 to pass it to a subsequent monitor command, the GDB front end
116 command will evaluate the address argument. It is for example
118 (gdb) memcheck who_points_at &some_struct sizeof(some_struct)
121 $2 = (some_struct_type *) 0x1130a0 <some_struct>
122 (gdb) p sizeof(some_struct)
124 (gdb) monitor who_point_at 0x1130a0 40
126 * The vgdb utility now supports extended-remote protocol when
127 invoked with --multi. In this mode the GDB run command is
128 supported. Which means you don't need to run gdb and valgrind
129 from different terminals. So for example to start your program
130 in gdb and run it under valgrind you can do:
132 (gdb) set remote exec-file prog
134 (gdb) target extended-remote | vgdb --multi
137 * The behaviour of realloc with a size of zero can now
138 be changed for tools that intercept malloc. Those
139 tools are memcheck, helgrind, drd, massif and dhat.
140 Realloc implementations generally do one of two things
141 - free the memory like free() and return NULL
142 (GNU libc and ptmalloc).
143 - either free the memory and then allocate a
144 minimum sized block or just return the
145 original pointer. Return NULL if the
146 allocation of the minimum sized block fails
147 (jemalloc, musl, snmalloc, Solaris, macOS).
148 When Valgrind is configured and built it will
149 try to match the OS and libc behaviour. However
150 if you are using a non-default library to replace
151 malloc and family (e.g., musl on a glibc Linux or
152 tcmalloc on FreeBSD) then you can use a command line
153 option to change the behaviour of Valgrind:
154 --realloc-zero-bytes-frees=yes|no [yes on Linux glibc, no otherwise]
156 * ================== PLATFORM CHANGES =================
158 * Make the address space limit on FreeBSD amd64 128Gbytes
159 (the same as Linux and Solaris, it was 32Gbytes)
161 * ==================== TOOL CHANGES ===================
164 - When doing a delta leak_search, it is now possible to only
165 output the new loss records compared to the previous leak search.
166 This is available in the memcheck monitor command 'leak_search'
167 by specifying the "new" keyword or in your program by using
168 the client request VALGRIND_DO_NEW_LEAK_CHECK.
169 Whenever a "delta" leak search is done (i.e. when specifying
170 "new" or "increased" or "changed" in the monitor command),
171 the new loss records have a "new" marker.
172 - Valgrind now contains python code that defines GDB memcheck
173 front end monitor commands. See CORE CHANGES.
174 - Performs checks for the use of realloc with a size of zero.
175 This is non-portable and a source of errors. If memcheck
176 detects such a usage it will generate an error
177 realloc() with size 0
178 followed by the usual callstacks.
179 A switch has been added to allow this to be turned off:
180 --show-realloc-size-zero=yes|no [yes]
183 - The option ---history-backtrace-size=<number> allows to configure
184 the number of entries to record in the stack traces of "old"
185 accesses. Previously, this number was hardcoded to 8.
186 - Valgrind now contains python code that defines GDB helgrind
187 front end monitor commands. See CORE CHANGES.
190 - `--cache-sim=no` is now the default. The cache simulation is old and
191 unlikely to match any real modern machine. This means only the `Ir`
192 event are gathered by default, but that is by far the most useful
194 - `cg_annotate`, `cg_diff`, and `cg_merge` have been rewritten in
195 Python. As a result, they all have more flexible command line
196 argument handling, e.g. supporting `--show-percs` and
197 `--no-show-percs` forms as well as the existing `--show-percs=yes`
198 and `--show-percs=no`.
199 - `cg_annotate` has some functional changes.
200 - It's much faster, e.g. 3-4x on common cases.
201 - It now supports diffing (with `--diff`, `--mod-filename`, and
202 `--mod-funcname`) and merging (by passing multiple data files).
203 - It now provides more information at the file and function level.
204 There are now "File:function" and "Function:file" sections. These
205 are very useful for programs that use inlining a lot.
206 - Support for user-annotated files and the `-I`/`--include` option
207 has been removed, because it was of little use and blocked other
209 - The `--auto` option is renamed `--annotate`, though the old
210 `--auto=yes`/`--auto=no` forms are still supported.
211 - `cg_diff` and `cg_merge` are now deprecated, because `cg_annotate`
212 now does a better job of diffing and merging.
213 - The Cachegrind output file format has changed very slightly, but in
214 ways nobody is likely to notice.
217 - Valgrind now contains python code that defines GDB callgrind
218 front end monitor commands. See CORE CHANGES.
221 - Valgrind now contains python code that defines GDB massif
222 front end monitor commands. See CORE CHANGES.
225 - A new kind of user request has been added which allows you to
226 override the 1024 byte limit on access count histograms for blocks
227 of memory. The client request is DHAT_HISTOGRAM_MEMORY.
229 * ==================== FIXED BUGS ====================
231 The following bugs have been fixed or resolved. Note that "n-i-bz"
232 stands for "not in bugzilla" -- that is, a bug that was reported to us
233 but never got a bugzilla entry. We encourage you to file bugs in
234 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
235 than mailing the developers (or mailing lists) directly -- bugs that
236 are not entered into bugzilla tend to get forgotten about or ignored.
238 170510 Don't warn about ioctl of size 0 without direction hint
239 241072 List tools in --help output
240 327548 false positive while destroying mutex
241 382034 Testcases build fixes for musl
242 351857 confusing error message about valid command line option
243 374596 inconsistent RDTSCP support on x86_64
244 392331 Spurious lock not held error from inside pthread_cond_timedwait
245 397083 Likely false positive "uninitialised value(s)" for __wmemchr_avx2 and __wmemcmp_avx2_movbe
246 400793 pthread_rwlock_timedwrlock false positive
247 419054 Unhandled syscall getcpu on arm32
248 433873 openat2 syscall unimplemented on Linux
249 434057 Add stdio mode to valgrind's gdbserver
250 435441 valgrind fails to interpose malloc on musl 1.2.2 due to weak symbol name and no libc soname
251 436413 Warn about realloc of size zero
252 439685 compiler warning in callgrind/main.c
253 444110 priv/guest_ppc_toIR.c:36198:31: warning: duplicated 'if' condition.
254 444487 hginfo test detects an extra lock inside data symbol "_rtld_local"
255 444488 Use glibc.pthread.stack_cache_size tunable
256 444568 drd/tests/pth_barrier_thr_cr fails on Fedora 38
257 445743 "The impossible happened: mutex is locked simultaneously by two threads"
258 while using mutexes with priority inheritance and signals
259 449309 Missing loopback device ioctl(s)
260 459476 vgdb: allow address reuse to avoid "address already in use" errorsuse" errors
261 460356 s390: Sqrt32Fx4 -- cannot reduce tree
262 462830 WARNING: unhandled amd64-freebsd syscall: 474
263 463027 broken check for MPX instruction support in assembler
264 464103 Enhancement: add a client request to DHAT to mark memory to be histogrammed
265 464476 Firefox fails to start under Valgrind
266 464609 Valgrind memcheck should support Linux pidfd_open
267 464680 Show issues caused by memory policies like selinux deny_execmem
268 464859 Build failures with GCC-13 (drd tsan_unittest)
269 464969 D language demangling
270 465435 m_libcfile.c:66 (vgPlain_safe_fd): Assertion 'newfd >= VG_(fd_hard_limit)' failed.
271 466104 aligned_alloc problems, part 1
272 467036 Add time cost statistics for Regtest
273 467482 Build failure on aarch64 Alpine
274 467714 fdleak_* and rlimit tests fail when parent process has more than
275 64 descriptors opened
276 467839 Gdbserver: Improve compatibility of library directory name
277 468401 [PATCH] Add a style file for clang-format
278 468556 Build failure for vgdb
279 468606 build: remove "Valgrind relies on GCC" check/output
280 469097 ppc64(be) doesn't support SCV syscall instruction
281 n-i-bz FreeBSD rfork syscall fail with EINVAL or ENOSYS rather than VG_(unimplemented)
283 To see details of a given bug, visit
284 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
285 where XXXXXX is the bug number as listed above.
287 * ==================== KNOWN ISSUES ===================
289 * configure --enable-lto=yes is know to not work in all setups.
290 See bug 469049. Workaround: Build without LTO.
292 (3.21.0.RC1: 14 Apr 2023)
293 (3.21.0.RC2: 21 Apr 2023)
295 Release 3.20.0 (24 Oct 2022)
296 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
298 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
299 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
300 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
301 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
302 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
303 AMD64/macOS 10.13 and nanoMIPS/Linux.
305 * ==================== CORE CHANGES ===================
307 * The option "--vgdb-stop-at=event1,event2,..." accepts the new value abexit.
308 This indicates to invoke gdbserver when your program exits abnormally
309 (i.e. with a non zero exit code).
310 * Fix Rust v0 name demangling.
311 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
312 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
313 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with
315 * The option --enable-debuginfod=<no|yes> [default: yes] has been added on
317 * More DWARF5 support as generated by clang14.
319 * ==================== FIXED BUGS ====================
321 The following bugs have been fixed or resolved. Note that "n-i-bz"
322 stands for "not in bugzilla" -- that is, a bug that was reported to us
323 but never got a bugzilla entry. We encourage you to file bugs in
324 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
325 than mailing the developers (or mailing lists) directly -- bugs that
326 are not entered into bugzilla tend to get forgotten about or ignored.
328 131186 writev reports error in (vector[...])
329 434764 iconv_open causes ld.so v2.28+ to use optimised strncmp
330 446754 Improve error codes from alloc functions under memcheck
331 452274 memcheck crashes with Assertion 'sci->status.what == SsIdle' failed
332 452779 Valgrind fails to build on FreeBSD 13.0 with llvm-devel (15.0.0)
333 453055 shared_timed_mutex drd test fails with "Lock shared failed" message
334 453602 Missing command line option to enable/disable debuginfod
335 452802 Handle lld 9+ split RW PT_LOAD segments correctly
336 454040 s390x: False-positive memcheck:cond in memmem on arch13 systems
337 456171 [PATCH] FreeBSD: Don't record address errors when accessing the 'kern.ps_strings' sysctl struct
338 n-i-bz Implement vgdb invoker on FreeBSD
339 458845 PowerPC: The L field for the dcbf and sync instruction should be
341 458915 Remove register cache to fix 458915 gdbserver causes wrong syscall return
342 459031 Documentation on --error-exitcode incomplete
343 459477 XERROR messages lacks ending '\n' in vgdb
344 462007 Implicit int in none/tests/faultstatus.c
346 To see details of a given bug, visit
347 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
348 where XXXXXX is the bug number as listed above.
350 (3.20.0.RC1: 20 Oct 2022)
353 Release 3.19.0 (11 Apr 2022)
354 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
356 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
357 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
358 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
359 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
360 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
361 AMD64/macOS 10.13 and nanoMIPS/Linux.
363 * ==================== CORE CHANGES ===================
365 * Fix Rust v0 name demangling.
366 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
367 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
368 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with HardenedBSD
370 * ================== PLATFORM CHANGES =================
373 - ignore the "v8.x" architecture levels, only look at actual CPU features
374 present. Fixes mismatch detected between RDMA and atomics features
375 preventing startup on some QEMU configurations.
376 - Implement LD{,A}XP and ST{,L}XP
377 - Fix incorrect code emitted for doubleword CAS.
380 - Fix sys_ipc semtimedop syscall
381 - Fix VFLRX and WFLRX instructions
382 - Fix EXRL instruction with negative offset
385 - Reimplement the vbpermq instruction support to generate less Iops and
386 avoid overflowing internal buffers.
387 - Fix checking for scv support to avoid "Facility 'SCV' unavailable (12),
388 exception" messages in dmsg.
389 - Fix setting condition code for Vector Compare quad word instructions.
390 - Fix fix lxsibzx, lxsihzx and lxsihzx instructions so they only load
391 their respective sized data.
392 - Fix the prefixed stq instruction in PC relative mode.
394 * ==================== TOOL CHANGES ===================
397 - Speed up --track-origins=yes for large (in the range of hundreds to
398 thousands of megabytes) mmap/munmaps.
400 - Several fixes for new versions of libstd++ using new posix try_lock
403 * ==================== FIXED BUGS ====================
405 The following bugs have been fixed or resolved. Note that "n-i-bz"
406 stands for "not in bugzilla" -- that is, a bug that was reported to us
407 but never got a bugzilla entry. We encourage you to file bugs in
408 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
409 than mailing the developers (or mailing lists) directly -- bugs that
410 are not entered into bugzilla tend to get forgotten about or ignored.
412 403802 leak_cpp_interior fails with some reachable blocks different than expected
413 435732 memcheck/tests/leak_cpp_interior fails with gcc11
414 444242 s390x: Valgrind crashes on EXRL with negative offset
415 444399 arm64: unhandled instruction 0xC87F2D89 (LD{,A}XP and ST{,L}XP).
417 444481 gdb_server test failures on s390x
418 444495 dhat/tests/copy fails on s390x
419 444552 memcheck/tests/sem fails on s390x with glibc 2.34
420 444571 PPC, fix the lxsibzx and lxsihzx so they only load their respective
422 444836 PPC, pstq instruction for R=1 is not storing to the correct address.
423 444925 fexecve syscall wrapper not properly implemented
424 445032 valgrind/memcheck crash with SIGSEGV when SIGVTALRM timer used and
426 445211 Fix out of tree builds
427 445300 [PATCH] Fix building tests with Musl
428 445011 SIGCHLD is sent when valgrind uses debuginfod-find
429 445354 arm64 backend: incorrect code emitted for doubleword CAS
430 445415 arm64 front end: alignment checks missing for atomic instructions
431 445504 Using C++ condition_variable results in bogus "mutex is locked simultaneously by two threads" warning
432 445607 Unhandled amd64-freebsd syscall: 247
433 445668 Inline stack frame generation is broken for Rust binaries
434 445916 Demangle Rust v0 symbols with .llvm suffix
435 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false positives
436 446138 DRD/Helgrind with std::timed_mutex::try_lock_until false positives
437 446281 Add a DRD suppression for fwrite
438 446103 Memcheck: `--track-origins=yes` causes extreme slowdowns for large mmap/munmap
439 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false
440 446251 TARGET_SIGNAL_THR added to enum target_signal
441 446823 FreeBSD - missing syscalls when using libzm4
442 447991 s390x: Valgrind indicates illegal instruction on wflrx
443 447995 Valgrind segfault on power10 due to hwcap checking code
444 449483 Powerpc: vcmpgtsq., vcmpgtuq,, vcmpequq. instructions not setting the
445 condition code correctly.
446 449672 ppc64 --track-origins=yes failures because of bad cmov addHRegUse
447 449838 sigsegv liburing the 'impossible' happened for io_uring_setup
448 450025 Powerc: ACC file not implemented as a logical overlay of the VSR
450 450437 Warn for execve syscall with argv or argv[0] being NULL
451 450536 Powerpc: valgrind throws 'facility scv unavailable exception'
452 451626 Syscall param bpf(attr->raw_tracepoint.name) points to unaddressable byte(s)
453 451827 [ppc64le] VEX temporary storage exhausted with several vbpermq instructions
454 451843 valgrind fails to start on a FreeBSD system which enforces W^X
456 To see details of a given bug, visit
457 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
458 where XXXXXX is the bug number as listed above.
460 (3.19.0.RC1: 02 Apr 2022)
461 (3.19.0.RC2: 08 Apr 2022)
464 Release 3.18.0 (15 Oct 2021)
465 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
467 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
468 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
469 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
470 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
471 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
472 AMD64/macOS 10.13 and nanoMIPS/Linux.
474 * ==================== CORE CHANGES ===================
476 * The libiberty demangler has been updated, which brings support for
477 Rust v0 name demangling. [Update: alas, due to a bug, this support
478 isn't working in 3.18.0.]
480 * __libc_freeres isn't called anymore after the program recieves a
481 fatal signal. Causing some internal glibc resources to hang around,
482 but preventing any crashes after the program has ended.
484 * The DWARF reader is now very much faster at startup when just
485 --read-inline-info=yes (the default in most cases) is given.
487 * glibc 2.34, which moved various functions from libpthread.so into
488 libc.so, is now supported.
490 * ================== PLATFORM CHANGES =================
494 - v8.2 scalar and vector FABD, FACGE, FACGT and FADD.
495 - v8.2 FP compare & conditional compare instructions.
496 - Zero variants of v8.2 FP compare instructions.
500 - Support the miscellaneous-instruction-extensions facility 3 and
501 the vector-enhancements facility 2. This enables programs
502 compiled with "-march=arch13" or "-march=z15" to be executed
507 - ISA 3.1 support is now complete
508 - ISA 3.0 support for the darn instruction added.
509 - ISA 3.0 support for the vector system call instruction scv added.
510 - ISA 3.0 support for the copy, paste and cpabort instructions added.
512 * Support for X86/FreeBSD and AMD64/FreeBSD has been added.
514 * ==================== OTHER CHANGES ====================
516 * Memcheck on amd64: minor fixes to remove some false positive
519 * ==================== FIXED BUGS ====================
521 The following bugs have been fixed or resolved. Note that "n-i-bz"
522 stands for "not in bugzilla" -- that is, a bug that was reported to us
523 but never got a bugzilla entry. We encourage you to file bugs in
524 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
525 than mailing the developers (or mailing lists) directly -- bugs that
526 are not entered into bugzilla tend to get forgotten about or ignored.
528 208531 [PATCH]: FreeBSD support for valgrind
529 368960 WARNING: unhandled amd64-linux syscall: 163 (acct)
530 407589 [Linux] Add support for C11 aligned_alloc() and GNU reallocarray()
531 423963 Error in child thread when CLONE_PIDFD is used
532 426148 crash with "impossible happened" when running BPF CO-RE programs
533 429375 PPC ISA 3.1 support is missing, part 9
534 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
535 431306 Update demangler to support Rust v0 name mangling
536 432387 s390x: z15 instructions support
537 433437 FreeBSD support, part 1
538 433438 FreeBSD support, part 2
539 433439 FreeBSD support, part 3
540 433469 FreeBSD support, part 4
541 433473 FreeBSD support, part 5
542 433477 FreeBSD support, part 6
543 433479 FreeBSD support, part 7
544 433504 FreeBSD support, part 8
545 433506 FreeBSD support, part 9
546 433507 FreeBSD support, part 10
547 433508 FreeBSD support, part 11
548 433510 FreeBSD support, part 12
549 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
550 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
551 434296 s390x: False-positive memcheck diagnostics from vector string
553 434840 PPC64 darn instruction not supported
554 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
555 435908 valgrind tries to fetch from deubginfod for files which already
556 have debug information
557 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2
558 439046 valgrind is unusably large when linked with lld
559 439090 Implement close_range(2)
560 439326 Valgrind 3.17.0 won't compile with Intel 2021 oneAPI compilers
561 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
562 440670 unhandled ppc64le-linux syscall: 252 statfs64 and 253 fstatfs64
563 440906 Fix impossible constraint issue in P10 testcase.
564 441512 Remove a unneeded / unnecessary prefix check.
565 441534 Update the expected output for test_isa_3_1_VRT.
566 442061 very slow execution under Fedora 34 (readdwarf3)
567 443031 Gcc -many change requires explicit .machine directives
568 443033 Add support for the ISA 3.0 mcrxrx instruction
569 443034 Sraw, srawi, srad, sradi, mfs
570 443178 Powerpc, test jm-mfspr expected output needs to be updated.
571 443179 Need new test for the lxvx and stxvx instructions on ISA 2.07 and
573 443180 The subnormal test and the ISA 3.0 test generate compiler warnings
574 443314 In the latest GIT version, Valgrind with "--trace-flags" crashes
576 443605 Don't call final_tidyup (__libc_freeres) on FatalSignal
578 To see details of a given bug, visit
579 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
580 where XXXXXX is the bug number as listed below.
582 (3.18.0.RC1: 12 Oct 2021)
583 (3.18.0: 15 Oct 2021)
587 Release 3.17.0 (19 Mar 2021)
588 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
590 3.17.0 fixes a number of bugs and adds some functional changes: support for
591 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
592 some new instructions for Arm64, S390 and POWER. There are also some tool
595 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
596 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
597 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
598 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
599 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
601 * ==================== CORE CHANGES ===================
603 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
606 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
607 debugging information. When a debuginfo file cannot be found locally,
608 Valgrind is able to query debuginfod servers for the file using its
609 build-id. See the user manual for more information about debuginfod support.
611 * ================== PLATFORM CHANGES =================
615 - Inaccuracies resulting from double-rounding in the simulation of
616 floating-point multiply-add/subtract instructions have been fixed. These
617 should now behave exactly as the hardware does.
619 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
620 ongoing. Support for the half-word variants of at least the following
621 instructions has been added:
623 FABS <Vd>.<T>, <Vn>.<T>
625 FNEG <Vd>.<T>, <Vn>.<T>
627 FSQRT <Vd>.<T>, <Vn>.<T>
632 - Implement the new instructions/features that were added to z/Architecture
633 with the vector-enhancements facility 1. Also cover the instructions from
634 the vector-packed-decimal facility that are defined outside the chapter
635 "Vector Decimal Instructions", but not the ones from that chapter itself.
637 For a detailed list of newly supported instructions see the updates to
638 `docs/internals/s390-opcodes.csv'.
