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 an distro could use it to install the scripts in a
16 safe load location --with-gdbscripts-dir=%{_datadir}/gdb/auto-load
17 It is also possible to configure --without-gdb-scripts-dir so no
18 .debug_gdb_scripts section is added to the vgpreload library and
19 no valgrind-monitor python scripts are installed at all.
21 * ================== PLATFORM CHANGES =================
24 * ==================== TOOL CHANGES ===================
27 * ==================== FIXED BUGS ====================
29 The following bugs have been fixed or resolved. Note that "n-i-bz"
30 stands for "not in bugzilla" -- that is, a bug that was reported to us
31 but never got a bugzilla entry. We encourage you to file bugs in
32 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
33 than mailing the developers (or mailing lists) directly -- bugs that
34 are not entered into bugzilla tend to get forgotten about or ignored.
36 426751 Valgrind reports «still reachable» memory using musl (alpine running inside docker)
37 460192 Add epoll_pwait2
38 469049 link failure on ppc64 (big endian) valgrind 3.20
39 469146 massif --ignore-fn does not ignore inlined functions
40 469768 Make it possible to install gdb scripts in a different location
41 470121 Can't run callgrind_control with valgrind 3.21.0 because of perl errors
42 470132 s390x: Assertion failure on VGM instruction
43 470520 Multiple realloc zero errors crash in MC_(eq_Error)
44 470713 Failure on the Yosys project: valgrind: m_libcfile.c:1802
45 (Bool vgPlain_realpath(const HChar *, HChar *)):
46 Assertion 'resolved' failed
47 470830 Don't print actions vgdb me ... continue for vgdb --multi mode
48 470978 s390x: Valgrind cannot start qemu-kvm when "sysctl vm.allocate_pgste=0"
49 471311 gdb --multi mode stdout redirecting to stderr
50 471807 Add support for lazy reading and downloading of DWARF debuginfo
51 472219 Syscall param ppoll(ufds.events) points to uninitialised byte(s)
52 472963 Broken regular expression in configure.ac
53 473604 Fix bug472219.c compile failure with Clang 16
55 To see details of a given bug, visit
56 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
57 where XXXXXX is the bug number as listed above.
59 Release 3.21.0 (28 Apr 2023)
60 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
62 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
63 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
64 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
65 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
66 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
67 AMD64/macOS 10.13 and nanoMIPS/Linux.
69 * ==================== CORE CHANGES ===================
71 * When GDB is used to debug a program running under valgrind using
72 the valgrind gdbserver, GDB will automatically load some
73 python code provided in valgrind defining GDB front end commands
74 corresponding to the valgrind monitor commands.
75 These GDB front end commands accept the same format as
76 the monitor commands directly sent to the Valgrind gdbserver.
77 These GDB front end commands provide a better integration
78 in the GDB command line interface, so as to use for example
79 GDB auto-completion, command specific help, searching for
80 a command or command help matching a regexp, ...
81 For relevant monitor commands, GDB will evaluate arguments
82 to make the use of monitor commands easier.
83 For example, instead of having to print the address of a variable
84 to pass it to a subsequent monitor command, the GDB front end
85 command will evaluate the address argument. It is for example
87 (gdb) memcheck who_points_at &some_struct sizeof(some_struct)
90 $2 = (some_struct_type *) 0x1130a0 <some_struct>
91 (gdb) p sizeof(some_struct)
93 (gdb) monitor who_point_at 0x1130a0 40
95 * The vgdb utility now supports extended-remote protocol when
96 invoked with --multi. In this mode the GDB run command is
97 supported. Which means you don't need to run gdb and valgrind
98 from different terminals. So for example to start your program
99 in gdb and run it under valgrind you can do:
101 (gdb) set remote exec-file prog
103 (gdb) target extended-remote | vgdb --multi
106 * The behaviour of realloc with a size of zero can now
107 be changed for tools that intercept malloc. Those
108 tools are memcheck, helgrind, drd, massif and dhat.
109 Realloc implementations generally do one of two things
110 - free the memory like free() and return NULL
111 (GNU libc and ptmalloc).
112 - either free the memory and then allocate a
113 minimum sized block or just return the
114 original pointer. Return NULL if the
115 allocation of the minimum sized block fails
116 (jemalloc, musl, snmalloc, Solaris, macOS).
117 When Valgrind is configured and built it will
118 try to match the OS and libc behaviour. However
119 if you are using a non-default library to replace
120 malloc and family (e.g., musl on a glibc Linux or
121 tcmalloc on FreeBSD) then you can use a command line
122 option to change the behaviour of Valgrind:
123 --realloc-zero-bytes-frees=yes|no [yes on Linux glibc, no otherwise]
125 * ================== PLATFORM CHANGES =================
127 * Make the address space limit on FreeBSD amd64 128Gbytes
128 (the same as Linux and Solaris, it was 32Gbytes)
130 * ==================== TOOL CHANGES ===================
133 - When doing a delta leak_search, it is now possible to only
134 output the new loss records compared to the previous leak search.