640 Since the miscellaneous instruction extensions facility 2 was already
641 added in Valgrind 3.16.0, this completes the support necessary to run
642 general programs built with `--march=z14' under Valgrind. The
643 vector-packed-decimal facility is currently not exploited by the standard
644 toolchain and libraries.
648 - Various bug fixes. Fix for the sync field to limit setting just two of
649 the two bits in the L-field. Fix the write size for the stxsibx and
650 stxsihx instructions. Fix the modsw and modsd instructions.
652 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
653 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
654 conversion instructions are still missing.
656 * ==================== TOOL CHANGES ====================
658 * General tool changes
660 - All the tools and their vgpreload libraries are now installed under
661 libexec because they cannot be executed directly and should be run through
662 the valgrind executable. This should be an internal, not user visible,
663 change, but might impact valgrind packagers.
665 - The --track-fds option now respects -q, --quiet and won't output anything
666 if no file descriptors are leaked. It also won't report the standard stdin
667 (0), stdout (1) or stderr (2) descriptors as being leaked with
668 --trace-fds=yes anymore. To track whether the standard file descriptors
669 are still open at the end of the program run use --trace-fds=all.
673 - DHAT has been extended, with two new modes of operation. The new
674 --mode=copy flag triggers copy profiling, which records calls to memcpy,
675 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
676 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
677 the new dhat/dhat.h file. This is useful for learning more about hot code
678 paths. See the user manual for more information about the new modes.
680 - Because of these changes, DHAT's file format has changed. DHAT output
681 files produced with earlier versions of DHAT will not work with this
682 version of DHAT's viewer, and DHAT output files produced with this version
683 of DHAT will not work with earlier versions of DHAT's viewer.
685 * ==================== FIXED BUGS ====================
687 The following bugs have been fixed or resolved. Note that "n-i-bz"
688 stands for "not in bugzilla" -- that is, a bug that was reported to us
689 but never got a bugzilla entry. We encourage you to file bugs in
690 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
691 than mailing the developers (or mailing lists) directly -- bugs that
692 are not entered into bugzilla tend to get forgotten about or ignored.
694 To see details of a given bug, visit
695 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
696 where XXXXXX is the bug number as listed below.
698 140178 open("/proc/self/exe", ...); doesn't quite work
699 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
700 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
701 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
702 345077 linux syscall execveat support (linux 3.19)
703 361770 Missing F_ADD_SEALS
704 369029 handle linux syscalls sched_getattr and sched_setattr
705 384729 __libc_freeres inhibits cross-platform valgrind
706 388787 Support for C++17 new/delete
707 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
708 396656 Warnings while reading debug info
709 397605 ioctl FICLONE mishandled
710 401416 Compile failure with openmpi 4.0
711 408663 Suppression file for musl libc
712 404076 s390x: z14 vector instructions not implemented
713 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
714 (actually affected all x86 and nanomips regardless of host bitness)
715 413547 regression test does not check for Arm 64 features.
716 414268 Enable AArch64 feature detection and decoding for v8.x instructions
717 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
718 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
719 422261 platform selection fails for unqualified client name
720 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
721 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
722 423195 PPC ISA 3.1 support is missing, part 1
723 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
724 424012 crash with readv/writev having invalid but not NULL arg2 iovec
725 424298 amd64: Implement RDSEED
726 425232 PPC ISA 3.1 support is missing, part 2
727 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
728 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
729 426123 PPC ISA 3.1 support is missing, part 3
730 426144 Fix "condition variable has not been initialized" on Fedora 33.
731 427400 PPC ISA 3.1 support is missing, part 4
732 427401 PPC ISA 3.1 support is missing, part 5
733 427404 PPC ISA 3.1 support is missing, part 6
734 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
735 427787 Support new faccessat2 linux syscall (439)
736 427969 debuginfo section duplicates a section in the main ELF file
737 428035 drd: Unbreak the musl build
738 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
739 428716 cppcheck detects potential leak in VEX/useful/smchash.c
740 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
741 429352 PPC ISA 3.1 support is missing, part 7
742 429354 PPC ISA 3.1 support is missing, part 8
743 429692 unhandled ppc64le-linux syscall: 147 (getsid)
744 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
746 429952 Errors when building regtest with clang
747 430354 ppc stxsibx and stxsihx instructions write too much data
748 430429 valgrind.h doesn't compile on s390x with clang
749 430485 expr_is_guardable doesn't handle Iex_Qop
750 431556 Complete arm64 FADDP v8.2 instruction support
751 432102 Add support for DWARF5 as produced by GCC11
752 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
753 432381 drd: Process STACK_REGISTER client requests
754 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
755 432672 vg_regtest: test-specific environment variables not reset between tests
756 432809 VEX should support REX.W + POPF
757 432861 PPC modsw and modsd give incorrect results for 1 mod 12
758 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
759 432215 Add debuginfod functionality
760 433323 Use pkglibexecdir as vglibdir
761 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
762 433629 valgrind/README has type "abd" instead of "and"
763 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
764 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
765 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
766 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
767 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
769 (3.17.0.RC1: 13 Mar 2021)
770 (3.17.0.RC2: 17 Mar 2021)
771 (3.17.0: 19 Mar 2021)
775 Release 3.16.1 (22 June 2020)
776 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
778 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
779 fixes character encoding problems in the documentation HTML.
781 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
782 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
784 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
788 Release 3.16.0 (27 May 2020)
789 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
791 3.16.0 is a feature release with many improvements and the usual collection of
794 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
795 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
796 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
797 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
798 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
800 * ==================== CORE CHANGES ===================
802 * It is now possible to dynamically change the value of many command line
803 options while your program (or its children) are running under Valgrind.
805 To see the list of dynamically changeable options, run
806 "valgrind --help-dyn-options".
808 You can change the options from the shell by using vgdb to launch
809 the monitor command "v.clo <clo option>...".
810 The same monitor command can be used from a gdb connected
811 to the valgrind gdbserver.
812 Your program can also change the dynamically changeable options using
813 the client request VALGRIND_CLO_CHANGE(option).
815 * ================== PLATFORM CHANGES =================
817 * MIPS: preliminary support for nanoMIPS instruction set has been added.
819 * ==================== TOOL CHANGES ====================
823 - The implicit memcpy done by each call to realloc now counts towards the
824 read and write counts of resized heap blocks, making those counts higher
829 - cg_annotate's --auto and --show-percs options now default to 'yes', because
830 they are usually wanted.
834 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
835 because they are usually wanted.
837 - The command option --collect-systime has been enhanced to specify
838 the unit used to record the elapsed time spent during system calls.
839 The command option now accepts the values no|yes|msec|usec|nsec,
840 where yes is a synonym of msec. When giving the value nsec, the
841 system cpu time of system calls is also recorded.
845 - Several memcheck options are now dynamically changeable.
846 Use valgrind --help-dyn-options to list them.
848 - The release 3.15 introduced a backward incompatible change for
849 some suppression entries related to preadv and pwritev syscalls.
850 When reading a suppression entry using the unsupported 3.14 format,
851 valgrind will now produce a warning to say the suppression entry will not
852 work, and suggest the needed change.
854 - Significantly fewer false positive errors on optimised code generated by
855 Clang and GCC. In particular, Memcheck now deals better with the
856 situation where the compiler will transform C-level "A && B" into "B && A"
857 under certain circumstances (in which the transformation is valid).
858 Handling of integer equality/non-equality checks on partially defined
859 values is also improved on some architectures.
863 - The exprimental Stack and Global Array Checking tool has been removed.
864 It only ever worked on x86 and amd64, and even on those it had a
865 high false positive rate and was slow. An alternative for detecting
866 stack and global array overruns is using the AddressSanitizer (ASAN)
867 facility of the GCC and Clang compilers, which require you to rebuild
868 your code with -fsanitize=address.
870 * ==================== OTHER CHANGES ====================
872 * New and modified GDB server monitor features:
874 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
876 - The gdbserver monitor commands that require an address and an optional
877 length argument now accepts the alternate 'C like' syntax "address[length]".
878 For example, the memcheck command "monitor who_points_at 0x12345678 120"
879 can now also be given as "monitor who_points_at 0x12345678[120]".
881 * ==================== FIXED BUGS ====================
883 The following bugs have been fixed or resolved. Note that "n-i-bz"
884 stands for "not in bugzilla" -- that is, a bug that was reported to us
885 but never got a bugzilla entry. We encourage you to file bugs in
886 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
887 than mailing the developers (or mailing lists) directly -- bugs that
888 are not entered into bugzilla tend to get forgotten about or ignored.
890 To see details of a given bug, visit
891 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
892 where XXXXXX is the bug number as listed below.
894 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
895 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
896 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
897 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
898 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
899 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
900 400872 Add nanoMIPS support to Valgrind
901 403212 drd/tests/trylock hangs on FreeBSD
902 404406 s390x: z14 miscellaneous instructions not implemented
903 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
904 406561 mcinfcallWSRU gdbserver_test fails on ppc64
905 406824 Unsupported baseline
906 407218 Add support for the copy_file_range syscall
907 407307 Intercept stpcpy also in ld.so for arm64
908 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
910 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
911 408009 Expose rdrand and f16c even on avx if host cpu supports them
912 408091 Missing pkey syscalls
913 408414 Add support for missing for preadv2 and pwritev2 syscalls
914 409141 Valgrind hangs when SIGKILLed
915 409206 Support for Linux PPS and PTP ioctls
916 409367 exit_group() after signal to thread waiting in futex() causes hangs
917 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
918 409780 References to non-existent configure.in
919 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
920 410599 Non-deterministic behaviour of pth_self_kill_15_other test
921 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
922 411134 Allow the user to change a set of command line options during execution
923 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
924 412344 Problem setting mips flags with specific paths
925 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
926 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
927 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
928 413603 callgrind_annotate/cg_annotate truncate function names at '#'
929 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
930 415136 ARMv8.1 Compare-and-Swap instructions are not supported
931 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
932 416239 valgrind crashes when handling clock_adjtime
933 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
934 416286 DRD reports "conflicting load" error on std::mutex::lock()
935 416301 s390x: "compare and signal" not supported
936 416387 finit_module and bpf syscalls are unhandled on arm64
937 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
938 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
939 416753 new 32bit time syscalls for 2038+
940 417075 pwritev(vector[...]) suppression ignored
941 417075 is not fixed, but incompatible supp entries are detected
942 and a warning is produced for these.
943 417187 [MIPS] Conditional branch problem since 'grail' changes
944 417238 Test memcheck/tests/vbit-test fails on mips64 BE
945 417266 Make memcheck/tests/linux/sigqueue usable with musl
946 417281 s390x: /bin/true segfaults with "grail" enabled
947 417427 commit to fix vki_siginfo_t definition created numerous regression
949 417452 s390_insn_store_emit: dst->tag for HRcVec128
950 417578 Add suppressions for glibc DTV leaks
951 417906 clone with CLONE_VFORK and no CLONE_VM fails
952 418004 Grail code additions break ppc64.
953 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
954 418997 s390x: Support Iex_ITE for float and vector types
955 419503 s390x: Avoid modifying registers returned from isel functions
956 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
957 421570 std_mutex fails on Arm v8.1 h/w
958 434035 vgdb might crash if valgrind is killed
959 n-i-bz Fix minor one time leaks in dhat.
960 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
961 n-i-bz Add support for the Linux io_uring system calls
962 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
963 n-i-bz Fix non-glibc build of test suite with s390x_features
964 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
965 423195 PPC ISA 3.1 support is missing, part 1
967 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
968 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
969 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
973 Release 3.15.0 (12 April 2019)
974 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
976 3.15.0 is a feature release with many improvements and the usual collection of
979 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
980 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
981 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
982 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
983 support for X86/macOS 10.13 and AMD64/macOS 10.13.
985 * ==================== CORE CHANGES ===================
987 * The XTree Massif output format now makes use of the information obtained
988 when specifying --read-inline-info=yes.
990 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
992 * ==================== TOOL CHANGES ====================
996 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
997 it has been promoted from an experimental tool to a regular tool. Run it
998 with --tool=dhat instead of --tool=exp-dhat.
1000 - DHAT now prints only minimal data when the program ends, instead writing
1001 the bulk of the profiling data to a file. As a result, the --show-top-n
1002 and --sort-by options have been removed.
1004 - Profile results can be viewed with the new viewer, dh_view.html. When
1005 a run ends, a short message is printed, explaining how to view the result.
1007 - See the documentation for more details.
1011 - cg_annotate has a new option, --show-percs, which prints percentages next
1012 to all event counts.
1016 - callgrind_annotate has a new option, --show-percs, which prints percentages
1017 next to all event counts.
1019 - callgrind_annotate now inserts commas in call counts, and
1020 sort the caller/callee lists in the call tree.
1024 - The default value for --read-inline-info is now "yes" on
1025 Linux/Android/Solaris. It is still "no" on other OS.
1029 - The option --xtree-leak=yes (to output leak result in xtree format)
1030 automatically activates the option --show-leak-kinds=all, as xtree
1031 visualisation tools such as kcachegrind can in any case select what kind
1032 of leak to visualise.
1034 - There has been further work to avoid false positives. In particular,
1035 integer equality on partially defined inputs (C == and !=) is now handled
1038 * ==================== OTHER CHANGES ====================
1040 * The new option --show-error-list=no|yes displays, at the end of the run, the
1041 list of detected errors and the used suppressions. Prior to this change,
1042 showing this information could only be done by specifying "-v -v", but that
1043 also produced a lot of other possibly-non-useful messages. The option -s is
1044 equivalent to --show-error-list=yes.
1046 * ==================== FIXED BUGS ====================
1048 The following bugs have been fixed or resolved. Note that "n-i-bz"
1049 stands for "not in bugzilla" -- that is, a bug that was reported to us
1050 but never got a bugzilla entry. We encourage you to file bugs in
1051 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1052 than mailing the developers (or mailing lists) directly -- bugs that
1053 are not entered into bugzilla tend to get forgotten about or ignored.
1055 To see details of a given bug, visit
1056 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1057 where XXXXXX is the bug number as listed below.
1059 385411 s390x: z13 vector floating-point instructions not implemented
1060 397187 z13 vector register support for vgdb gdbserver
1061 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
1062 398870 Please add support for instruction vcvtps2ph
1063 399287 amd64 front end: Illegal Instruction vcmptrueps
1064 399301 Use inlined frames in Massif XTree output.
1065 399322 Improve callgrind_annotate output
1066 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
1067 400164 helgrind test encounters mips x-compiler warnings and assembler error
1068 400490 s390x: VRs allocated as if separate from FPRs
1069 400491 s390x: Operand of LOCH treated as unsigned integer
1070 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
1071 options, which specify a mips64 processor
1072 401112 LLVM 5.0 generates comparison against partially initialized data
1073 401277 More bugs in z13 support
1074 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
1075 401578 drd: crashes sometimes on fork()
1076 401627 memcheck errors with glibc avx2 optimized wcsncmp
1077 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
1078 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
1079 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
1081 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
1082 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
1083 402123 invalid assembler opcodes for mips32r2
1084 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
1085 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
1086 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
1087 402351 mips64 libvexmultiarch_test fails on s390x
1088 402369 Overhaul DHAT
1089 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
1090 402480 Do not use %rsp in clobber list
1091 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
1092 402515 Implement new option --show-error-list=no|yes / -s
1093 402519 POWER 3.0 addex instruction incorrectly implemented
1094 402781 Redo the cache used to process indirect branch targets
1095 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
1096 403552 s390x: wrong facility bit checked for vector facility
1097 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
1098 404638 Add VG_(replaceIndexXA)
1099 404843 s390x: backtrace sometimes ends prematurely
1100 404888 autotools cleanup series
1101 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
1102 405182 Valgrind fails to build with Clang
1103 405205 filter_libc: remove the line holding the futex syscall error entirely
1104 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
1105 the upper and lower 32-bits of the 64-bit result
1106 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
1107 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
1108 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
1110 405403 s390x disassembler cannot be used on x86
1111 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
1112 405458 MIPS mkFormVEC arguments swapped?
1113 405716 drd: Fix an integer overflow in the stack margin calculation
1114 405722 Support arm64 core dump
1115 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
1116 of the 64-bit destination field.
1117 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
1118 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
1119 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
1121 406256 PPC64, vector floating point instructions don't handle subnormal
1122 according to VSCR[NJ] bit setting.
1123 406352 cachegrind/callgrind fails ann tests because of missing a.c
1124 406354 dhat is broken on x86 (32bit)
1125 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
1126 406357 gdbserver_tests fails because of gdb output change
1127 406360 memcheck/tests/libstdc++.supp needs more supression variants
1128 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
1129 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
1130 407340 PPC64, does not support the vlogefp, vexptefp instructions.
1131 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
1132 n-i-bz Fix callgrind_annotate non deterministic order for equal total
1133 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
1134 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
1135 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
1137 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
1138 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
1139 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
1143 Release 3.14.0 (9 October 2018)
1144 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1146 3.14.0 is a feature release with many improvements and the usual collection of
1149 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1150 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1151 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1152 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
1153 support for X86/macOS 10.13, AMD64/macOS 10.13.
1155 * ==================== CORE CHANGES ===================
1157 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
1158 debug info for unloaded code. This allows saved stack traces (e.g. for
1159 memory leaks) to include file/line info for code that has been dlclose'd (or
1160 similar). See the user manual for more information and known limitations.
1162 * Ability to specify suppressions based on source file name and line number.
1164 * Majorly overhauled register allocator. No end-user changes, but the JIT
1165 generates code a bit more quickly now.
1167 * ================== PLATFORM CHANGES =================
1169 * Preliminary support for macOS 10.13 has been added.
1171 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
1173 * mips: support for MIPS SIMD architecture (MSA) has been added.
1175 * mips: support for MIPS N32 ABI has been added.
1177 * s390: partial support for vector instructions (integer and string) has been
1180 * ==================== TOOL CHANGES ====================
1182 * Helgrind: Addition of a flag
1183 --delta-stacktrace=no|yes [yes on linux amd64/x86]
1184 which specifies how full history stack traces should be computed.
1185 Setting this to =yes can speed up Helgrind by 25% when using
1186 --history-level=full.
1188 * Memcheck: reduced false positive rate for optimised code created by Clang 6
1189 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
1190 blocks more carefully to determine where it can avoid expensive definedness
1191 checks without loss of precision. This is controlled by the flag
1192 --expensive-definedness-checks=no|auto|yes [auto].
1194 * ==================== OTHER CHANGES ====================
1196 * Valgrind is now buildable with link-time optimisation (LTO). A new
1197 configure option --enable-lto=yes allows building Valgrind with LTO. If the
1198 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
1199 Note that if you are doing Valgrind development, --enable-lto=yes massively
1200 slows down the build process.
1202 * ==================== FIXED BUGS ====================
1204 The following bugs have been fixed or resolved. Note that "n-i-bz"
1205 stands for "not in bugzilla" -- that is, a bug that was reported to us
1206 but never got a bugzilla entry. We encourage you to file bugs in
1207 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1208 than mailing the developers (or mailing lists) directly -- bugs that
1209 are not entered into bugzilla tend to get forgotten about or ignored.
1211 To see details of a given bug, visit
1212 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1213 where XXXXXX is the bug number as listed below.
1215 79362 Debug info is lost for .so files when they are dlclose'd
1216 208052 strlcpy error when n = 0
1217 255603 exp-sgcheck Assertion '!already_present' failed
1218 338252 building valgrind with -flto (link time optimisation) fails
1219 345763 MIPS N32 ABI support
1220 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
1221 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
1222 372347 Replacement problem of the additional c++14/c++17 new/delete operators
1223 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1224 376257 helgrind history full speed up using a cached stack
1225 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
1227 379748 Fix missing pselect syscall (OS X 10.11)
1228 379754 Fix missing syscall ulock_wait (OS X 10.12)
1229 380397 s390x: __GI_strcspn() replacemenet needed
1230 381162 possible array overrun in VEX register allocator
1231 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
1232 381274 powerpc too chatty even with --sigill-diagnostics=no
1233 381289 epoll_pwait can have a NULL sigmask
1234 381553 VEX register allocator v3
1235 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
1236 381769 Use ucontext_t instead of struct ucontext
1237 381805 arm32 needs ld.so index hardwire for new glibc security fixes
1238 382256 gz compiler flag test doesn't work for gold
1239 382407 vg_perf needs "--terse" command line option
1240 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
1241 382563 MIPS MSA ASE support
1242 382998 xml-socket doesn't work
1243 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
1244 383723 Fix missing kevent_qos syscall (macOS 10.11)
1245 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
1246 384096 Mention AddrCheck at Memcheck's command line option [..]
1247 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
1248 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
1249 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
1250 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
1251 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
1252 384337 performance improvements to VEX register allocator v2 and v3
1253 384526 reduce number of spill insns generated by VEX register allocator v3
1254 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
1255 384631 Sanitise client args as printed with -v
1256 384633 Add a simple progress-reporting facility
1257 384987 VEX regalloc: allocate caller-save registers for short lived vregs
1258 385055 PPC VEX temporary storage exhausted
1259 385182 PPC64 is missing support for the DSCR
1260 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
1261 385207 PPC64, generate_store_FPRF() generates too many Iops
1262 385208 PPC64, xxperm instruction exhausts temporary memory
1263 385210 PPC64, vpermr instruction could exhaust temporary memory
1264 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
1265 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
1266 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
1267 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
1268 385408 s390x: z13 vector "support" instructions not implemented
1269 385409 s390x: z13 vector integer instructions not implemented
1270 385410 s390x: z13 vector string instructions not implemented
1271 385412 s390x: new non-vector z13 instructions not implemented
1272 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
1273 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
1274 385939 Optionally exit on the first error
1275 386318 valgrind.org/info/tools.html is missing SGCheck
1276 386425 running valgrind + wine on armv7l gives illegal opcode
1277 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
1278 387410 MIPSr6 support
1279 387664 Memcheck: make expensive-definedness-checks be the default
1280 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
1281 387766 asm shifts cause false positive "Conditional jump or move depends
1282 on uninitialised value"
1283 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
1284 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
1285 388786 Support bpf syscall in amd64 Linux
1286 388862 Add replacements for wmemchr and wcsnlen on Linux
1287 389065 valgrind meets gcc flag -Wlogical-op
1288 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
1289 390471 suppression by specification of source-file line number
1290 390723 make xtree dump files world wide readable, similar to log files
1291 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
1292 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
1293 392118 unhandled amd64-linux syscall: 332 (statx)
1294 392449 callgrind not clearing the number of calls properly
1295 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
1296 lxssp, stxssp and stxvl instructions.
1297 393023 callgrind_control risks using the wrong vgdb
1298 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
1299 393099 posix_memalign() invalid write if alignment == 0
1300 393146 failing assert "is_DebugInfo_active(di)"
1301 395709 PPC64 is missing support for the xvnegsp instruction
1302 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
1304 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
1305 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
1306 396839 s390x: Trap instructions not implemented
1307 396887 arch_prctl should return EINVAL on unknown option
1308 == 397286 crash before launching binary (Unsupported arch_prctl option)
1309 == 397393 valgrind: the 'impossible' happened: (Archlinux)
1310 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
1311 396906 compile tests failure on mips32-linux: broken inline asm in tests on
1313 397012 glibc ld.so uses arch_prctl on i386
1314 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
1315 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
1316 397424 glibc 2.27 and gdb_server tests
1317 398028 Assertion `cfsi_fits` failing in simple C program
1318 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
1320 n-i-bz Fix missing workq_ops operations (macOS)
1321 n-i-bz fix bug in strspn replacement
1322 n-i-bz Add support for the Linux BLKFLSBUF ioctl
1323 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
1324 n-i-bz Fix possible stack trashing by semctl syscall wrapping
1325 n-i-bz Add support for the Linux membarrier() system call
1326 n-i-bz x86 front end: recognise and handle UD2 correctly
1327 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
1328 correctly aligned before entering the handler.
1330 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
1331 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
1332 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
1336 Release 3.13.0 (15 June 2017)
1337 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1339 3.13.0 is a feature release with many improvements and the usual collection of
1342 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1343 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1344 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1345 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
1347 * ==================== CORE CHANGES ===================
1349 * The translation cache size has been increased to keep up with the demands of
1350 large applications. The maximum number of sectors has increased from 24 to
1351 48. The default number of sectors has increased from 16 to 32 on all
1352 targets except Android, where the increase is from 6 to 12.
1354 * The amount of memory that Valgrind can use has been increased from 64GB to
1355 128GB. In particular this means your application can allocate up to about
1356 60GB when running on Memcheck.
1358 * Valgrind's default load address has been changed from 0x3800'0000 to
1359 0x5800'0000, so as to make it possible to load larger executables. This
1360 should make it possible to load executables of size at least 1200MB.
1362 * A massive spaceleak caused by reading compressed debuginfo files has been
1363 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
1366 * The C++ demangler has been updated.
1368 * Support for demangling Rust symbols has been added.
1370 * A new representation of stack traces, the "XTree", has been added. An XTree
1371 is a tree of stacktraces with data associated with the stacktraces. This is
1372 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
1373 consumption of your program. Reporting is controlled by the new options
1374 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
1376 A report can also be produced on demand using the gdbserver monitor command
1377 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
1378 format' and 'massif format. The existing visualisers for these formats (e.g.
1379 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
1380 analyse these reports.
1382 Memcheck can also produce XTree leak reports using the Callgrind file
1383 format. For more details, see the user manual.
1385 * ================== PLATFORM CHANGES =================
1387 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
1389 * amd64: fixes for JIT failure problems on long AVX2 code blocks
1391 * amd64 and x86: support for CET prefixes has been added
1393 * arm32: a few missing ARMv8 instructions have been implemented
1395 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
1396 Store-Conditional instructions has been added. This is to deal with
1397 processor implementations that implement the LL/SC specifications strictly
1398 and as a result cause Valgrind to hang in certain situations. The
1399 alternative implementation is automatically enabled at startup, as required.
1400 You can use the option --sim-hints=fallback-llsc to force-enable it if you
1403 * Support for OSX 10.12 has been improved.
1405 * On Linux, clone handling has been improved to honour CLONE_VFORK that
1406 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
1407 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
1408 CLONE_VM exact semantics will (still) not work.
1410 * The TileGX/Linux port has been removed because it appears to be both unused
1413 * ==================== TOOL CHANGES ====================
1417 - Memcheck should give fewer false positives when running optimised
1418 Clang/LLVM generated code.
1420 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1422 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
1423 to produce the end of execution leak report in a xtree callgrind format
1426 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
1427 the leak report in an xtree file.
1431 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1433 - For some workloads (typically, for big applications), Massif memory
1434 consumption and CPU consumption has decreased significantly.
1438 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1440 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
1441 for Ada gnat compiled applications.
1443 * ==================== OTHER CHANGES ====================
1445 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
1446 append the inner guest stacktrace to the inner host stacktrace. This helps
1447 to investigate the errors reported by the outer, when they are caused by the
1448 inner guest program (such as an inner regtest). See README_DEVELOPERS for
1451 * To allow fast detection of callgrind files by desktop environments and file
1452 managers, the format was extended to have an optional first line that
1453 uniquely identifies the format ("# callgrind format"). Callgrind creates
1454 this line now, as does the new xtree functionality.
1456 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
1457 have a new %n format letter that is replaced by a sequence number.
1459 * "--version -v" now shows the SVN revision numbers from which Valgrind was
1462 * ==================== FIXED BUGS ====================
1464 The following bugs have been fixed or resolved. Note that "n-i-bz"
1465 stands for "not in bugzilla" -- that is, a bug that was reported to us
1466 but never got a bugzilla entry. We encourage you to file bugs in
1467 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1468 than mailing the developers (or mailing lists) directly -- bugs that
1469 are not entered into bugzilla tend to get forgotten about or ignored.
1471 To see details of a given bug, visit
1472 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1473 where XXXXXX is the bug number as listed below.
1475 162848 --log-file output isn't split when a program forks
1476 340777 Illegal instruction on mips (ar71xx)
1477 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
1478 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
1479 to a different stack.
1480 344139 x86 stack-seg overrides, needed by the Wine people
1481 344524 store conditional of guest applications always fail - observed on
1483 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
1484 352395 Please provide SVN revision info in --version -v
1485 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
1486 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
1487 INVALID_POSIX_THREADID' failed
1488 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
1489 barrier implementation
1490 358697 valgrind.h: Some code remains even when defining NVALGRIND
1491 359202 Add musl libc configure/compile
1492 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
1493 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
1494 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
1495 362223 assertion failed when .valgrindrc is a directory instead of a file
1496 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
1497 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
1498 368507 can't malloc chunks larger than about 34GB
1499 368529 Android arm target link error, missing atexit and pthread_atfork
1500 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
1501 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
1502 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
1503 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
1504 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
1505 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
1506 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
1507 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
1508 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
1509 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
1510 370028 Reduce the number of compiler warnings on MIPS platforms
1511 370635 arm64 missing syscall getcpu
1512 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
1513 371227 Clean AArch64 syscall table
1514 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
1515 371471 Valgrind complains about non legit memory leaks on placement new (C++)
1516 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
1517 371503 disInstr(arm64): unhandled instruction 0xF89F0000
1518 371869 support '%' in symbol Z-encoding
1519 371916 execution tree xtree concept
1520 372120 c++ demangler demangles symbols which are not c++
1521 372185 Support of valgrind on ARMv8 with 32 bit executable
1522 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1523 372195 Power PC, xxsel instruction is not always recognized.
1524 372504 Hanging on exit_group
1525 372600 process loops forever when fatal signals are arriving quickly
1526 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1527 373046 Stacks registered by core are never deregistered
1528 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1529 373086 Implement additional Xen hypercalls
1530 373192 Calling posix_spawn in glibc 2.24 completely broken
1531 373488 Support for fanotify API on ARM64 architecture
1532 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1533 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1534 373938 const IRExpr arguments for matchIRExpr()
1535 374719 some spelling fixes
1536 374963 increase valgrind's load address to prevent mmap failure
1537 375514 valgrind_get_tls_addr() does not work in case of static TLS
1538 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1539 for ML_(find_rx_mapping)()
1540 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1541 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1542 == 377159 "vex: the `impossible' happened" still present
1543 == 375150 Assertion 'tres.status == VexTransOK' failed
1544 == 378068 valgrind crashes on AVX2 function in FFmpeg
1545 376142 Segfaults on MIPS Cavium Octeon boards
1546 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1547 376455 Solaris: unhandled syscall lgrpsys(180)
1548 376518 Solaris: unhandled fast trap getlgrp(6)
1549 376611 ppc64 and arm64 don't know about prlimit64 syscall
1550 376729 PPC64, remove R2 from the clobber list
1552 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1553 to be wrongly marked as addressable
1554 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1555 PIE enabled by default
1556 377376 memcheck/tests/linux/getregset fails with glibc2.24
1557 377427 PPC64, lxv instruction failing on odd destination register
1558 377478 PPC64: ISA 3.0 setup fixes
1559 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1560 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1561 and 2 args for FUTEX_TRYLOCK_PI
1562 377717 Fix massive space leak when reading compressed debuginfo sections
1563 377891 Update Xen 4.6 domctl wrappers
1564 377930 fcntl syscall wrapper is missing flock structure check
1565 378524 libvexmultiarch_test regression on s390x and ppc64
1566 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1567 378673 Update libiberty demangler
1568 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1569 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1570 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1571 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1572 (task_register_dyld_image_infos)
1573 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1574 (task_register_dyld_shared_cache_image_info)
1575 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1576 379473 MIPS: add support for rdhwr cycle counter register
1577 379504 remove TileGX/Linux port
1578 379525 Support more x86 nop opcodes
1579 379838 disAMode(x86): not an addr!
1580 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1581 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1582 379895 clock_gettime does not execute POST syscall wrapper
1583 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1584 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1585 380200 xtree generated callgrind files refer to files without directory name
1586 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1587 380397 s390x: __GI_strcspn() replacement needed
1588 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1590 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1591 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1592 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1596 Release 3.12.0 (20 October 2016)
1597 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1599 3.12.0 is a feature release with many improvements and the usual
1600 collection of bug fixes.
1602 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1603 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1604 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1605 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1606 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1607 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1609 * ================== PLATFORM CHANGES =================
1611 * POWER: Support for ISA 3.0 has been added
1613 * mips: support for O32 FPXX ABI has been added.
1614 * mips: improved recognition of different processors
1615 * mips: determination of page size now done at run time
1617 * amd64: Partial support for AMD FMA4 instructions.
1619 * arm, arm64: Support for v8 crypto and CRC instructions.
1621 * Improvements and robustification of the Solaris port.
1623 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1625 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1626 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1627 where possible. Valgrind's support for 32-bit x86 has stagnated in
1628 recent years and has fallen far behind that for 64-bit x86
1629 instructions. By contrast 64-bit x86 is well supported, up to and
1632 * ==================== TOOL CHANGES ====================
1636 - Added meta mempool support for describing a custom allocator which:
1637 - Auto-frees all chunks assuming that destroying a pool destroys all
1639 - Uses itself to allocate other memory blocks
1641 - New flag --ignore-range-below-sp to ignore memory accesses below
1642 the stack pointer, if you really have to. The related flag
1643 --workaround-gcc296-bugs=yes is now deprecated. Use
1644 --ignore-range-below-sp=1024-1 as a replacement.
1648 - Improved thread startup time significantly on non-Linux platforms.
1652 - Added collection of the metric "tot-blocks-allocd"
1654 * ==================== OTHER CHANGES ====================
1656 * Replacement/wrapping of malloc/new related functions is now done not just
1657 for system libraries by default, but for any globally defined malloc/new
1658 related function (both in shared libraries and statically linked alternative
1659 malloc implementations). The dynamic (runtime) linker is excluded, though.
1660 To only intercept malloc/new related functions in
1661 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1662 "nouserintercepts" can be any non-existing library name).
1663 This new functionality is not implemented for MacOS X.
1665 * The maximum number of callers in a suppression entry is now equal to
1666 the maximum size for --num-callers (500).
1667 Note that --gen-suppressions=yes|all similarly generates suppressions
1668 containing up to --num-callers frames.
1670 * New and modified GDB server monitor features:
1672 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1673 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1676 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1677 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1680 * Valgrind is able to read compressed debuginfo sections in two formats:
1681 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1682 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1684 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1685 for the most common use case (x86_64-linux, Memcheck) has been
1688 * Improved performance for programs that do a lot of discarding of
1689 instruction address ranges of 8KB or less.
1691 * The C++ symbol demangler has been updated.
1693 * More robustness against invalid syscall parameters on Linux.
1695 * ==================== FIXED BUGS ====================
1697 The following bugs have been fixed or resolved. Note that "n-i-bz"
1698 stands for "not in bugzilla" -- that is, a bug that was reported to us
1699 but never got a bugzilla entry. We encourage you to file bugs in
1700 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1701 than mailing the developers (or mailing lists) directly -- bugs that
1702 are not entered into bugzilla tend to get forgotten about or ignored.
1704 To see details of a given bug, visit
1705 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1706 where XXXXXX is the bug number as listed below.
1708 191069 Exiting due to signal not reported in XML output
1709 199468 Suppressions: stack size limited to 25
1710 while --num-callers allows more frames
1711 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1712 278744 cvtps2pd with redundant RexW
1713 303877 valgrind doesn't support compressed debuginfo sections.
1714 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1715 348345 Assertion fails for negative lineno
1716 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1717 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1718 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1719 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1720 352197 mips: mmap2() not wrapped correctly for page size > 4096
1721 353083 arm64 doesn't implement various xattr system calls
1722 353084 arm64 doesn't support sigpending system call
1723 353137 www: update info for Supported Platforms
1724 353138 www: update "The Valgrind Developers" page
1725 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1728 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1729 353398 WARNING: unhandled amd64-solaris syscall: 207
1730 353660 XML in auxwhat tag not escaping reserved symbols properly
1731 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1732 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1733 353802 ELF debug info reader confused with multiple .rodata sections
1734 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1735 353917 unhandled amd64-solaris syscall fchdir(120)
1736 353920 unhandled amd64-solaris syscall: 170
1737 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1738 354392 unhandled amd64-solaris syscall: 171
1739 354797 Vbit test does not include Iops for Power 8 instruction support
1740 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1744 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1745 355188 valgrind should intercept all malloc related global functions
1746 355454 do not intercept malloc related symbols from the runtime linker
1747 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1748 356044 Dwarf line info reader misinterprets is_stmt register
1749 356112 mips: replace addi with addiu
1750 356393 valgrind (vex) crashes because isZeroU happened
1753 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1754 356678 arm64-linux: unhandled syscall 232 (mincore)
1755 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1756 356823 Unsupported ARM instruction: stlex
1757 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1758 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1759 357673 crash if I try to run valgrind with a binary link with libcurl
1760 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1761 357871 pthread_spin_destroy not properly wrapped
1762 357887 Calls to VG_(fclose) do not close the file descriptor
1763 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1764 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1765 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1766 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1767 359181 Buffer Overflow during Demangling
1768 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1769 359289 s390x: popcnt (B9E1) not implemented
1770 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1771 359503 Add missing syscalls for aarch64 (arm64)
1772 359645 "You need libc6-dbg" help message could be more helpful
1773 359703 s390: wire up separate socketcalls system calls
1774 359724 getsockname might crash - deref_UInt should call safe_to_deref
1775 359733 amd64 implement ld.so strchr/index override like x86
1776 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1777 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1779 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1780 359871 Incorrect mask handling in ppoll
1781 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1782 360008 Contents of Power vr registers contents is not printed correctly when
1783 the --vgdb-shadow-registers=yes option is used
1784 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1785 non-zero shadow bits
1786 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1787 360425 arm64 unsupported instruction ldpsw
1789 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1790 360571 Error about the Android Runtime reading below the stack pointer on ARM
1791 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1792 360749 kludge for multiple .rodata sections on Solaris no longer needed
1793 360752 raise the number of reserved fds in m_main.c from 10 to 12
1794 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1795 361226 s390x: risbgn (EC59) not implemented
1796 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1797 361354 ppc64[le]: wire up separate socketcalls system calls
1798 361615 Inconsistent termination for multithreaded process terminated by signal
1799 361926 Unhandled Solaris syscall: sysfs(84)
1800 362009 V dumps core on unimplemented functionality before threads are created
1801 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1802 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1803 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1804 362953 Request for an update to the Valgrind Developers page
1805 363680 add renameat2() support
1806 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1807 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1808 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1809 364058 clarify in manual limitations of array overruns detections
1810 364413 pselect sycallwrapper mishandles NULL sigmask
1811 364728 Power PC, missing support for several HW registers in
1812 get_otrack_shadow_offset_wrk()
1813 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1814 365273 Invalid write to stack location reported after signal handler runs
1815 365912 ppc64BE segfault during jm-insns test (RELRO)
1816 366079 FPXX Support for MIPS32 Valgrind
1817 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1818 366344 Multiple unhandled instruction for Aarch64
1819 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1820 367995 Integration of memcheck with custom memory allocator
1821 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1822 368412 False positive result for altivec capability check
1823 368416 Add tc06_two_races_xml.exp output for ppc64
1824 368419 Perf Events ioctls not implemented
1825 368461 mmapunmap test fails on ppc64
1826 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1827 369000 AMD64 fma4 instructions unsupported.
1828 369169 ppc64 fails jm_int_isa_2_07 test
1829 369175 jm_vec_isa_2_07 test crashes on ppc64
1830 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1831 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1832 369359 msghdr_foreachfield can crash when handling bad iovec
1833 369360 Bad sigprocmask old or new sets can crash valgrind
1834 369361 vmsplice syscall wrapper crashes on bad iovec
1835 369362 Bad sigaction arguments crash valgrind
1836 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1837 369402 Bad set/get_thread_area pointer crashes valgrind
1838 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1839 369446 valgrind crashes on unknown fcntl command
1840 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1841 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1842 370265 ISA 3.0 HW cap stuff needs updating
1843 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1844 372195 Power PC, xxsel instruction is not always recognized
1846 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1847 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1848 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1849 n-i-bz Fix ppoll_alarm exclusion on OS X
1850 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1851 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1852 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1853 n-i-bz amd64: memcheck false positive with shr %edx
1854 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1855 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1856 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1857 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1859 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1860 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1861 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1865 Release 3.11.0 (22 September 2015)
1866 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1868 3.11.0 is a feature release with many improvements and the usual
1869 collection of bug fixes.
1871 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1872 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1873 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1874 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1875 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1876 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1878 * ================== PLATFORM CHANGES =================
1880 * Support for Solaris/x86 and Solaris/amd64 has been added.
1882 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1884 * Preliminary support for the Tilera TileGX architecture has been added.
1886 * s390x: It is now required for the host to have the "long displacement"
1887 facility. The oldest supported machine model is z990.
1889 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1890 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1892 * The JIT's register allocator is significantly faster, making the JIT
1893 as a whole somewhat faster, so JIT-intensive activities, for example
1894 program startup, are modestly faster, around 5%.
1896 * There have been changes to the default settings of several command
1897 line flags, as detailed below.
1899 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1900 capable hosts, the simulated CPUID will now indicate AVX2 support.
1902 * ==================== TOOL CHANGES ====================
1906 - The default value for --leak-check-heuristics has been changed from
1907 "none" to "all". This helps to reduce the number of possibly
1908 lost blocks, in particular for C++ applications.
1910 - The default value for --keep-stacktraces has been changed from
1911 "malloc-then-free" to "malloc-and-free". This has a small cost in
1912 memory (one word per malloc-ed block) but allows Memcheck to show the
1913 3 stacktraces of a dangling reference: where the block was allocated,
1914 where it was freed, and where it is acccessed after being freed.
1916 - The default value for --partial-loads-ok has been changed from "no" to
1917 "yes", so as to avoid false positive errors resulting from some kinds
1918 of vectorised loops.
1920 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1921 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1922 than get_vbits when you need to associate byte data value with
1923 their corresponding validity bits.
1925 - The 'block_list' monitor command has been enhanced:
1926 o it can print a range of loss records
1927 o it now accepts an optional argument 'limited <max_blocks>'
1928 to control the number of blocks printed.
1929 o if a block has been found using a heuristic, then
1930 'block_list' now shows the heuristic after the block size.
1931 o the loss records/blocks to print can be limited to the blocks
1932 found via specified heuristics.
1934 - The C helper functions used to instrument loads on
1935 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1936 replaced by handwritten assembly sequences. This gives speedups
1937 in the region of 0% to 7% for those targets only.
1939 - A new command line option, --expensive-definedness-checks=yes|no,
1940 has been added. This is useful for avoiding occasional invalid
1941 uninitialised-value errors in optimised code. Watch out for
1942 runtime degradation, as this can be up to 25%. As always, though,
1943 the slowdown is highly application specific. The default setting
1948 - A new monitor command 'all_snapshots <filename>' dumps all
1949 snapshots taken so far.
1953 - Significant memory reduction and moderate speedups for
1954 --history-level=full for applications accessing a lot of memory
1955 with many different stacktraces.
1957 - The default value for --conflict-cache-size=N has been doubled to
1958 2000000. Users that were not using the default value should
1959 preferably also double the value they give.
1961 The default was changed due to the changes in the "full history"
1962 implementation. Doubling the value gives on average a slightly more
1963 complete history and uses similar memory (or significantly less memory
1964 in the worst case) than the previous implementation.
1966 - The Helgrind monitor command 'info locks' now accepts an optional
1967 argument 'lock_addr', which shows information about the lock at the
1970 - When using --history-level=full, the new Helgrind monitor command
1971 'accesshistory <addr> [<len>]' will show the recorded accesses for
1972 <len> (or 1) bytes at <addr>.
1974 * ==================== OTHER CHANGES ====================
1976 * The default value for the --smc-check option has been changed from
1977 "stack" to "all-non-file" on targets that provide automatic D-I
1978 cache coherence (x86, amd64 and s390x). The result is to provide,
1979 by default, transparent support for JIT generated and self-modifying
1980 code on all targets.
1982 * Mac OS X only: the default value for the --dsymutil option has been
1983 changed from "no" to "yes", since any serious usage on Mac OS X
1984 always required it to be "yes".
1986 * The command line options --db-attach and --db-command have been removed.
1987 They were deprecated in 3.10.0.
1989 * When a process dies due to a signal, Valgrind now shows the signal
1990 and the stacktrace at default verbosity (i.e. verbosity 1).
1992 * The address description logic used by Memcheck and Helgrind now
1993 describes addresses in anonymous segments, file mmap-ed segments,
1994 shared memory segments and the brk data segment.
1996 * The new option --error-markers=<begin>,<end> can be used to mark the
1997 begin/end of errors in textual output mode, to facilitate
1998 searching/extracting errors in output files that mix valgrind errors
1999 with program output.
2001 * The new option --max-threads=<number> can be used to change the number
2002 of threads valgrind can handle. The default is 500 threads which
2003 should be more than enough for most applications.
2005 * The new option --valgrind-stacksize=<number> can be used to change the
2006 size of the private thread stacks used by Valgrind. This is useful
2007 for reducing memory use or increasing the stack size if Valgrind
2008 segfaults due to stack overflow.
2010 * The new option --avg-transtab-entry-size=<number> can be used to specify
2011 the expected instrumented block size, either to reduce memory use or
2012 to avoid excessive retranslation.
2014 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
2016 * New and modified GDB server monitor features:
2018 - When a signal is reported in GDB, you can now use the GDB convenience
2019 variable $_siginfo to examine detailed signal information.
2021 - Valgrind's gdbserver now allows the user to change the signal
2022 to deliver to the process. So, use 'signal SIGNAL' to continue execution
2023 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
2024 continue without passing the signal to the process.
2026 - With GDB >= 7.10, the command 'target remote'
2027 will automatically load the executable file of the process running
2028 under Valgrind. This means you do not need to specify the executable
2029 file yourself, GDB will discover it itself. See GDB documentation about
2030 'qXfer:exec-file:read' packet for more info.
2032 * ==================== FIXED BUGS ====================
2034 The following bugs have been fixed or resolved. Note that "n-i-bz"
2035 stands for "not in bugzilla" -- that is, a bug that was reported to us
2036 but never got a bugzilla entry. We encourage you to file bugs in
2037 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2038 than mailing the developers (or mailing lists) directly -- bugs that
2039 are not entered into bugzilla tend to get forgotten about or ignored.
2041 To see details of a given bug, visit
2042 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2043 where XXXXXX is the bug number as listed below.
2045 116002 VG_(printf): Problems with justification of strings and integers
2046 155125 avoid cutting away file:lineno after long function name
2047 197259 Unsupported arch_prtctl PR_SET_GS option
2048 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
2049 201216 Fix Valgrind does not support pthread_sigmask() on OS X
2050 201435 Fix Darwin: -v does not show kernel version
2051 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
2052 211256 Fixed an outdated comment regarding the default platform.
2053 211529 Incomplete call stacks for code compiled by newer versions of MSVC
2054 211926 Avoid compilation warnings in valgrind.h with -pedantic
2055 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
2057 226609 Crediting upstream authors in man page
2058 231257 Valgrind omits path when executing script from shebang line
2059 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
2060 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
2061 269360 s390x: Fix addressing mode selection for compare-and-swap
2062 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
2064 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
2065 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
2066 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
2067 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
2068 330147 libmpiwrap PMPI_Get_count returns undefined value
2069 333051 mmap of huge pages fails due to incorrect alignment
2071 334802 valgrind does not always explain why a given option is bad
2072 335618 mov.w rN, pc/sp (ARM32)
2073 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
2078 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
2079 338602 AVX2 bit in CPUID missing
2080 338606 Strange message for scripts with invalid interpreter
2081 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
2082 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
2083 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
2085 339156 gdbsrv not called for fatal signal
2086 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
2087 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
2088 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
2089 339442 Fix testsuite build failure on OS X 10.9
2090 339542 Enable compilation with Intel's ICC compiler
2091 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
2092 339688 Mac-specific ASM does not support .version directive (cpuid,
2093 tronical and pushfpopf tests)
2094 339745 Valgrind crash when check Marmalade app (partial fix)
2095 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
2096 339778 Linux/TileGx platform support for Valgrind
2097 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
2098 339789 Fix none/tests/execve test on Mac OS X 10.9
2099 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
2100 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
2101 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
2102 340392 Allow user to select more accurate definedness checking in memcheck
2103 to avoid invalid complaints on optimised code
2104 340430 Fix some grammatical weirdness in the manual.
2105 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
2106 341419 Signal handler ucontext_t not filled out correctly on OS X
2107 341539 VG_(describe_addr) should not describe address as belonging to client
2108 segment if it is past the heap end
2109 341613 Enable building of manythreads and thread-exits tests on Mac OS X
2110 341615 Fix none/tests/darwin/access_extended test on Mac OS X
2111 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
2112 341789 aarch64: shmat fails with valgrind on ARMv8
2113 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
2114 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
2115 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
2116 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
2117 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
2118 342221 socket connect false positive uninit memory for unknown af family
2119 342353 Allow dumping full massif output while valgrind is still running
2120 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
2124 342603 Add I2C_SMBUS ioctl support
2125 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
2126 342683 Mark memory past the initial brk limit as unaddressable
2127 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
2128 342795 Internal glibc __GI_mempcpy call should be intercepted
2129 342841 s390x: Support instructions fiebr(a) and fidbr(a)
2130 343012 Unhandled syscall 319 (memfd_create)
2131 343069 Patch updating v4l2 API support
2132 343173 helgrind crash during stack unwind
2133 343219 fix GET_STARTREGS for arm
2134 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
2135 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
2136 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
2137 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
2138 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
2139 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
2140 343597 ppc64le: incorrect use of offseof macro
2141 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
2142 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
2143 343732 Unhandled syscall 144 (setgid) on aarch64
2144 343733 Unhandled syscall 187 (msgctl and related) on aarch64
2145 343802 s390x: False positive "conditional jump or move depends on [..]
2146 343902 --vgdb=yes doesn't break when --xml=yes is used
2147 343967 Don't warn about setuid/setgid/setcap executable for directories
2148 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
2149 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
2150 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
2151 344054 www - update info for Solaris/illumos
2152 344416 'make regtest' does not work cleanly on OS X
2153 344235 Remove duplicate include of pub_core_aspacemgr.h
2154 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
2155 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
2156 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
2157 344314 callgrind_annotate ... warnings about commands containing newlines
2158 344318 socketcall should wrap recvmmsg and sendmmsg
2159 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
2160 344416 Fix 'make regtest' does not work cleanly on OS X
2161 344499 Fix compilation for Linux kernel >= 4.0.0
2162 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
2163 unix:349 (__pthread_fchdir)
2164 344559 Garbage collection of unused segment names in address space manager
2165 344560 Fix stack traces missing penultimate frame on OS X
2166 344621 Fix memcheck/tests/err_disable4 test on OS X
2167 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
2168 344702 Fix missing libobjc suppressions on OS X 10.10
2170 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
2171 344939 Fix memcheck/tests/xml1 on OS X 10.10
2172 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
2173 345079 Fix build problems in VEX/useful/test_main.c
2174 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
2175 345177 arm64: prfm (reg) not implemented
2176 345215 Performance improvements for the register allocator
2177 345248 add support for Solaris OS in valgrind
2178 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
2179 345394 Fix memcheck/tests/strchr on OS X
2180 345637 Fix memcheck/tests/sendmsg on OS X
2181 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
2182 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
2183 345887 Fix an assertion in the address space manager
2184 345928 amd64: callstack only contains current function for small stacks
2185 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
2186 345987 MIPS64: Implement cavium LHX instruction
2187 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
2188 346185 Fix typo saving altivec register v24
2189 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
2190 and LibVEX_GuestPPC64_get_CR()
2191 346270 Regression tests none/tests/jm_vec/isa_2_07 and
2192 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
2193 346307 fuse filesystem syscall deadlocks
2194 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
2195 346411 MIPS: SysRes::_valEx handling is incorrect
2196 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
2197 346474 PPC64 Power 8, spr TEXASRU register not supported
2198 346487 Compiler generates "note" about a future ABI change for PPC64
2199 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
2200 and causing spurious "invalid read of size 8" warnings
2201 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
2202 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
2203 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
2204 347322 Power PC regression test cleanup
2205 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
2207 347389 unhandled syscall: 373 (Linux ARM syncfs)
2208 347686 Patch set to cleanup PPC64 regtests
2209 347978 Remove bash dependencies where not needed
2210 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
2211 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
2213 348102 Patch updating v4l2 API support
2214 348247 amd64 front end: jno jumps wrongly when overflow is not set
2215 348269 Improve mmap MAP_HUGETLB support.
2216 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
2217 348345 Assertion fails for negative lineno
2218 348377 Unsupported ARM instruction: yield
2219 348565 Fix detection of command line option availability for clang
2220 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
2221 348728 Fix broken check for VIDIOC_G_ENC_INDEX
2222 348748 Fix redundant condition
2223 348890 Fix clang warning about unsupported --param inline-unit-growth=900
2224 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
2225 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
2226 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
2227 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
2228 349626 Implemented additional Xen hypercalls
2229 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
2230 349790 Clean up of the hardware capability checking utilities.
2231 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
2232 349874 Fix typos in source code
2233 349879 memcheck: add handwritten assembly for helperc_LOADV*
2234 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
2235 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
2236 350202 Add limited param to 'monitor block_list'
2237 350290 s390x: Support instructions fixbr(a)
2238 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
2239 350809 Fix none/tests/async-sigs for Solaris
2240 350811 Remove reference to --db-attach which has been removed.
2241 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
2242 350854 hard-to-understand code in VG_(load_ELF)()
2243 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2244 351386 Solaris: Cannot run ld.so.1 under Valgrind
2245 351474 Fix VG_(iseqsigset) as obvious
2246 351531 Typo in /include/vki/vki-xen-physdev.h header guard
2247 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
2248 351858 ldsoexec support on Solaris
2249 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
2250 352130 helgrind reports false races for printfs using mempcpy on FILE* state
2251 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
2252 352320 arm64 crash on none/tests/nestedfs
2253 352765 Vbit test fails on Power 6
2254 352768 The mbar instruction is missing from the Power PC support
2255 352769 Power PC program priority register (PPR) is not supported
2256 n-i-bz Provide implementations of certain compiler builtins to support
2257 compilers that may not provide those
2258 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
2259 n-i-bz Fix compilation on distros with glibc < 2.5
2260 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
2261 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
2262 n-i-bz mremap did not work properly on shared memory
2263 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
2264 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
2266 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
2267 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
2268 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
2272 Release 3.10.1 (25 November 2014)
2273 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2274 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
2275 and backports fixes for all reported missing AArch64 ARMv8 instructions
2276 and syscalls from the trunk. If you package or deliver 3.10.0 for others
2277 to use, you might want to consider upgrading to 3.10.1 instead.
2279 The following bugs have been fixed or resolved. Note that "n-i-bz"
2280 stands for "not in bugzilla" -- that is, a bug that was reported to us
2281 but never got a bugzilla entry. We encourage you to file bugs in
2282 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2283 than mailing the developers (or mailing lists) directly -- bugs that
2284 are not entered into bugzilla tend to get forgotten about or ignored.
2286 To see details of a given bug, visit
2287 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2288 where XXXXXX is the bug number as listed below.
2290 335440 arm64: ld1 (single structure) is not implemented
2291 335713 arm64: unhanded instruction: prfm (immediate)
2292 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
2293 339182 ppc64: AvSplat ought to load destination vector register with [..]
2294 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
2295 339433 ppc64 lxvw4x instruction uses four 32-byte loads
2296 339645 Use correct tag names in sys_getdents/64 wrappers
2297 339706 Fix false positive for ioctl(TIOCSIG) on linux
2298 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
2299 339853 arm64 times syscall unknown
2300 339855 arm64 unhandled getsid/setsid syscalls
2301 339858 arm64 dmb sy not implemented
2302 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
2303 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
2304 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
2306 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
2307 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
2308 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
2309 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
2310 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
2311 340509 arm64: unhandled instruction fcvtas
2312 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
2313 340632 arm64: unhandled instruction fcvtas
2314 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
2315 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
2316 340788 warning: unhandled syscall: 318 (getrandom)
2317 340807 disInstr(arm): unhandled instruction: 0xEE989B20
2318 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
2319 340922 arm64: unhandled getgroups/setgroups syscalls
2320 350251 Fix typo in VEX utility program (test_main.c).
2321 350407 arm64: unhandled instruction ucvtf (vector, integer)
2322 350809 none/tests/async-sigs breaks when run under cron on Solaris
2323 350811 update README.solaris after r15445
2324 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
2325 350854 strange code in VG_(load_ELF)()
2326 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2327 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
2328 n-i-bz Add missing ]] to terminate CDATA.
2329 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
2330 n-i-bz Enable sys_fadvise64_64 on arm32.
2331 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
2332 n-i-bz Add test cases for all known arm64 load/store instructions.
2333 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
2334 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
2335 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
2336 n-i-bz arm64: Implement "fcvtpu w, s".
2337 n-i-bz arm64: implement ADDP and various others
2338 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
2339 n-i-bz arm64: enable FCVT{A,N}S X,S.
2341 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
2345 Release 3.10.0 (10 September 2014)
2346 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2348 3.10.0 is a feature release with many improvements and the usual
2349 collection of bug fixes.
2351 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
2352 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
2353 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
2354 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
2355 significantly improved relative to the 3.9.0 release.
2357 * ================== PLATFORM CHANGES =================
2359 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
2360 is mostly complete, and is usable, but some SIMD instructions are as
2363 * Support for little-endian variant of the 64-bit POWER architecture.
2365 * Support for Android on MIPS32.
2367 * Support for 64bit FPU on MIPS32 platforms.
2369 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
2371 * Configuration for and running on Android targets has changed.
2372 See README.android in the source tree for details.
2374 * ================== DEPRECATED FEATURES =================
2376 * --db-attach is now deprecated and will be removed in the next
2377 valgrind feature release. The built-in GDB server capabilities are
2378 superior and should be used instead. Learn more here:
2379 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
2381 * ==================== TOOL CHANGES ====================
2385 - Client code can now selectively disable and re-enable reporting of
2386 invalid address errors in specific ranges using the new client
2387 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
2388 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
2390 - Leak checker: there is a new leak check heuristic called
2391 "length64". This is used to detect interior pointers pointing 8
2392 bytes inside a block, on the assumption that the first 8 bytes
2393 holds the value "block size - 8". This is used by
2394 sqlite3MemMalloc, for example.
2396 - Checking of system call parameters: if a syscall parameter
2397 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
2398 several fields not initialised, an error is now reported for each
2399 field. Previously, an error was reported only for the first
2400 uninitialised field.
2402 - Mismatched alloc/free checking: a new flag
2403 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
2404 such checks if necessary.
2408 - Improvements to error messages:
2410 o Race condition error message involving heap allocated blocks also
2411 show the thread number that allocated the raced-on block.
2413 o All locks referenced by an error message are now announced.
2414 Previously, some error messages only showed the lock addresses.
2416 o The message indicating where a lock was first observed now also
2417 describes the address/location of the lock.
2419 - Helgrind now understands the Ada task termination rules and
2420 creates a happens-before relationship between a terminated task
2421 and its master. This avoids some false positives and avoids a big
2422 memory leak when a lot of Ada tasks are created and terminated.
2423 The interceptions are only activated with forthcoming releases of
2424 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
2426 - A new GDB server monitor command "info locks" giving the list of
2427 locks, their location, and their status.
2431 - callgrind_control now supports the --vgdb-prefix argument,
2432 which is needed if valgrind was started with this same argument.
2434 * ==================== OTHER CHANGES ====================
2436 * Unwinding through inlined function calls. Stack unwinding can now
2437 make use of Dwarf3 inlined-unwind information if it is available.
2438 The practical effect is that inlined calls become visible in stack
2439 traces. The suppression matching machinery has been adjusted
2440 accordingly. This is controlled by the new option
2441 --read-inline-info=yes|no. Currently this is enabled by default
2442 only on Linux and Android targets and only for the tools Memcheck,
2445 * Valgrind can now read EXIDX unwind information on 32-bit ARM
2446 targets. If an object contains both CFI and EXIDX unwind
2447 information, Valgrind will prefer the CFI over the EXIDX. This
2448 facilitates unwinding through system libraries on arm-android
2451 * Address description logic has been improved and is now common
2452 between Memcheck and Helgrind, resulting in better address
2453 descriptions for some kinds of error messages.
2455 * Error messages about dubious arguments (eg, to malloc or calloc) are
2456 output like other errors. This means that they can be suppressed
2457 and they have a stack trace.
2459 * The C++ demangler has been updated for better C++11 support.
2461 * New and modified GDB server monitor features:
2463 - Thread local variables/storage (__thread) can now be displayed.
2465 - The GDB server monitor command "v.info location <address>"
2466 displays information about an address. The information produced
2467 depends on the tool and on the options given to valgrind.
2468 Possibly, the following are described: global variables, local
2469 (stack) variables, allocated or freed blocks, ...
2471 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
2472 ask the GDB server to stop at the start of program execution, at
2473 the end of the program execution and on Valgrind internal errors.
2475 - A new monitor command "v.info stats" shows various Valgrind core
2476 and tool statistics.
2478 - A new monitor command "v.set hostvisibility" allows the GDB server
2479 to provide access to Valgrind internal host status/memory.
2481 * A new option "--aspace-minaddr=<address>" can in some situations
2482 allow the use of more memory by decreasing the address above which
2483 Valgrind maps memory. It can also be used to solve address
2484 conflicts with system libraries by increasing the default value.
2485 See user manual for details.
2487 * The amount of memory used by Valgrind to store debug info (unwind
2488 info, line number information and symbol data) has been
2489 significantly reduced, even though Valgrind now reads more
2490 information in order to support unwinding of inlined function calls.
2492 * Dwarf3 handling with --read-var-info=yes has been improved:
2494 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
2496 - Code compiled with
2497 -ffunction-sections -fdata-sections -Wl,--gc-sections
2498 no longer causes assertion failures.
2500 * Improved checking for the --sim-hints= and --kernel-variant=
2501 options. Unknown strings are now detected and reported to the user
2504 * The semantics of stack start/end boundaries in the valgrind.h
2505 VALGRIND_STACK_REGISTER client request has been clarified and
2506 documented. The convention is that start and end are respectively
2507 the lowest and highest addressable bytes of the stack.
2509 * ==================== FIXED BUGS ====================
2511 The following bugs have been fixed or resolved. Note that "n-i-bz"
2512 stands for "not in bugzilla" -- that is, a bug that was reported to us
2513 but never got a bugzilla entry. We encourage you to file bugs in
2514 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2515 than mailing the developers (or mailing lists) directly -- bugs that
2516 are not entered into bugzilla tend to get forgotten about or ignored.
2518 To see details of a given bug, visit
2519 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2520 where XXXXXX is the bug number as listed below.
2522 175819 Support for ipv6 socket reporting with --track-fds
2523 232510 make distcheck fails
2524 249435 Analyzing wine programs with callgrind triggers a crash
2525 278972 support for inlined function calls in stacktraces and suppression
2527 291310 FXSAVE instruction marks memory as undefined on amd64
2528 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2529 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2530 315199 vgcore file for threaded app does not show which thread crashed
2531 315952 tun/tap ioctls are not supported
2532 323178 Unhandled instruction: PLDW register (ARM)
2533 323179 Unhandled instruction: PLDW immediate (ARM)
2534 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2535 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2536 325124 [MIPSEL] Compilation error
2537 325477 Phase 4 support for IBM Power ISA 2.07
2538 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2539 325628 Phase 5 support for IBM Power ISA 2.07
2540 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2541 325751 Missing the two privileged Power PC Transactional Memory Instructions
2542 325816 Phase 6 support for IBM Power ISA 2.07
2543 325856 Make SGCheck fail gracefully on unsupported platforms
2544 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2545 326436 DRD: False positive in libstdc++ std::list::push_back
2546 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2547 326462 Refactor vgdb to isolate invoker stuff into separate module
2548 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2549 326623 DRD: false positive conflict report in a field assignment
2550 326724 Valgrind does not compile on OSX 1.9 Mavericks
2551 326816 Intercept for __strncpy_sse2_unaligned missing?
2552 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2553 326983 Clear direction flag after tests on amd64.
2554 327212 Do not prepend the current directory to absolute path names.
2555 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2556 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2557 327284 s390x: Fix translation of the risbg instruction
2558 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2559 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2560 327916 DW_TAG_typedef may have no name
2561 327943 s390x: add a redirection for the 'index' function
2562 328100 XABORT not implemented
2563 328205 Implement additional Xen hypercalls
2564 328454 add support Backtraces with ARM unwind tables (EXIDX)
2565 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2566 328711 valgrind.1 manpage "memcheck options" section is badly generated
2567 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2568 329612 Incorrect handling of AT_BASE for image execution
2569 329694 clang warns about using uninitialized variable
2570 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2571 330228 mmap must align to VKI_SHMLBA on mips32
2572 330257 LLVM does not support `-mno-dynamic-no-pic` option
2573 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2574 330459 --track-fds=yes doesn't track eventfds
2575 330469 Add clock_adjtime syscall support
2576 330594 Missing sysalls on PowerPC / uClibc
2577 330622 Add test to regression suite for POWER instruction: dcbzl
2578 330939 Support for AMD's syscall instruction on x86
2580 330941 Typo in PRE(poll) syscall wrapper
2581 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2582 331254 Fix expected output for memcheck/tests/dw4
2583 331255 Fix race condition in test none/tests/coolo_sigaction
2584 331257 Fix type of jump buffer in test none/tests/faultstatus
2585 331305 configure uses bash specific syntax
2586 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2587 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2588 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2589 331829 Unexpected ioctl opcode sign extension
2590 331830 ppc64: WARNING: unhandled syscall: 96/97
2591 331839 drd/tests/sem_open specifies invalid semaphore name
2592 331847 outcome of drd/tests/thread_name is nondeterministic
2593 332037 Valgrind cannot handle Thumb "add pc, reg"
2594 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2595 consistency checks enabled
2596 332263 intercepts for pthread_rwlock_timedrdlock and
2597 pthread_rwlock_timedwrlock are incorrect
2598 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2600 332276 Implement additional Xen hypercalls
2601 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2602 332765 Fix ms_print to create temporary files in a proper directory
2603 333072 drd: Add semaphore annotations
2604 333145 Tests for missaligned PC+#imm access for arm
2605 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2606 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2607 333248 WARNING: unhandled syscall: unix:443
2608 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2609 333501 cachegrind: assertion: Cache set count is not a power of two.
2612 333666 Recognize MPX instructions and bnd prefix.
2613 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2614 333817 Valgrind reports the memory areas written to by the SG_IO
2616 334049 lzcnt fails silently (x86_32)
2617 334384 Valgrind does not have support Little Endian support for
2619 334585 recvmmsg unhandled (+patch) (arm)
2620 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2621 334727 Build fails with -Werror=format-security
2622 334788 clarify doc about --log-file initial program directory
2623 334834 PPC64 Little Endian support, patch 2
2624 334836 PPC64 Little Endian support, patch 3 testcase fixes
2625 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2626 335034 Unhandled ioctl: HCIGETDEVLIST
2627 335155 vgdb, fix error print statement.
2628 335262 arm64: movi 8bit version is not supported
2629 335263 arm64: dmb instruction is not implemented
2630 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2631 335496 arm64: sbc/abc instructions are not implemented
2632 335554 arm64: unhandled instruction: abs
2633 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2634 335735 arm64: unhandled instruction: cnt
2635 335736 arm64: unhandled instruction: uaddlv
2636 335848 arm64: unhandled instruction: {s,u}cvtf
2637 335902 arm64: unhandled instruction: sli
2638 335903 arm64: unhandled instruction: umull (vector)
2639 336055 arm64: unhandled instruction: mov (element)
2640 336062 arm64: unhandled instruction: shrn{,2}
2641 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2642 336189 arm64: unhandled Instruction: mvn
2643 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2644 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2645 336772 Make moans about unknown ioctls more informative
2646 336957 Add a section about the Solaris/illumos port on the webpage
2647 337094 ifunc wrapper is broken on ppc64
2648 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2649 337528 leak check heuristic for block prefixed by length as 64bit number
2650 337740 Implement additional Xen hypercalls
2651 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2652 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2653 337871 deprecate --db-attach
2654 338023 Add support for all V4L2/media ioctls
2655 338024 inlined functions are not shown if DW_AT_ranges is used
2656 338106 Add support for 'kcmp' syscall
2657 338115 DRD: computed conflict set differs from actual after fork
2658 338160 implement display of thread local storage in gdbsrv
2659 338205 configure.ac and check for -Wno-tautological-compare
2660 338300 coredumps are missing one byte of every segment
2661 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2662 338499 --sim-hints parsing broken due to wrong order in tokens
2663 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2664 338681 Unable to unwind through clone thread created on i386-linux
2665 338698 race condition between gdbsrv and vgdb on startup
2666 338703 helgrind on arm-linux gets false positives in dynamic loader
2667 338791 alt dwz files can be relative of debug/main file
2668 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2669 338932 build V-trunk with gcc-trunk
2670 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2671 345079 Fix build problems in VEX/useful/test_main.c
2672 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2673 n-i-bz s390x: Fix memory corruption for multithreaded applications
2674 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2675 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2676 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2677 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2678 n-i-bz Add command-line option --default-suppressions
2679 n-i-bz Add support for BLKDISCARDZEROES ioctl
2680 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2681 n-i-bz Add support for sys_pivot_root and sys_unshare
2683 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2684 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2685 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2689 Release 3.9.0 (31 October 2013)
2690 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2691 3.9.0 is a feature release with many improvements and the usual
2692 collection of bug fixes.
2694 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2695 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2696 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2697 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2699 * ================== PLATFORM CHANGES =================
2701 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2702 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2704 * Support for MIPS DSP ASE on MIPS32 platforms.
2706 * Support for s390x Decimal Floating Point instructions on hosts that
2707 have the DFP facility installed.
2709 * Support for POWER8 (Power ISA 2.07) instructions
2711 * Support for Intel AVX2 instructions. This is available only on 64
2714 * Initial support for Intel Transactional Synchronization Extensions,
2717 * Initial support for Hardware Transactional Memory on POWER.
2719 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2720 run large GUI apps tolerably well.
2722 * ==================== TOOL CHANGES ====================
2726 - Improvements in handling of vectorised code, leading to
2727 significantly fewer false error reports. You need to use the flag
2728 --partial-loads-ok=yes to get the benefits of these changes.
2730 - Better control over the leak checker. It is now possible to
2731 specify which leak kinds (definite/indirect/possible/reachable)
2732 should be displayed, which should be regarded as errors, and which
2733 should be suppressed by a given leak suppression. This is done
2734 using the options --show-leak-kinds=kind1,kind2,..,
2735 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2736 "match-leak-kinds:" line in suppression entries, respectively.
2738 Note that generated leak suppressions contain this new line and
2739 are therefore more specific than in previous releases. To get the
2740 same behaviour as previous releases, remove the "match-leak-kinds:"
2741 line from generated suppressions before using them.
2743 - Reduced "possible leak" reports from the leak checker by the use
2744 of better heuristics. The available heuristics provide detection
2745 of valid interior pointers to std::stdstring, to new[] allocated
2746 arrays with elements having destructors and to interior pointers
2747 pointing to an inner part of a C++ object using multiple
2748 inheritance. They can be selected individually using the
2749 option --leak-check-heuristics=heur1,heur2,...
2751 - Better control of stacktrace acquisition for heap-allocated
2752 blocks. Using the --keep-stacktraces option, it is possible to
2753 control independently whether a stack trace is acquired for each
2754 allocation and deallocation. This can be used to create better
2755 "use after free" errors or to decrease Valgrind's resource
2756 consumption by recording less information.
2758 - Better reporting of leak suppression usage. The list of used
2759 suppressions (shown when the -v option is given) now shows, for
2760 each leak suppressions, how many blocks and bytes it suppressed
2761 during the last leak search.
2765 - False errors resulting from the use of statically initialised
2766 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2769 - False errors resulting from the use of pthread_cond_waits that
2770 timeout, have been removed.
2772 * ==================== OTHER CHANGES ====================
2774 * Some attempt to tune Valgrind's space requirements to the expected
2775 capabilities of the target:
2777 - The default size of the translation cache has been reduced from 8
2778 sectors to 6 on Android platforms, since each sector occupies
2779 about 40MB when using Memcheck.
2781 - The default size of the translation cache has been increased to 16
2782 sectors on all other platforms, reflecting the fact that large
2783 applications require instrumentation and storage of huge amounts
2784 of code. For similar reasons, the number of memory mapped
2785 segments that can be tracked has been increased by a factor of 6.
2787 - In all cases, the maximum number of sectors in the translation
2788 cache can be controlled by the new flag --num-transtab-sectors.
2790 * Changes in how debug info (line numbers, etc) is read:
2792 - Valgrind no longer temporarily mmaps the entire object to read
2793 from it. Instead, reading is done through a small fixed sized
2794 buffer. This avoids virtual memory usage spikes when Valgrind
2795 reads debuginfo from large shared objects.
2797 - A new experimental remote debug info server. Valgrind can read
2798 debug info from a different machine (typically, a build host)
2799 where debuginfo objects are stored. This can save a lot of time
2800 and hassle when running Valgrind on resource-constrained targets
2801 (phones, tablets) when the full debuginfo objects are stored
2802 somewhere else. This is enabled by the --debuginfo-server=
2805 - Consistency checking between main and debug objects can be
2806 disabled using the --allow-mismatched-debuginfo option.
2808 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2809 scanning can recover stack traces in some cases when the normal
2810 unwind mechanisms fail. Stack scanning is best described as "a
2811 nasty, dangerous and misleading hack" and so is disabled by default.
2812 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2815 * Detection and merging of recursive stack frame cycles. When your
2816 program has recursive algorithms, this limits the memory used by
2817 Valgrind for recorded stack traces and avoids recording
2818 uninteresting repeated calls. This is controlled by the command
2819 line option --merge-recursive-frame and by the monitor command
2820 "v.set merge-recursive-frames".
2822 * File name and line numbers for used suppressions. The list of used
2823 suppressions (shown when the -v option is given) now shows, for each
2824 used suppression, the file name and line number where the suppression
2827 * New and modified GDB server monitor features:
2829 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2830 that can be used to execute gdbserver monitor commands from the
2833 - A new monitor command, "v.info open_fds", that gives the list of
2834 open file descriptors and additional details.
2836 - An optional message in the "v.info n_errs_found" monitor command,
2837 for example "v.info n_errs_found test 1234 finished", allowing a
2838 comment string to be added to the process output, perhaps for the
2839 purpose of separating errors of different tests or test phases.
2841 - A new monitor command "v.info execontext" that shows information
2842 about the stack traces recorded by Valgrind.
2844 - A new monitor command "v.do expensive_sanity_check_general" to run
2845 some internal consistency checks.
2847 * New flag --sigill-diagnostics to control whether a diagnostic
2848 message is printed when the JIT encounters an instruction it can't
2849 translate. The actual behavior -- delivery of SIGILL to the
2850 application -- is unchanged.
2852 * The maximum amount of memory that Valgrind can use on 64 bit targets
2853 has been increased from 32GB to 64GB. This should make it possible
2854 to run applications on Memcheck that natively require up to about 35GB.
2856 * ==================== FIXED BUGS ====================
2858 The following bugs have been fixed or resolved. Note that "n-i-bz"
2859 stands for "not in bugzilla" -- that is, a bug that was reported to us
2860 but never got a bugzilla entry. We encourage you to file bugs in
2861 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2862 than mailing the developers (or mailing lists) directly -- bugs that
2863 are not entered into bugzilla tend to get forgotten about or ignored.
2865 To see details of a given bug, visit
2866 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2867 where XXXXXX is the bug number as listed below.
2869 123837 system call: 4th argument is optional, depending on cmd
2870 135425 memcheck should tell you where Freed blocks were Mallocd
2871 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2872 207815 Adds some of the drm ioctls to syswrap-linux.c
2873 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2874 252955 Impossible to compile with ccache
2875 253519 Memcheck reports auxv pointer accesses as invalid reads.
2876 263034 Crash when loading some PPC64 binaries
2877 269599 Increase deepest backtrace
2878 274695 s390x: Support "compare to/from logical" instructions (z196)
2879 275800 s390x: Autodetect cache info (part 2)
2880 280271 Valgrind reports possible memory leaks on still-reachable std::string
2881 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2882 289578 Backtraces with ARM unwind tables (stack scan flags)
2883 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2884 304832 ppc32: build failure
2885 305431 Use find_buildid shdr fallback for separate .debug files
2886 305728 Add support for AVX2 instructions
2887 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2888 306035 s390x: Fix IR generation for LAAG and friends
2889 306054 s390x: Condition code computation for convert-to-int/logical
2890 306098 s390x: alternate opcode form for convert to/from fixed
2891 306587 Fix cache line detection from auxiliary vector for PPC.
2892 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2893 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2894 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2895 307101 sys_capget second argument can be NULL
2896 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2897 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2898 307113 s390x: DFP support
2899 307141 valgrind does't work in mips-linux system
2900 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2901 307285 x86_amd64 feature test for avx in test suite is wrong
2902 307290 memcheck overlap testcase needs memcpy version filter
2903 307463 Please add "&limit=0" to the "all open bugs" link
2904 307465 --show-possibly-lost=no should reduce the error count / exit code
2905 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2906 307729 pkgconfig support broken valgrind.pc
2907 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2908 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2909 308089 Unhandled syscall on ppc64: prctl
2910 308135 PPC32 MPC8xx has 16 bytes cache size
2911 308321 testsuite memcheck filter interferes with gdb_filter
2913 308341 vgdb should report process exit (or fatal signal)
2914 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2915 308495 Remove build dependency on installed Xen headers
2916 308573 Internal error on 64-bit instruction executed in 32-bit mode
2918 308627 pmovmskb validity bit propagation is imprecise
2919 308644 vgdb command for having the info for the track-fds option
2920 308711 give more info about aspacemgr and arenas in out_of_memory
2921 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2922 308718 ARM implement SMLALBB family of instructions
2923 308886 Missing support for PTRACE_SET/GETREGSET
2924 308930 syscall name_to_handle_at (303 on amd64) not handled
2925 309229 V-bit tester does not report number of tests generated
2926 309323 print unrecognized instuction on MIPS
2927 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2928 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2929 309430 Self hosting ppc64 encounters a vassert error on operand type
2930 309600 valgrind is a bit confused about 0-sized sections
2931 309823 Generate errors for still reachable blocks
2932 309921 PCMPISTRI validity bit propagation is imprecise
2933 309922 none/tests/ppc64/test_dfp5 sometimes fails
2934 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2935 310424 --read-var-info does not properly describe static variables
2936 310792 search additional path for debug symbols
2937 310931 s390x: Message-security assist (MSA) instruction extension [..]
2938 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2939 311318 ARM: "128-bit constant is not implemented" error message
2940 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2941 311690 V crashes because it redirects branches inside of a redirected function
2942 311880 x86_64: make regtest hangs at shell_valid1
2943 311922 WARNING: unhandled syscall: 170
2945 312171 ppc: insn selection for DFP
2946 312571 Rounding mode call wrong for the DFP Iops [..]
2947 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2948 312913 Dangling pointers error should also report the alloc stack trace
2949 312980 Building on Mountain Lion generates some compiler warnings
2950 313267 Adding MIPS64/Linux port to Valgrind
2953 313811 Buffer overflow in assert_fail
2954 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2955 314269 ppc: dead code in insn selection
2956 314718 ARM: implement integer divide instruction (sdiv and udiv)
2957 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2958 315441 sendmsg syscall should ignore unset msghdr msg_flags
2959 315534 msgrcv inside a thread causes valgrind to hang (block)
2960 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2961 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2962 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2963 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2964 316144 valgrind.1 manpage contains unknown ??? strings [..]
2965 316145 callgrind command line options in manpage reference (unknown) [..]
2966 316145 callgrind command line options in manpage reference [..]
2967 316181 drd: Fixed a 4x slowdown for certain applications
2968 316503 Valgrind does not support SSE4 "movntdqa" instruction
2969 316535 Use of |signed int| instead of |size_t| in valgrind messages
2970 316696 fluidanimate program of parsec 2.1 stuck
2971 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2972 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2973 317186 "Impossible happens" when occurs VCVT instruction on ARM
2974 317318 Support for Threading Building Blocks "scalable_malloc"
2975 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2976 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2977 317463 bmi testcase IR SANITY CHECK FAILURE
2978 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2979 318050 libmpiwrap fails to compile with out-of-source build
2980 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2981 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2982 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2983 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2984 318932 Add missing PPC64 and PPC32 system call support
2985 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2986 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2987 319494 VEX Makefile-gcc standalone build update after r2702
2988 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2989 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2990 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2991 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2992 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2993 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2994 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2995 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2996 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2997 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2998 320895 add fanotify support (patch included)
2999 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
3000 321065 Valgrind updates for Xen 4.3
3001 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
3002 321363 Unhandled instruction: SSAX (ARM + Thumb)
3003 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
3004 321466 Unhandled instruction: SHASX (ARM + Thumb)
3005 321467 Unhandled instruction: SHSAX (ARM + Thumb)
3006 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
3007 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
3008 321620 Unhandled instruction: UASX (ARM + Thumb)
3009 321621 Unhandled instruction: USAX (ARM + Thumb)
3010 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
3011 321693 Unhandled instruction: LDRSBT (Thumb)
3012 321694 Unhandled instruction: UQASX (ARM + Thumb)
3013 321696 Unhandled instruction: UQSAX (Thumb + ARM)
3014 321697 Unhandled instruction: UHASX (ARM + Thumb)
3015 321703 Unhandled instruction: UHSAX (ARM + Thumb)
3016 321704 Unhandled instruction: REVSH (ARM + Thumb)
3017 321730 Add cg_diff and cg_merge man pages
3018 321738 Add vgdb and valgrind-listener man pages
3020 321891 Unhandled instruction: LDRHT (Thumb)
3021 321960 pthread_create() then alloca() causing invalid stack write errors
3022 321969 ppc32 and ppc64 don't support [lf]setxattr
3023 322254 Show threadname together with tid if set by application
3024 322294 Add initial support for IBM Power ISA 2.07
3025 322368 Assertion failure in wqthread_hijack under OS X 10.8
3026 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
3027 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
3028 322851 0bXXX binary literal syntax is not standard
3029 323035 Unhandled instruction: LDRSHT(Thumb)
3030 323036 Unhandled instruction: SMMLS (ARM and Thumb)
3031 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
3032 323175 Unhandled instruction: SMLALD (ARM + Thumb)
3033 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
3034 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
3035 323437 Phase 2 support for IBM Power ISA 2.07
3036 323713 Support mmxext (integer sse) subset on i386 (athlon)
3037 323803 Transactional memory instructions are not supported for Power
3038 323893 SSE3 not available on amd cpus in valgrind
3039 323905 Probable false positive from Valgrind/drd on close()
3040 323912 valgrind.h header isn't compatible for mingw64
3041 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
3042 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
3043 324181 mmap does not handle MAP_32BIT
3044 324227 memcheck false positive leak when a thread calls exit+block [..]
3045 324421 Support for fanotify API on ARM architecture
3046 324514 gdbserver monitor cmd output behaviour consistency [..]
3047 324518 ppc64: Emulation of dcbt instructions does not handle [..]
3048 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
3049 324582 When access is made to freed memory, report both allocation [..]
3050 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
3051 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
3052 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
3053 324834 Unhandled instructions in Microsoft C run-time for x86_64
3054 324894 Phase 3 support for IBM Power ISA 2.07
3055 326091 drd: Avoid false race reports from optimized strlen() impls
3056 326113 valgrind libvex hwcaps error on AMD64
3057 n-i-bz Some wrong command line options could be ignored
3058 n-i-bz patch to allow fair-sched on android
3059 n-i-bz report error for vgdb snapshot requested before execution
3060 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
3062 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
3066 Release 3.8.1 (19 September 2012)
3067 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3068 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
3069 that occur moderately frequently in real use cases, adds support for
3070 some missing instructions on ARM, and fixes a deadlock condition on
3071 MacOSX. If you package or deliver 3.8.0 for others to use, you might
3072 want to consider upgrading to 3.8.1 instead.
3074 The following bugs have been fixed or resolved. Note that "n-i-bz"
3075 stands for "not in bugzilla" -- that is, a bug that was reported to us
3076 but never got a bugzilla entry. We encourage you to file bugs in
3077 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3078 than mailing the developers (or mailing lists) directly -- bugs that
3079 are not entered into bugzilla tend to get forgotten about or ignored.
3081 To see details of a given bug, visit
3082 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3083 where XXXXXX is the bug number as listed below.
3086 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
3087 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
3088 298281 wcslen causes false(?) uninitialised value warnings
3089 301281 valgrind hangs on OS X when the process calls system()
3090 304035 disInstr(arm): unhandled instruction 0xE1023053
3091 304867 implement MOVBE instruction in x86 mode
3092 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
3093 305042 amd64: implement 0F 7F encoding of movq between two registers
3094 305199 ARM: implement QDADD and QDSUB
3095 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
3096 305513 killed by fatal signal: SIGSEGV
3097 305690 DRD reporting invalid semaphore when sem_trywait fails
3098 305926 Invalid alignment checks for some AVX instructions
3099 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
3100 306310 3.8.0 release tarball missing some files
3101 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
3102 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
3103 n-i-bz shmat of a segment > 4Gb does not work
3104 n-i-bz simulate_control_c script wrong USR1 signal number on mips
3105 n-i-bz vgdb ptrace calls wrong on mips [...]
3106 n-i-bz Fixes for more MPI false positives
3107 n-i-bz exp-sgcheck's memcpy causes programs to segfault
3108 n-i-bz OSX build w/ clang: asserts at startup
3109 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
3110 n-i-bz fix a couple of union tag-vs-field mixups
3111 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
3113 The following bugs were fixed in 3.8.0 but not listed in this NEWS
3116 254088 Valgrind should know about UD2 instruction
3119 304754 NEWS blows TeX's little mind
3121 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
3125 Release 3.8.0 (10 August 2012)
3126 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3127 3.8.0 is a feature release with many improvements and the usual
3128 collection of bug fixes.
3130 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3131 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
3132 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
3133 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
3134 There is initial support for MacOSX 10.8, but it is not usable for
3135 serious work at present.
3137 * ================== PLATFORM CHANGES =================
3139 * Support for MIPS32 platforms running Linux. Valgrind has been
3140 tested on MIPS32 and MIPS32r2 platforms running different Debian
3141 Squeeze and MeeGo distributions. Both little-endian and big-endian
3142 cores are supported. The tools Memcheck, Massif and Lackey have
3143 been tested and are known to work. See README.mips for more details.
3145 * Preliminary support for Android running on x86.
3147 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
3149 * Support for Intel AVX instructions and for AES instructions. This
3150 support is available only for 64 bit code.
3152 * Support for POWER Decimal Floating Point instructions.
3154 * ==================== TOOL CHANGES ====================
3156 * Non-libc malloc implementations are now supported. This is useful
3157 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
3158 Using the new option --soname-synonyms, such tools can be informed
3159 that the malloc implementation is either linked statically into the
3160 executable, or is present in some other shared library different
3161 from libc.so. This makes it possible to process statically linked
3162 programs, and programs using other malloc libraries, for example
3163 TCMalloc or JEMalloc.
3165 * For tools that provide their own replacement for malloc et al, the
3166 option --redzone-size=<number> allows users to specify the size of
3167 the padding blocks (redzones) added before and after each client
3168 allocated block. Smaller redzones decrease the memory needed by
3169 Valgrind. Bigger redzones increase the chance to detect blocks
3170 overrun or underrun. Prior to this change, the redzone size was
3171 hardwired to 16 bytes in Memcheck.
3175 - The leak_check GDB server monitor command now can
3176 control the maximum nr of loss records to output.
3178 - Reduction of memory use for applications allocating
3179 many blocks and/or having many partially defined bytes.
3181 - Addition of GDB server monitor command 'block_list' that lists
3182 the addresses/sizes of the blocks of a leak search loss record.
3184 - Addition of GDB server monitor command 'who_points_at' that lists
3185 the locations pointing at a block.
3187 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
3188 detect an invalid access of these redzones, by marking them
3189 noaccess. Similarly, if a redzone size is given for a memory
3190 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
3191 This still allows to find some bugs if the user has forgotten to
3192 mark the pool superblock noaccess.
3194 - Performance of memory leak check has been improved, especially in
3195 cases where there are many leaked blocks and/or many suppression
3196 rules used to suppress leak reports.
3198 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
3199 more precise analysis, which is important for LLVM/Clang
3200 generated code. This is at the cost of somewhat reduced
3201 performance. Note there is no change to analysis precision or
3202 costs on Linux targets.
3206 - Added even more facilities that can help finding the cause of a data
3207 race, namely the command-line option --ptrace-addr and the macro
3208 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
3210 - Fixed a subtle bug that could cause false positive data race reports.
3212 * ==================== OTHER CHANGES ====================
3214 * The C++ demangler has been updated so as to work well with C++
3215 compiled by up to at least g++ 4.6.
3217 * Tool developers can make replacement/wrapping more flexible thanks
3218 to the new option --soname-synonyms. This was reported above, but
3219 in fact is very general and applies to all function
3220 replacement/wrapping, not just to malloc-family functions.
3222 * Round-robin scheduling of threads can be selected, using the new
3223 option --fair-sched= yes. Prior to this change, the pipe-based
3224 thread serialisation mechanism (which is still the default) could
3225 give very unfair scheduling. --fair-sched=yes improves
3226 responsiveness of interactive multithreaded applications, and
3227 improves repeatability of results from the thread checkers Helgrind
3230 * For tool developers: support to run Valgrind on Valgrind has been
3231 improved. We can now routinely Valgrind on Helgrind or Memcheck.
3233 * gdbserver now shows the float shadow registers as integer
3234 rather than float values, as the shadow values are mostly
3235 used as bit patterns.
3237 * Increased limit for the --num-callers command line flag to 500.
3239 * Performance improvements for error matching when there are many
3240 suppression records in use.
3242 * Improved support for DWARF4 debugging information (bug 284184).
3244 * Initial support for DWZ compressed Dwarf debug info.
3246 * Improved control over the IR optimiser's handling of the tradeoff
3247 between performance and precision of exceptions. Specifically,
3248 --vex-iropt-precise-memory-exns has been removed and replaced by
3249 --vex-iropt-register-updates, with extended functionality. This
3250 allows the Valgrind gdbserver to always show up to date register
3253 * Modest performance gains through the use of translation chaining for
3256 * ==================== FIXED BUGS ====================
3258 The following bugs have been fixed or resolved. Note that "n-i-bz"
3259 stands for "not in bugzilla" -- that is, a bug that was reported to us
3260 but never got a bugzilla entry. We encourage you to file bugs in
3261 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3262 than mailing the developers (or mailing lists) directly -- bugs that
3263 are not entered into bugzilla tend to get forgotten about or ignored.
3265 To see details of a given bug, visit
3266 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3267 where XXXXXX is the bug number as listed below.
3269 197914 Building valgrind from svn now requires automake-1.10
3270 203877 increase to 16Mb maximum allowed alignment for memalign et al
3271 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
3272 247386 make perf does not run all performance tests
3273 270006 Valgrind scheduler unfair
3274 270777 Adding MIPS/Linux port to Valgrind
3275 270796 s390x: Removed broken support for the TS insn
3276 271438 Fix configure for proper SSE4.2 detection
3277 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
3278 273475 Add support for AVX instructions
3279 274078 improved configure logic for mpicc
3280 276993 fix mremap 'no thrash checks'
3281 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
3282 281482 memcheck incorrect byte allocation count in realloc() for silly argument
3283 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
3284 283413 Fix wrong sanity check
3285 283671 Robustize alignment computation in LibVEX_Alloc
3286 283961 Adding support for some HCI IOCTLs
3287 284124 parse_type_DIE: confused by: DWARF 4
3288 284864 == 273475 (Add support for AVX instructions)
3289 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
3290 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
3291 285725 == 273475 (Add support for AVX instructions)
3292 286261 add wrapper for linux I2C_RDWR ioctl
3293 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
3294 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
3295 286384 configure fails "checking for a supported version of gcc"
3296 286497 == 273475 (Add support for AVX instructions)
3297 286596 == 273475 (Add support for AVX instructions)
3298 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
3299 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
3300 287260 Incorrect conditional jump or move depends on uninitialised value(s)
3301 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
3302 287307 == 273475 (Add support for AVX instructions)
3303 287858 VG_(strerror): unknown error
3304 288298 (MacOSX) unhandled syscall shm_unlink
3305 288995 == 273475 (Add support for AVX instructions)
3306 289470 Loading of large Mach-O thin binaries fails.
3307 289656 == 273475 (Add support for AVX instructions)
3308 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
3309 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
3310 289839 s390x: Provide support for unicode conversion instructions
3311 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
3312 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
3313 290655 Add support for AESKEYGENASSIST instruction
3314 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
3315 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
3316 291253 ES register not initialised in valgrind simulation
3317 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
3318 291865 s390x: Support the "Compare Double and Swap" family of instructions
3319 292300 == 273475 (Add support for AVX instructions)
3320 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
3321 292493 == 273475 (Add support for AVX instructions)
3322 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
3323 292627 Missing support for some SCSI ioctls
3324 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
3325 292841 == 273475 (Add support for AVX instructions)
3326 292993 implement the getcpu syscall on amd64-linux
3327 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
3328 293088 Add some VEX sanity checks for ppc64 unhandled instructions
3329 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
3330 293754 PCMPxSTRx not implemented for 16-bit characters
3331 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
3332 293808 CLFLUSH not supported by latest VEX for amd64
3333 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
3334 294048 MPSADBW instruction not implemented
3335 294055 regtest none/tests/shell fails when locale is not set to C
3336 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
3337 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
3338 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
3339 294260 disInstr_AMD64: disInstr miscalculated next %rip
3340 294523 --partial-loads-ok=yes causes false negatives
3341 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
3342 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
3343 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
3344 295089 can not annotate source for both helgrind and drd
3345 295221 POWER Processor decimal floating point instruction support missing
3346 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
3347 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
3348 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
3349 295617 ARM - Add some missing syscalls
3350 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
3351 296229 Linux user input device ioctls missing wrappers
3352 296318 ELF Debug info improvements (more than one rx/rw mapping)
3353 296422 Add translation chaining support
3354 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
3355 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
3356 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
3357 297078 gdbserver signal handling problems [..]
3358 297147 drd false positives on newly allocated memory
3359 297329 disallow decoding of IBM Power DFP insns on some machines
3360 297497 POWER Processor decimal floating point instruction support missing
3361 297701 Another alias for strncasecmp_l in libc-2.13.so
3362 297911 'invalid write' not reported when using APIs for custom mem allocators.
3363 297976 s390x: revisit EX implementation
3364 297991 Valgrind interferes with mmap()+ftell()
3365 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
3366 297993 Fix compilation of valgrind with gcc -g3.
3367 298080 POWER Processor DFP support missing, part 3
3368 298227 == 273475 (Add support for AVX instructions)
3369 298335 == 273475 (Add support for AVX instructions)
3370 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
3371 298394 s390x: Don't bail out on an unknown machine model. [..]
3372 298421 accept4() syscall (366) support is missing for ARM
3373 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
3374 298732 valgrind installation problem in ubuntu with kernel version 3.x
3375 298862 POWER Processor DFP instruction support missing, part 4
3376 298864 DWARF reader mis-parses DW_FORM_ref_addr
3377 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
3378 299053 Support DWARF4 DW_AT_high_pc constant form
3379 299104 == 273475 (Add support for AVX instructions)
3380 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
3381 299629 dup3() syscall (358) support is missing for ARM
3382 299694 POWER Processor DFP instruction support missing, part 5
3383 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
3384 299803 == 273475 (Add support for AVX instructions)
3385 299804 == 273475 (Add support for AVX instructions)
3386 299805 == 273475 (Add support for AVX instructions)
3387 300140 ARM - Missing (T1) SMMUL
3388 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
3389 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
3390 300414 FCOM and FCOMP unimplemented for amd64 guest
3391 301204 infinite loop in canonicaliseSymtab with ifunc symbol
3392 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
3393 301265 add x86 support to Android build
3394 301984 configure script doesn't detect certain versions of clang
3395 302205 Fix compiler warnings for POWER VEX code and POWER test cases
3396 302287 Unhandled movbe instruction on Atom processors
3397 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
3398 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
3399 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
3400 302656 == 273475 (Add support for AVX instructions)
3401 302709 valgrind for ARM needs extra tls support for android emulator [..]
3402 302827 add wrapper for CDROM_GET_CAPABILITY
3403 302901 Valgrind crashes with dwz optimized debuginfo
3404 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
3405 303116 Add support for the POWER instruction popcntb
3406 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
3407 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
3408 303466 == 273475 (Add support for AVX instructions)
3409 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
3410 303963 strstr() function produces wrong results under valgrind callgrind
3411 304054 CALL_FN_xx macros need to enforce stack alignment
3412 304561 tee system call not supported
3413 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
3414 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
3415 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
3416 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
3417 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
3418 n-i-bz s390x: Shadow registers can now be examined using vgdb
3420 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
3421 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
3425 Release 3.7.0 (5 November 2011)
3426 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3427 3.7.0 is a feature release with many significant improvements and the
3428 usual collection of bug fixes.
3430 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3431 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
3432 Support for recent distros and toolchain components (glibc 2.14, gcc
3433 4.6, MacOSX 10.7) has been added.
3435 * ================== PLATFORM CHANGES =================
3437 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
3438 analyse 64-bit programs running on z/Architecture. Most user space
3439 instructions up to and including z10 are supported. Valgrind has
3440 been tested extensively on z9, z10, and z196 machines running SLES
3441 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
3442 known to work well. Callgrind, Helgrind, and DRD work reasonably
3443 well on z9 and later models. See README.s390 for more details.
3445 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
3446 64-bit processes are supported. Some complex threaded applications
3447 (Firefox) are observed to hang when run as 32 bit applications,
3448 whereas 64-bit versions run OK. The cause is unknown. Memcheck
3449 will likely report some false errors. In general, expect some rough
3450 spots. This release also supports MacOSX 10.6, but drops support
3453 * Preliminary support for Android (on ARM). Valgrind can now run
3454 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
3455 README.android for more details, plus instructions on how to get
3458 * Support for the IBM Power ISA 2.06 (Power7 instructions)
3460 * General correctness and performance improvements for ARM/Linux, and,
3461 by extension, ARM/Android.
3463 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
3464 instruction set support is under development but is not available in
3467 * Support for AIX5 has been removed.
3469 * ==================== TOOL CHANGES ====================
3471 * Memcheck: some incremental changes:
3473 - reduction of memory use in some circumstances
3475 - improved handling of freed memory, which in some circumstances
3476 can cause detection of use-after-free that would previously have
3479 - fix of a longstanding bug that could cause false negatives (missed
3480 errors) in programs doing vector saturated narrowing instructions.
3482 * Helgrind: performance improvements and major memory use reductions,
3483 particularly for large, long running applications which perform many
3484 synchronisation (lock, unlock, etc) events. Plus many smaller
3487 - display of locksets for both threads involved in a race
3489 - general improvements in formatting/clarity of error messages
3491 - addition of facilities and documentation regarding annotation
3492 of thread safe reference counted C++ classes
3494 - new flag --check-stack-refs=no|yes [yes], to disable race checking
3495 on thread stacks (a performance hack)
3497 - new flag --free-is-write=no|yes [no], to enable detection of races
3498 where one thread accesses heap memory but another one frees it,
3499 without any coordinating synchronisation event
3501 * DRD: enabled XML output; added support for delayed thread deletion
3502 in order to detect races that occur close to the end of a thread
3503 (--join-list-vol); fixed a memory leak triggered by repeated client
3504 memory allocatation and deallocation; improved Darwin support.
3506 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
3508 * exp-sgcheck: this tool has been reduced in scope so as to improve
3509 performance and remove checking that Memcheck does better.
3510 Specifically, the ability to check for overruns for stack and global
3511 arrays is unchanged, but the ability to check for overruns of heap
3512 blocks has been removed. The tool has accordingly been renamed to
3513 exp-sgcheck ("Stack and Global Array Checking").
3515 * ==================== OTHER CHANGES ====================
3517 * GDB server: Valgrind now has an embedded GDB server. That means it
3518 is possible to control a Valgrind run from GDB, doing all the usual
3519 things that GDB can do (single stepping, breakpoints, examining
3520 data, etc). Tool-specific functionality is also available. For
3521 example, it is possible to query the definedness state of variables
3522 or memory from within GDB when running Memcheck; arbitrarily large
3523 memory watchpoints are supported, etc. To use the GDB server, start
3524 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3527 * Improved support for unfriendly self-modifying code: a new option
3528 --smc-check=all-non-file is available. This adds the relevant
3529 consistency checks only to code that originates in non-file-backed
3530 mappings. In effect this confines the consistency checking only to
3531 code that is or might be JIT generated, and avoids checks on code
3532 that must have been compiled ahead of time. This significantly
3533 improves performance on applications that generate code at run time.
3535 * It is now possible to build a working Valgrind using Clang-2.9 on
3538 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3539 These enable and disable error reporting on a per-thread, and
3540 nestable, basis. This is useful for hiding errors in particularly
3541 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3542 now uses this facility.
3544 * Added the --mod-funcname option to cg_diff.
3546 * ==================== FIXED BUGS ====================
3548 The following bugs have been fixed or resolved. Note that "n-i-bz"
3549 stands for "not in bugzilla" -- that is, a bug that was reported to us
3550 but never got a bugzilla entry. We encourage you to file bugs in
3551 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3552 mailing the developers (or mailing lists) directly -- bugs that are
3553 not entered into bugzilla tend to get forgotten about or ignored.
3555 To see details of a given bug, visit
3556 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3557 where XXXXXX is the bug number as listed below.
3559 79311 malloc silly arg warning does not give stack trace
3560 210935 port valgrind.h (not valgrind) to win32 to support client requests
3561 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3562 243404 Port to zSeries
3563 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3564 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3565 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3566 253206 Some fixes for the faultstatus testcase
3567 255223 capget testcase fails when running as root
3568 256703 xlc_dbl_u32.c testcase broken
3569 256726 Helgrind tests have broken inline asm
3570 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3571 264800 testcase compile failure on zseries
3572 265762 make public VEX headers compilable by G++ 3.x
3573 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3574 266753 configure script does not give the user the option to not use QtCore
3575 266931 gen_insn_test.pl is broken
3576 266961 ld-linux.so.2 i?86-linux strlen issues
3577 266990 setns instruction causes false positive
3578 267020 Make directory for temporary files configurable at run-time.
3579 267342 == 267997 (segmentation fault on Mac OS 10.6)
3580 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3581 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3582 267488 regtest: darwin support for 64-bit build
3583 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3584 267630 Add support for IBM Power ISA 2.06 -- stage 1
3585 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3586 267819 Add client request for informing the core about reallocation
3587 267925 laog data structure quadratic for a single sequence of lock
3588 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3589 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3590 268513 missed optimizations in fold_Expr
3591 268619 s390x: fpr - gpr transfer facility
3592 268620 s390x: reconsider "long displacement" requirement
3593 268621 s390x: improve IR generation for XC
3594 268715 s390x: FLOGR is not universally available
3595 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3596 268930 s390x: MHY is not universally available
3597 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3598 269079 Support ptrace system call on ARM
3599 269144 missing "Bad option" error message
3600 269209 conditional load and store facility (z196)
3601 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3602 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3603 269736 s390x: minor code generation tweaks
3604 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3605 269863 s390x: remove unused function parameters
3606 269864 s390x: tweak s390_emit_load_cc
3607 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3608 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3609 270115 s390x: rewrite some testcases
3610 270309 == 267997 (valgrind crash on startup)
3611 270320 add support for Linux FIOQSIZE ioctl() call
3612 270326 segfault while trying to sanitize the environment passed to execle
3613 270794 IBM POWER7 support patch causes regression in none/tests
3614 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3615 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3616 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3617 270959 s390x: invalid use of R0 as base register
3618 271042 VSX configure check fails when it should not
3619 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3620 271259 s390x: fix code confusion
3621 271337 == 267997 (Valgrind segfaults on MacOS X)
3622 271385 s390x: Implement Ist_MBE
3623 271501 s390x: misc cleanups
3624 271504 s390x: promote likely and unlikely
3625 271579 ppc: using wrong enum type
3626 271615 unhandled instruction "popcnt" (arch=amd10h)
3627 271730 Fix bug when checking ioctls: duplicate check
3628 271776 s390x: provide STFLE instruction support
3629 271779 s390x: provide clock instructions like STCK
3630 271799 Darwin: ioctls without an arg report a memory error
3631 271820 arm: fix type confusion
3632 271917 pthread_cond_timedwait failure leads to not-locked false positive
3633 272067 s390x: fix DISP20 macro
3634 272615 A typo in debug output in mc_leakcheck.c
3635 272661 callgrind_annotate chokes when run from paths containing regex chars
3636 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3637 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3638 272967 make documentation build-system more robust
3639 272986 Fix gcc-4.6 warnings with valgrind.h
3640 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3641 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3642 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3643 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3644 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3645 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3646 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3647 273778 exp-ptrcheck: unhandled sysno == 259
3648 274089 exp-ptrcheck: unhandled sysno == 208
3649 274378 s390x: Various dispatcher tweaks
3650 274447 WARNING: unhandled syscall: 340
3651 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3652 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3653 274926 valgrind does not build against linux-3
3654 275148 configure FAIL with glibc-2.14
3655 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3656 275168 Make Valgrind work for MacOSX 10.7 Lion
3657 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3658 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3659 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3660 275308 Fix implementation for ppc64 fres instruc
3661 275339 s390x: fix testcase compile warnings
3662 275517 s390x: Provide support for CKSM instruction
3663 275710 s390x: get rid of redundant address mode calculation
3664 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3665 275852 == 250101 (valgrind uses all swap space and is killed)
3666 276784 Add support for IBM Power ISA 2.06 -- stage 3
3667 276987 gdbsrv: fix tests following recent commits
3668 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3669 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3670 277471 Unhandled syscall: 340
3671 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3672 277653 ARM: support Thumb2 PLD instruction
3673 277663 ARM: NEON float VMUL by scalar incorrect
3674 277689 ARM: tests for VSTn with register post-index are broken
3675 277694 ARM: BLX LR instruction broken in ARM mode
3676 277780 ARM: VMOV.F32 (immediate) instruction is broken
3677 278057 fuse filesystem syscall deadlocks
3678 278078 Unimplemented syscall 280 on ppc32
3679 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3680 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3681 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3682 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3683 279027 Support for MVCL and CLCL instruction
3684 279027 s390x: Provide support for CLCL and MVCL instructions
3685 279062 Remove a redundant check in the insn selector for ppc.
3686 279071 JDK creates PTEST with redundant REX.W prefix
3687 279212 gdbsrv: add monitor cmd v.info scheduler.
3688 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3689 279698 memcheck discards valid-bits for packuswb
3690 279795 memcheck reports uninitialised values for mincore on amd64
3691 279994 Add support for IBM Power ISA 2.06 -- stage 3
3692 280083 mempolicy syscall check errors
3693 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3694 280710 s390x: config files for nightly builds
3695 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3696 280965 Valgrind breaks fcntl locks when program does mmap
3697 281138 WARNING: unhandled syscall: 340
3698 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3699 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3700 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3701 281468 s390x: handle do_clone and gcc clones in call traces
3702 281488 ARM: VFP register corruption
3703 281828 == 275284 (false memmove warning: "Source and destination overlap")
3704 281883 s390x: Fix system call wrapper for "clone".
3705 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3706 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3707 282238 SLES10: make check fails
3708 282979 strcasestr needs replacement with recent(>=2.12) glibc
3709 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3710 283243 Regression in ppc64 memcheck tests
3711 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3712 283427 re-connect epoll_pwait syscall on ARM linux
3713 283600 gdbsrv: android: port vgdb.c
3714 283709 none/tests/faultstatus needs to account for page size
3715 284305 filter_gdb needs enhancement to work on ppc64
3716 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3717 284472 Thumb2 ROR.W encoding T2 not implemented
3718 284621 XML-escape process command line in XML output
3719 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3720 that have non-power-of-2 sizes (also AMDs)
3721 n-i-bz don't be spooked by libraries mashed by elfhack
3722 n-i-bz don't be spooked by libxul.so linked with gold
3723 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3725 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3726 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3727 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3731 Release 3.6.1 (16 February 2011)
3732 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3733 3.6.1 is a bug fix release. It adds support for some SSE4
3734 instructions that were omitted in 3.6.0 due to lack of time. Initial
3735 support for glibc-2.13 has been added. A number of bugs causing
3736 crashing or assertion failures have been fixed.
3738 The following bugs have been fixed or resolved. Note that "n-i-bz"
3739 stands for "not in bugzilla" -- that is, a bug that was reported to us
3740 but never got a bugzilla entry. We encourage you to file bugs in
3741 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3742 mailing the developers (or mailing lists) directly -- bugs that are
3743 not entered into bugzilla tend to get forgotten about or ignored.
3745 To see details of a given bug, visit
3746 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3747 where XXXXXX is the bug number as listed below.
3749 188572 Valgrind on Mac should suppress setenv() mem leak
3750 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3751 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3752 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3753 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3754 254420 memory pool tracking broken
3755 254957 Test code failing to compile due to changes in memcheck.h
3756 255009 helgrind/drd: crash on chmod with invalid parameter
3757 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3758 255355 helgrind/drd: crash on threaded programs doing fork
3760 255418 (SSE4.x) rint call compiled with ICC
3761 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3762 255888 closing valgrindoutput tag outputted to log-stream on error
3763 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3764 255966 Slowness when using mempool annotations
3765 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3766 256600 super-optimized strcasecmp() false positive
3767 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3768 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3769 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3770 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3771 257276 Missing case in memcheck --track-origins=yes
3772 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3773 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3774 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3775 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3776 263099 callgrind_annotate counts Ir improperly [...]
3777 263877 undefined coprocessor instruction on ARMv7
3778 265964 configure FAIL with glibc-2.13
3779 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3780 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3781 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3782 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3783 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3785 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3789 Release 3.6.0 (21 October 2010)
3790 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3791 3.6.0 is a feature release with many significant improvements and the
3792 usual collection of bug fixes.
3794 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3795 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3796 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3798 -------------------------
3800 Here are some highlights. Details are shown further down:
3802 * Support for ARM/Linux.
3804 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3806 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3808 * Support for the SSE4.2 instruction set.
3810 * Enhancements to the Callgrind profiler, including the ability to
3811 handle CPUs with three levels of cache.
3813 * A new experimental heap profiler, DHAT.
3815 * A huge number of bug fixes and small enhancements.
3817 -------------------------
3819 Here are details of the above changes, together with descriptions of
3820 many other changes, and a list of fixed bugs.
3822 * ================== PLATFORM CHANGES =================
3824 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3825 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3826 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3828 This requires a CPU capable of running the ARMv7-A instruction set
3829 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3830 of the user space instruction set, including ARM and Thumb integer
3831 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3832 Cachegrind and Massif tools work properly; other tools work to
3835 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3836 with support for recent releases of the underlying toolchain
3837 components, notably gcc-4.5 and glibc-2.12.
3839 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3840 support also works much better on OS X 10.5, and is as solid as
3843 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3844 64-bit mode. In 32-bit mode, support is only available up to and
3845 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3846 supported in 64-bit mode, and 32-bit mode does in fact support the
3847 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3848 10.6 on 32-bit targets.
3850 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3851 and including version 2.05 is supported.
3853 * ==================== TOOL CHANGES ====================
3855 * Cachegrind has a new processing script, cg_diff, which finds the
3856 difference between two profiles. It's very useful for evaluating
3857 the performance effects of a change in a program.
3859 Related to this change, the meaning of cg_annotate's (rarely-used)
3860 --threshold option has changed; this is unlikely to affect many
3861 people, if you do use it please see the user manual for details.
3863 * Callgrind now can do branch prediction simulation, similar to
3864 Cachegrind. In addition, it optionally can count the number of
3865 executed global bus events. Both can be used for a better
3866 approximation of a "Cycle Estimation" as derived event (you need to
3867 update the event formula in KCachegrind yourself).
3869 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3870 rather than the L2 cache. This is to accommodate machines with
3871 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3872 cache configuration of such a machine it will run the simulation as
3873 if the L2 cache isn't present. This means the results are less
3874 likely to match the true result for the machine, but
3875 Cachegrind/Callgrind's results are already only approximate, and
3876 should not be considered authoritative. The results are still
3877 useful for giving a general idea about a program's locality.
3879 * Massif has a new option, --pages-as-heap, which is disabled by
3880 default. When enabled, instead of tracking allocations at the level
3881 of heap blocks (as allocated with malloc/new/new[]), it instead
3882 tracks memory allocations at the level of memory pages (as mapped by
3883 mmap, brk, etc). Each mapped page is treated as its own block.
3884 Interpreting the page-level output is harder than the heap-level
3885 output, but this option is useful if you want to account for every
3886 byte of memory used by a program.
3888 * DRD has two new command-line options: --free-is-write and
3889 --trace-alloc. The former allows to detect reading from already freed
3890 memory, and the latter allows tracing of all memory allocations and
3893 * DRD has several new annotations. Custom barrier implementations can
3894 now be annotated, as well as benign races on static variables.
3896 * DRD's happens before / happens after annotations have been made more
3897 powerful, so that they can now also be used to annotate e.g. a smart
3898 pointer implementation.
3900 * Helgrind's annotation set has also been drastically improved, so as
3901 to provide to users a general set of annotations to describe locks,
3902 semaphores, barriers and condition variables. Annotations to
3903 describe thread-safe reference counted heap objects have also been
3906 * Memcheck has a new command-line option, --show-possibly-lost, which
3907 is enabled by default. When disabled, the leak detector will not
3908 show possibly-lost blocks.
3910 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3911 has been added. DHAT keeps track of allocated heap blocks, and also
3912 inspects every memory reference to see which block (if any) is being
3913 accessed. This gives a lot of insight into block lifetimes,
3914 utilisation, turnover, liveness, and the location of hot and cold
3915 fields. You can use DHAT to do hot-field profiling.
3917 * ==================== OTHER CHANGES ====================
3919 * Improved support for unfriendly self-modifying code: the extra
3920 overhead incurred by --smc-check=all has been reduced by
3921 approximately a factor of 5 as compared with 3.5.0.
3923 * Ability to show directory names for source files in error messages.
3924 This is combined with a flexible mechanism for specifying which
3925 parts of the paths should be shown. This is enabled by the new flag
3928 * A new flag, --require-text-symbol, which will stop the run if a
3929 specified symbol is not found it a given shared object when it is
3930 loaded into the process. This makes advanced working with function
3931 intercepting and wrapping safer and more reliable.
3933 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3934 and control of Valgrind is now available for the tools Memcheck and
3935 Helgrind. XML output from Valgrind is available for Memcheck,
3936 Helgrind and exp-Ptrcheck.
3938 * More reliable stack unwinding on amd64-linux, particularly in the
3939 presence of function wrappers, and with gcc-4.5 compiled code.
3941 * Modest scalability (performance improvements) for massive
3942 long-running applications, particularly for those with huge amounts
3945 * Support for analyzing programs running under Wine with has been
3946 improved. The header files <valgrind/valgrind.h>,
3947 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3948 Windows-programs compiled with MinGW or one of the Microsoft Visual
3951 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3952 The 32-bit simulator was not affected. This did not occur often,
3953 but when it did would usually crash the program under test.
3956 * A large number of bugs were fixed. These are shown below.
3958 * A number of bugs were investigated, and were candidates for fixing,
3959 but are not fixed in 3.6.0, due to lack of developer time. They may
3960 get fixed in later releases. They are:
3962 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3963 212419 false positive "lock order violated" (A+B vs A)
3964 213685 Undefined value propagates past dependency breaking instruction
3965 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3966 237920 valgrind segfault on fork failure
3967 242137 support for code compiled by LLVM-2.8
3968 242423 Another unknown Intel cache config value
3969 243232 Inconsistent Lock Orderings report with trylock
3970 243483 ppc: callgrind triggers VEX assertion failure
3971 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3972 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3974 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3975 249435 Analyzing wine programs with callgrind triggers a crash
3976 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3977 250065 Handling large allocations
3978 250101 huge "free" memory usage due to m_mallocfree.c
3979 "superblocks fragmentation"
3980 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3981 252091 Callgrind on ARM does not detect function returns correctly
3982 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3983 254420 memory pool tracking broken
3984 n-i-bz support for adding symbols for JIT generated code
3987 The following bugs have been fixed or resolved. Note that "n-i-bz"
3988 stands for "not in bugzilla" -- that is, a bug that was reported to us
3989 but never got a bugzilla entry. We encourage you to file bugs in
3990 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3991 mailing the developers (or mailing lists) directly -- bugs that are
3992 not entered into bugzilla tend to get forgotten about or ignored.
3994 To see details of a given bug, visit
3995 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3996 where XXXXXX is the bug number as listed below.
3998 135264 dcbzl instruction missing
4000 153699 Valgrind should report unaligned reads with movdqa
4002 190429 Valgrind reports lost of errors in ld.so
4003 with x86_64 2.9.90 glibc
4004 197266 valgrind appears to choke on the xmms instruction
4006 197988 Crash when demangling very large symbol names
4007 202315 unhandled syscall: 332 (inotify_init1)
4008 203256 Add page-level profiling to Massif
4009 205093 dsymutil=yes needs quotes, locking (partial fix)
4010 205241 Snow Leopard 10.6 support (partial fix)
4011 206600 Leak checker fails to upgrade indirect blocks when their
4012 parent becomes reachable
4013 210935 port valgrind.h (not valgrind) to win32 so apps run under
4014 wine can make client requests
4015 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
4016 within Linux ip-stack checksum functions
4017 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
4019 213685 Undefined value propagates past dependency breaking instruction
4021 215914 Valgrind inserts bogus empty environment variable
4023 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
4024 222545 shmat fails under valgind on some arm targets
4025 222560 ARM NEON support
4028 232509 Docs build fails with formatting inside <title></title> elements
4030 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
4031 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
4032 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
4033 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
4034 237485 symlink (syscall 57) is not supported on Mac OS
4035 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
4037 238208 is_just_below_ESP doesn't take into account red-zone
4038 238345 valgrind passes wrong $0 when executing a shell script
4039 238679 mq_timedreceive syscall doesn't flag the reception buffer
4041 238696 fcntl command F_DUPFD_CLOEXEC not supported
4042 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4043 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4044 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
4046 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
4052 242606 unhandled syscall: setegid (in Ptrcheck)
4053 242814 Helgrind "Impossible has happened" during
4054 QApplication::initInstance();
4055 243064 Valgrind attempting to read debug information from iso
4056 243270 Make stack unwinding in Valgrind wrappers more reliable
4057 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
4058 sysno = 277 (mq_open)
4059 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
4060 244493 ARM VFP d16-d31 registers support
4061 244670 add support for audit_session_self syscall on Mac OS 10.6
4062 244921 The xml report of helgrind tool is not well format
4063 244923 In the xml report file, the <preamble> not escape the
4064 xml char, eg '<','&','>'
4065 245535 print full path names in plain text reports
4066 245925 x86-64 red zone handling problem
4067 246258 Valgrind not catching integer underruns + new [] s
4068 246311 reg/reg cmpxchg doesn't work on amd64
4069 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
4070 246888 Improve Makefile.vex.am
4071 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
4072 to [f]chmod_extended
4073 247526 IBM POWER6 (ISA 2.05) support is incomplete
4074 247561 Some leak testcases fails due to reachable addresses in
4076 247875 sizeofIRType to handle Ity_I128
4077 247894 [PATCH] unhandled syscall sys_readahead
4078 247980 Doesn't honor CFLAGS passed to configure
4079 248373 darwin10.supp is empty in the trunk
4080 248822 Linux FIBMAP ioctl has int parameter instead of long
4081 248893 [PATCH] make readdwarf.c big endianess safe to enable
4082 unwinding on big endian systems
4083 249224 Syscall 336 not supported (SYS_proc_info)
4085 249775 Incorrect scheme for detecting NEON capabilities of host CPU
4086 249943 jni JVM init fails when using valgrind
4087 249991 Valgrind incorrectly declares AESKEYGENASSIST support
4089 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
4090 250799 frexp$fenv_access_off function generates SIGILL
4091 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
4092 251251 support pclmulqdq insn
4093 251362 valgrind: ARM: attach to debugger either fails or provokes
4095 251674 Unhandled syscall 294
4098 254257 Add support for debugfiles found by build-id
4099 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
4100 254646 Wrapped functions cause stack misalignment on OS X
4101 (and possibly Linux)
4102 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
4104 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
4108 Release 3.5.0 (19 August 2009)
4109 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4110 3.5.0 is a feature release with many significant improvements and the
4111 usual collection of bug fixes. The main improvement is that Valgrind
4112 now works on Mac OS X.
4114 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
4115 and X86/Darwin. Support for recent distros and toolchain components
4116 (glibc 2.10, gcc 4.5) has been added.
4118 -------------------------
4120 Here is a short summary of the changes. Details are shown further
4123 * Support for Mac OS X (10.5.x).
4125 * Improvements and simplifications to Memcheck's leak checker.
4127 * Clarification and simplifications in various aspects of Valgrind's
4130 * XML output for Helgrind and Ptrcheck.
4132 * Performance and stability improvements for Helgrind and DRD.
4134 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
4136 * A new experimental tool, BBV, useful for computer architecture
4139 * Improved Wine support, including ability to read Windows PDB
4142 -------------------------
4144 Here are details of the above changes, followed by descriptions of
4145 many other minor changes, and a list of fixed bugs.
4148 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
4149 called "Darwin" because that is the name of the OS core, which is the
4150 level that Valgrind works at.)
4154 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
4155 because it would require work and 10.4 is only becoming less common.
4157 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
4158 fairly well. For 10.5.x, 32-bit programs are the default even on
4159 64-bit machines, so it handles most current programs.
4161 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
4162 officially supported, but simple programs at least will probably work.
4163 However, start-up is slow.
4165 - PowerPC machines are not supported.
4167 Things that don't work:
4169 - The Ptrcheck tool.
4171 - Objective-C garbage collection.
4175 - If you have Rogue Amoeba's "Instant Hijack" program installed,
4176 Valgrind will fail with a SIGTRAP at start-up. See
4177 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
4182 - You will likely find --dsymutil=yes a useful option, as error
4183 messages may be imprecise without it.
4185 - Mac OS X support is new and therefore will be less robust than the
4186 Linux support. Please report any bugs you find.
4188 - Threaded programs may run more slowly than on Linux.
4190 Many thanks to Greg Parker for developing this port over several years.
4193 * Memcheck's leak checker has been improved.
4195 - The results for --leak-check=summary now match the summary results
4196 for --leak-check=full. Previously they could differ because
4197 --leak-check=summary counted "indirectly lost" blocks and
4198 "suppressed" blocks as "definitely lost".
4200 - Blocks that are only reachable via at least one interior-pointer,
4201 but are directly pointed to by a start-pointer, were previously
4202 marked as "still reachable". They are now correctly marked as
4205 - The default value for the --leak-resolution option has been
4206 changed from "low" to "high". In general, this means that more
4207 leak reports will be produced, but each leak report will describe
4208 fewer leaked blocks.
4210 - With --leak-check=full, "definitely lost" and "possibly lost"
4211 leaks are now considered as proper errors, ie. they are counted
4212 for the "ERROR SUMMARY" and affect the behaviour of
4213 --error-exitcode. These leaks are not counted as errors if
4214 --leak-check=summary is specified, however.
4216 - Documentation for the leak checker has been improved.
4219 * Various aspects of Valgrind's text output have changed.
4221 - Valgrind's start-up message has changed. It is shorter but also
4222 includes the command being run, which makes it easier to use
4223 --trace-children=yes. An example:
4225 - Valgrind's shut-down messages have also changed. This is most
4226 noticeable with Memcheck, where the leak summary now occurs before
4227 the error summary. This change was necessary to allow leaks to be
4228 counted as proper errors (see the description of the leak checker
4229 changes above for more details). This was also necessary to fix a
4230 longstanding bug in which uses of suppressions against leaks were
4231 not "counted", leading to difficulties in maintaining suppression
4232 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
4234 - Behavior of -v has changed. In previous versions, -v printed out
4235 a mixture of marginally-user-useful information, and tool/core
4236 statistics. The statistics printing has now been moved to its own
4237 flag, --stats=yes. This means -v is less verbose and more likely
4238 to convey useful end-user information.
4240 - The format of some (non-XML) stack trace entries has changed a
4241 little. Previously there were six possible forms:
4243 0x80483BF: really (a.c:20)
4244 0x80483BF: really (in /foo/a.out)
4246 0x80483BF: (within /foo/a.out)
4247 0x80483BF: ??? (a.c:20)
4250 The third and fourth of these forms have been made more consistent
4251 with the others. The six possible forms are now:
4253 0x80483BF: really (a.c:20)
4254 0x80483BF: really (in /foo/a.out)
4255 0x80483BF: really (in ???)
4256 0x80483BF: ??? (in /foo/a.out)
4257 0x80483BF: ??? (a.c:20)
4260 Stack traces produced when --xml=yes is specified are different
4264 * Helgrind and Ptrcheck now support XML output, so they can be used
4265 from GUI tools. Also, the XML output mechanism has been
4268 - The XML format has been overhauled and generalised, so it is more
4269 suitable for error reporting tools in general. The Memcheck
4270 specific aspects of it have been removed. The new format, which
4271 is an evolution of the old format, is described in
4272 docs/internals/xml-output-protocol4.txt.
4274 - Memcheck has been updated to use the new format.
4276 - Helgrind and Ptrcheck are now able to emit output in this format.
4278 - The XML output mechanism has been overhauled. XML is now output
4279 to its own file descriptor, which means that:
4281 * Valgrind can output text and XML independently.
4283 * The longstanding problem of XML output being corrupted by
4284 unexpected un-tagged text messages is solved.
4286 As before, the destination for text output is specified using
4287 --log-file=, --log-fd= or --log-socket=.
4289 As before, XML output for a tool is enabled using --xml=yes.
4291 Because there's a new XML output channel, the XML output
4292 destination is now specified by --xml-file=, --xml-fd= or
4295 Initial feedback has shown this causes some confusion. To
4296 clarify, the two envisaged usage scenarios are:
4298 (1) Normal text output. In this case, do not specify --xml=yes
4299 nor any of --xml-file=, --xml-fd= or --xml-socket=.
4301 (2) XML output. In this case, specify --xml=yes, and one of
4302 --xml-file=, --xml-fd= or --xml-socket= to select the XML
4303 destination, one of --log-file=, --log-fd= or --log-socket=
4304 to select the destination for any remaining text messages,
4305 and, importantly, -q.
4307 -q makes Valgrind completely silent on the text channel,
4308 except in the case of critical failures, such as Valgrind
4309 itself segfaulting, or failing to read debugging information.
4310 Hence, in this scenario, it suffices to check whether or not
4311 any output appeared on the text channel. If yes, then it is
4312 likely to be a critical error which should be brought to the
4313 attention of the user. If no (the text channel produced no
4314 output) then it can be assumed that the run was successful.
4316 This allows GUIs to make the critical distinction they need to
4317 make (did the run fail or not?) without having to search or
4318 filter the text output channel in any way.
4320 It is also recommended to use --child-silent-after-fork=yes in
4324 * Improvements and changes in Helgrind:
4326 - XML output, as described above
4328 - Checks for consistent association between pthread condition
4329 variables and their associated mutexes are now performed.
4331 - pthread_spinlock functions are supported.
4333 - Modest performance improvements.
4335 - Initial (skeletal) support for describing the behaviour of
4336 non-POSIX synchronisation objects through ThreadSanitizer
4337 compatible ANNOTATE_* macros.
4339 - More controllable tradeoffs between performance and the level of
4340 detail of "previous" accesses in a race. There are now three
4343 * --history-level=full. This is the default, and was also the
4344 default in 3.4.x. It shows both stacks involved in a race, but
4345 requires a lot of memory and can be very slow in programs that
4346 do many inter-thread synchronisation events.
4348 * --history-level=none. This only shows the later stack involved
4349 in a race. This can be much faster than --history-level=full,
4350 but makes it much more difficult to find the other access
4351 involved in the race.
4353 The new intermediate setting is
4355 * --history-level=approx
4357 For the earlier (other) access, two stacks are presented. The
4358 earlier access is guaranteed to be somewhere in between the two
4359 program points denoted by those stacks. This is not as useful
4360 as showing the exact stack for the previous access (as per
4361 --history-level=full), but it is better than nothing, and it's
4362 almost as fast as --history-level=none.
4365 * New features and improvements in DRD:
4367 - The error messages printed by DRD are now easier to interpret.
4368 Instead of using two different numbers to identify each thread
4369 (Valgrind thread ID and DRD thread ID), DRD does now identify
4370 threads via a single number (the DRD thread ID). Furthermore
4371 "first observed at" information is now printed for all error
4372 messages related to synchronization objects.
4374 - Added support for named semaphores (sem_open() and sem_close()).
4376 - Race conditions between pthread_barrier_wait() and
4377 pthread_barrier_destroy() calls are now reported.
4379 - Added support for custom allocators through the macros
4380 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
4381 in <valgrind/valgrind.h>). An alternative for these two macros is
4382 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
4385 - Added support for annotating non-POSIX synchronization objects
4386 through several new ANNOTATE_*() macros.
4388 - OpenMP: added support for the OpenMP runtime (libgomp) included
4389 with gcc versions 4.3.0 and 4.4.0.
4393 - Added two new command-line options (--first-race-only and
4394 --segment-merging-interval).
4397 * Genuinely atomic support for x86/amd64/ppc atomic instructions
4399 Valgrind will now preserve (memory-access) atomicity of LOCK-
4400 prefixed x86/amd64 instructions, and any others implying a global
4401 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
4403 This means that Valgrinded processes will "play nicely" in
4404 situations where communication with other processes, or the kernel,
4405 is done through shared memory and coordinated with such atomic
4406 instructions. Prior to this change, such arrangements usually
4407 resulted in hangs, races or other synchronisation failures, because
4408 Valgrind did not honour atomicity of such instructions.
4411 * A new experimental tool, BBV, has been added. BBV generates basic
4412 block vectors for use with the SimPoint analysis tool, which allows
4413 a program's overall behaviour to be approximated by running only a
4414 fraction of it. This is useful for computer architecture
4415 researchers. You can run BBV by specifying --tool=exp-bbv (the
4416 "exp-" prefix is short for "experimental"). BBV was written by
4420 * Modestly improved support for running Windows applications under
4421 Wine. In particular, initial support for reading Windows .PDB debug
4422 information has been added.
4425 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
4426 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
4430 * The Valgrind client requests VALGRIND_PRINTF and
4431 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
4432 the string was always printed immediately on its own line. Now, the
4433 string will be added to a buffer but not printed until a newline is
4434 encountered, or other Valgrind output is printed (note that for
4435 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
4436 "other Valgrind output"). This allows you to use multiple
4437 VALGRIND_PRINTF calls to build up a single output line, and also to
4438 print multiple output lines with a single request (by embedding
4439 multiple newlines in the string).
4442 * The graphs drawn by Massif's ms_print program have changed slightly:
4444 - The half-height chars '.' and ',' are no longer drawn, because
4445 they are confusing. The --y option can be used if the default
4446 y-resolution is not high enough.
4448 - Horizontal lines are now drawn after the top of a snapshot if
4449 there is a gap until the next snapshot. This makes it clear that
4450 the memory usage has not dropped to zero between snapshots.
4453 * Something that happened in 3.4.0, but wasn't clearly announced: the
4454 option --read-var-info=yes can be used by some tools (Memcheck,
4455 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
4456 variable type and location information. This makes those tools
4457 start up more slowly and increases memory consumption, but
4458 descriptions of data addresses in error messages become more
4462 * exp-Omega, an experimental instantaneous leak-detecting tool, was
4463 disabled in 3.4.0 due to a lack of interest and maintenance,
4464 although the source code was still in the distribution. The source
4465 code has now been removed from the distribution. For anyone
4466 interested, the removal occurred in SVN revision r10247.
4469 * Some changes have been made to the build system.
4471 - VEX/ is now integrated properly into the build system. This means
4472 that dependency tracking within VEX/ now works properly, "make
4473 install" will work without requiring "make" before it, and
4474 parallel builds (ie. 'make -j') now work (previously a
4475 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
4476 was effectively ignored).
4478 - The --with-vex configure option has been removed. It was of
4479 little use and removing it simplified the build system.
4481 - The location of some install files has changed. This should not
4482 affect most users. Those who might be affected:
4484 * For people who use Valgrind with MPI programs, the installed
4485 libmpiwrap.so library has moved from
4486 $(INSTALL)/<platform>/libmpiwrap.so to
4487 $(INSTALL)/libmpiwrap-<platform>.so.
4489 * For people who distribute standalone Valgrind tools, the
4490 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
4491 have moved to $(INSTALL)/libcoregrind-<platform>.a.
4493 These changes simplify the build system.
4495 - Previously, all the distributed suppression (*.supp) files were
4496 installed. Now, only default.supp is installed. This should not
4497 affect users as the other installed suppression files were not
4498 read; the fact that they were installed was a mistake.
4501 * KNOWN LIMITATIONS:
4503 - Memcheck is unusable with the Intel compiler suite version 11.1,
4504 when it generates code for SSE2-and-above capable targets. This
4505 is because of icc's use of highly optimised inlined strlen
4506 implementations. It causes Memcheck to report huge numbers of
4507 false errors even in simple programs. Helgrind and DRD may also
4510 Versions 11.0 and earlier may be OK, but this has not been
4514 The following bugs have been fixed or resolved. Note that "n-i-bz"
4515 stands for "not in bugzilla" -- that is, a bug that was reported to us
4516 but never got a bugzilla entry. We encourage you to file bugs in
4517 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4518 mailing the developers (or mailing lists) directly -- bugs that are
4519 not entered into bugzilla tend to get forgotten about or ignored.
4521 To see details of a given bug, visit
4522 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4523 where XXXXXX is the bug number as listed below.
4525 84303 How about a LockCheck tool?
4526 91633 dereference of null ptr in vgPlain_st_basetype
4527 97452 Valgrind doesn't report any pthreads problems
4528 100628 leak-check gets assertion failure when using
4529 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4530 108528 NPTL pthread cleanup handlers not called
4531 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4532 110128 mallinfo is not implemented...
4533 110770 VEX: Generated files not always updated when making valgrind
4534 111102 Memcheck: problems with large (memory footprint) applications
4535 115673 Vex's decoder should never assert
4536 117564 False positive: Syscall param clone(child_tidptr) contains
4537 uninitialised byte(s)
4538 119404 executing ssh from inside valgrind fails
4539 133679 Callgrind does not write path names to sources with dwarf debug
4541 135847 configure.in problem with non gnu compilers (and possible fix)
4542 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4543 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4544 136230 memcheck reports "possibly lost", should be "still reachable"
4545 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4546 137904 Valgrind reports a memory leak when using POSIX threads,
4548 139076 valgrind VT_GETSTATE error
4549 142228 complaint of elf_dynamic_do_rela in trivial usage
4550 145347 spurious warning with USBDEVFS_REAPURB
4551 148441 (wine) can't find memory leak in Wine, win32 binary
4553 148742 Leak-check fails assert on exit
4554 149878 add (proper) check for calloc integer overflow
4555 150606 Call graph is broken when using callgrind control
4556 152393 leak errors produce an exit code of 0. I need some way to
4557 cause leak errors to result in a nonzero exit code.
4558 157154 documentation (leak-resolution doc speaks about num-callers
4559 def=4) + what is a loss record
4560 159501 incorrect handling of ALSA ioctls
4561 162020 Valgrinding an empty/zero-byte file crashes valgrind
4562 162482 ppc: Valgrind crashes while reading stabs information
4563 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4564 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4565 163560 VEX/test_main.c is missing from valgrind-3.3.1
4566 164353 malloc_usable_size() doesn't return a usable size
4567 165468 Inconsistent formatting in memcheck manual -- please fix
4568 169505 main.c:286 (endOfInstr):
4569 Assertion 'ii->cost_offset == *cost_offset' failed
4570 177206 Generate default.supp during compile instead of configure
4571 177209 Configure valt_load_address based on arch+os
4572 177305 eventfd / syscall 323 patch lost
4573 179731 Tests fail to build because of inlining of non-local asm labels
4574 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4575 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4576 181594 Bogus warning for empty text segment
4577 181707 dwarf doesn't require enumerations to have name
4578 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4579 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4580 Assertion '!already_present' failed.
4581 185359 exp-ptrcheck: unhandled syscall getresuid()
4582 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4583 185816 Valgrind is unable to handle debug info for files with split
4584 debug info that are prelinked afterwards
4585 185980 [darwin] unhandled syscall: sem_open
4586 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4587 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4588 186790 Suppression pattern used for leaks are not reported
4589 186796 Symbols with length>200 in suppression files are ignored
4590 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4591 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4592 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4593 188046 bashisms in the configure script
4594 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4595 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4596 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4597 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4598 assertion fail "!lock->heldBy"
4599 188427 Add support for epoll_create1 (with patch)
4600 188530 Support for SIOCGSTAMPNS
4601 188560 Include valgrind.spec in the tarball
4602 188572 Valgrind on Mac should suppress setenv() mem leak
4603 189054 Valgrind fails to build because of duplicate non-local asm labels
4604 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4605 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4606 189763 drd assertion failure: s_threadinfo[tid].is_recording
4607 190219 unhandled syscall: 328 (x86-linux)
4608 190391 dup of 181394; see above
4609 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4610 190820 No debug information on powerpc-linux
4611 191095 PATCH: Improve usbdevfs ioctl handling
4612 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4614 191189 --xml=yes should obey --gen-suppressions=all
4615 191192 syslog() needs a suppression on macosx
4616 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4617 191761 getrlimit on MacOSX
4618 191992 multiple --fn-skip only works sometimes; dependent on order
4619 192634 V. reports "aspacem sync_check_mapping_callback:
4620 segment mismatch" on Darwin
4621 192954 __extension__ missing on 2 client requests
4622 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4623 194474 "INSTALL" file has different build instructions than "README"
4624 194671 Unhandled syscall (sem_wait?) from mac valgrind
4625 195069 memcheck: reports leak (memory still reachable) for
4627 195169 drd: (vgDrd_barrier_post_wait):
4628 Assertion 'r->sg[p->post_iteration]' failed.
4629 195268 valgrind --log-file doesn't accept ~/...
4630 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4631 195860 WARNING: unhandled syscall: unix:223
4632 196528 need a error suppression for pthread_rwlock_init under os x?
4633 197227 Support aio_* syscalls on Darwin
4634 197456 valgrind should reject --suppressions=(directory)
4635 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4636 197591 unhandled syscall 27 (mincore)
4637 197793 Merge DCAS branch to the trunk == 85756, 142103
4638 197794 Avoid duplicate filenames in Vex
4639 197898 make check fails on current SVN
4640 197901 make check fails also under exp-ptrcheck in current SVN
4641 197929 Make --leak-resolution=high the default
4642 197930 Reduce spacing between leak reports
4643 197933 Print command line of client at start-up, and shorten preamble
4644 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4645 198395 add BBV to the distribution as an experimental tool
4646 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4647 198649 callgrind_annotate doesn't cumulate counters
4648 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4649 199977 Valgrind complains about an unrecognized instruction in the
4650 atomic_incs test program
4651 200029 valgrind isn't able to read Fedora 12 debuginfo
4652 200760 darwin unhandled syscall: unix:284
4653 200827 DRD doesn't work on Mac OS X
4654 200990 VG_(read_millisecond_timer)() does not work correctly
4655 201016 Valgrind does not support pthread_kill() on Mac OS
4656 201169 Document --read-var-info
4657 201323 Pre-3.5.0 performance sanity checking
4658 201384 Review user manual for the 3.5.0 release
4659 201585 mfpvr not implemented on ppc
4660 201708 tests failing because x86 direction flag is left set
4661 201757 Valgrind doesn't handle any recent sys_futex additions
4662 204377 64-bit valgrind can not start a shell script
4663 (with #!/path/to/shell) if the shell is a 32-bit executable
4664 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4665 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4666 about memory allocation events with memory access tracing enabled
4667 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4669 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4673 Release 3.4.1 (28 February 2009)
4674 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4675 3.4.1 is a bug-fix release that fixes some regressions and assertion
4676 failures in debug info reading in 3.4.0, most notably incorrect stack
4677 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4678 other debug info problems are also fixed. A number of bugs in the
4679 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4681 In view of the fact that 3.4.0 contains user-visible regressions
4682 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4683 encouraged to ship 3.4.1 in preference to 3.4.0.
4685 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4686 bugzilla" -- that is, a bug that was reported to us but never got a
4687 bugzilla entry. We encourage you to file bugs in bugzilla
4688 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4689 developers (or mailing lists) directly -- bugs that are not entered
4690 into bugzilla tend to get forgotten about or ignored.
4692 n-i-bz Fix various bugs reading icc-11 generated debug info
4693 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4694 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4695 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4696 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4697 179618 exp-ptrcheck crashed / exit prematurely
4698 179624 helgrind: false positive races with pthread_create and
4699 recv/open/close/read
4700 134207 pkg-config output contains @VG_PLATFORM@
4701 176926 floating point exception at valgrind startup with PPC 440EPX
4702 181594 Bogus warning for empty text segment
4703 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4704 181707 Dwarf3 doesn't require enumerations to have name
4705 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4706 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4707 Assertion '!already_present' failed.
4708 185359 exp-ptrcheck unhandled syscall getresuid()
4710 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4711 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4715 Release 3.4.0 (2 January 2009)
4716 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4717 3.4.0 is a feature release with many significant improvements and the
4718 usual collection of bug fixes. This release supports X86/Linux,
4719 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4720 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4722 3.4.0 brings some significant tool improvements. Memcheck can now
4723 report the origin of uninitialised values, the thread checkers
4724 Helgrind and DRD are much improved, and we have a new experimental
4725 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4726 global arrays. In detail:
4728 * Memcheck is now able to track the origin of uninitialised values.
4729 When it reports an uninitialised value error, it will try to show
4730 the origin of the value, as either a heap or stack allocation.
4731 Origin tracking is expensive and so is not enabled by default. To
4732 use it, specify --track-origins=yes. Memcheck's speed will be
4733 essentially halved, and memory usage will be significantly
4734 increased. Nevertheless it can drastically reduce the effort
4735 required to identify the root cause of uninitialised value errors,
4736 and so is often a programmer productivity win, despite running more
4739 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4740 3.4.0, will be released shortly.
4742 * Helgrind's race detection algorithm has been completely redesigned
4743 and reimplemented, to address usability and scalability concerns:
4745 - The new algorithm has a lower false-error rate: it is much less
4746 likely to report races that do not really exist.
4748 - Helgrind will display full call stacks for both accesses involved
4749 in a race. This makes it easier to identify the root causes of
4752 - Limitations on the size of program that can run have been removed.
4754 - Performance has been modestly improved, although that is very
4757 - Direct support for Qt4 threading has been added.
4759 - pthread_barriers are now directly supported.
4761 - Helgrind works well on all supported Linux targets.
4763 * The DRD thread debugging tool has seen major improvements:
4765 - Greatly improved performance and significantly reduced memory
4768 - Support for several major threading libraries (Boost.Thread, Qt4,
4769 glib, OpenMP) has been added.
4771 - Support for atomic instructions, POSIX semaphores, barriers and
4772 reader-writer locks has been added.
4774 - Works now on PowerPC CPUs too.
4776 - Added support for printing thread stack usage at thread exit time.
4778 - Added support for debugging lock contention.
4780 - Added a manual for Drd.
4782 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4783 checks for misuses of pointers. In that sense it is a bit like
4784 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4785 detect overruns of stack and global arrays, it can detect
4786 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4787 detect accesses heap blocks that have been freed a very long time
4788 ago (millions of blocks in the past).
4790 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4791 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4792 of the main Valgrind documentation. As this is an experimental
4793 tool, we would be particularly interested in hearing about your
4794 experiences with it.
4796 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4797 longer built by default, although the code remains in the repository
4798 and the tarball. This is due to three factors: a perceived lack of
4799 users, a lack of maintenance, and concerns that it may not be
4800 possible to achieve reliable operation using the existing design.
4802 * As usual, support for the latest Linux distros and toolchain
4803 components has been added. It should work well on Fedora Core 10,
4804 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4805 state) is supported, as is glibc-2.9. The C++ demangler has been
4806 updated so as to work well with C++ compiled by even the most recent
4809 * You can now use frame-level wildcards in suppressions. This was a
4810 frequently-requested enhancement. A line "..." in a suppression now
4811 matches zero or more frames. This makes it easier to write
4812 suppressions which are precise yet insensitive to changes in
4815 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4817 * Very basic support for IBM Power6 has been added (64-bit processes only).
4819 * Valgrind is now cross-compilable. For example, it is possible to
4820 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4821 on a ppc32/64-linux target.
4823 * You can set the main thread's stack size at startup using the
4824 new --main-stacksize= flag (subject of course to ulimit settings).
4825 This is useful for running apps that need a lot of stack space.
4827 * The limitation that you can't use --trace-children=yes together
4828 with --db-attach=yes has been removed.
4830 * The following bugs have been fixed. Note that "n-i-bz" stands for
4831 "not in bugzilla" -- that is, a bug that was reported to us but
4832 never got a bugzilla entry. We encourage you to file bugs in
4833 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4834 mailing the developers (or mailing lists) directly.
4836 n-i-bz Make return types for some client requests 64-bit clean
4837 n-i-bz glibc 2.9 support
4838 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4839 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4840 n-i-bz Building in an env without gdb gives bogus gdb attach
4841 92456 Tracing the origin of uninitialised memory
4842 106497 Valgrind does not demangle some C++ template symbols
4844 151612 Suppression with "..." (frame-level wildcards in .supp files)
4845 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4846 159285 unhandled syscall:25 (stime, on x86-linux)
4847 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4848 160954 ppc build of valgrind crashes with illegal instruction (isel)
4849 160956 mallinfo implementation, w/ patch
4850 162092 Valgrind fails to start gnome-system-monitor
4851 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4852 163794 assertion failure with "--track-origins=yes"
4853 163933 sigcontext.err and .trapno must be set together
4854 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4855 164476 Missing kernel module loading system calls
4856 164669 SVN regression: mmap() drops posix file locks
4857 166581 Callgrind output corruption when program forks
4858 167288 Patch file for missing system calls on Cell BE
4859 168943 unsupported scas instruction pentium
4860 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4861 172417 x86->IR: 0x82 ...
4862 172563 amd64->IR: 0xD9 0xF5 - fprem1
4863 173099 .lds linker script generation error
4864 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4865 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4867 174908 --log-file value not expanded correctly for core file
4868 175044 Add lookup_dcookie for amd64
4869 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4871 Developer-visible changes:
4873 * Valgrind's debug-info reading machinery has been majorly overhauled.
4874 It can now correctly establish the addresses for ELF data symbols,
4875 which is something that has never worked properly before now.
4877 Also, Valgrind can now read DWARF3 type and location information for
4878 stack and global variables. This makes it possible to use the
4879 framework to build tools that rely on knowing the type and locations
4880 of stack and global variables, for example exp-Ptrcheck.
4882 Reading of such information is disabled by default, because most
4883 tools don't need it, and because it is expensive in space and time.
4884 However, you can force Valgrind to read it, using the
4885 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4886 make use of such information, if present, to provide source-level
4887 descriptions of data addresses in the error messages they create.
4889 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4890 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).