135 This is available in the memcheck monitor command 'leak_search'
136 by specifying the "new" keyword or in your program by using
137 the client request VALGRIND_DO_NEW_LEAK_CHECK.
138 Whenever a "delta" leak search is done (i.e. when specifying
139 "new" or "increased" or "changed" in the monitor command),
140 the new loss records have a "new" marker.
141 - Valgrind now contains python code that defines GDB memcheck
142 front end monitor commands. See CORE CHANGES.
143 - Performs checks for the use of realloc with a size of zero.
144 This is non-portable and a source of errors. If memcheck
145 detects such a usage it will generate an error
146 realloc() with size 0
147 followed by the usual callstacks.
148 A switch has been added to allow this to be turned off:
149 --show-realloc-size-zero=yes|no [yes]
152 - The option ---history-backtrace-size=<number> allows to configure
153 the number of entries to record in the stack traces of "old"
154 accesses. Previously, this number was hardcoded to 8.
155 - Valgrind now contains python code that defines GDB helgrind
156 front end monitor commands. See CORE CHANGES.
159 - `--cache-sim=no` is now the default. The cache simulation is old and
160 unlikely to match any real modern machine. This means only the `Ir`
161 event are gathered by default, but that is by far the most useful
163 - `cg_annotate`, `cg_diff`, and `cg_merge` have been rewritten in
164 Python. As a result, they all have more flexible command line
165 argument handling, e.g. supporting `--show-percs` and
166 `--no-show-percs` forms as well as the existing `--show-percs=yes`
167 and `--show-percs=no`.
168 - `cg_annotate` has some functional changes.
169 - It's much faster, e.g. 3-4x on common cases.
170 - It now supports diffing (with `--diff`, `--mod-filename`, and
171 `--mod-funcname`) and merging (by passing multiple data files).
172 - It now provides more information at the file and function level.
173 There are now "File:function" and "Function:file" sections. These
174 are very useful for programs that use inlining a lot.
175 - Support for user-annotated files and the `-I`/`--include` option
176 has been removed, because it was of little use and blocked other
178 - The `--auto` option is renamed `--annotate`, though the old
179 `--auto=yes`/`--auto=no` forms are still supported.
180 - `cg_diff` and `cg_merge` are now deprecated, because `cg_annotate`
181 now does a better job of diffing and merging.
182 - The Cachegrind output file format has changed very slightly, but in
183 ways nobody is likely to notice.
186 - Valgrind now contains python code that defines GDB callgrind
187 front end monitor commands. See CORE CHANGES.
190 - Valgrind now contains python code that defines GDB massif
191 front end monitor commands. See CORE CHANGES.
194 - A new kind of user request has been added which allows you to
195 override the 1024 byte limit on access count histograms for blocks
196 of memory. The client request is DHAT_HISTOGRAM_MEMORY.
198 * ==================== FIXED BUGS ====================
200 The following bugs have been fixed or resolved. Note that "n-i-bz"
201 stands for "not in bugzilla" -- that is, a bug that was reported to us
202 but never got a bugzilla entry. We encourage you to file bugs in
203 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
204 than mailing the developers (or mailing lists) directly -- bugs that
205 are not entered into bugzilla tend to get forgotten about or ignored.
207 170510 Don't warn about ioctl of size 0 without direction hint
208 241072 List tools in --help output
209 327548 false positive while destroying mutex
210 382034 Testcases build fixes for musl
211 351857 confusing error message about valid command line option
212 374596 inconsistent RDTSCP support on x86_64
213 392331 Spurious lock not held error from inside pthread_cond_timedwait
214 397083 Likely false positive "uninitialised value(s)" for __wmemchr_avx2 and __wmemcmp_avx2_movbe
215 400793 pthread_rwlock_timedwrlock false positive
216 419054 Unhandled syscall getcpu on arm32
217 433873 openat2 syscall unimplemented on Linux
218 434057 Add stdio mode to valgrind's gdbserver
219 435441 valgrind fails to interpose malloc on musl 1.2.2 due to weak symbol name and no libc soname
220 436413 Warn about realloc of size zero
221 439685 compiler warning in callgrind/main.c
222 444110 priv/guest_ppc_toIR.c:36198:31: warning: duplicated 'if' condition.
223 444487 hginfo test detects an extra lock inside data symbol "_rtld_local"
224 444488 Use glibc.pthread.stack_cache_size tunable
225 444568 drd/tests/pth_barrier_thr_cr fails on Fedora 38
226 445743 "The impossible happened: mutex is locked simultaneously by two threads"
227 while using mutexes with priority inheritance and signals
228 449309 Missing loopback device ioctl(s)
229 459476 vgdb: allow address reuse to avoid "address already in use" errorsuse" errors
230 460356 s390: Sqrt32Fx4 -- cannot reduce tree
231 462830 WARNING: unhandled amd64-freebsd syscall: 474
232 463027 broken check for MPX instruction support in assembler
233 464103 Enhancement: add a client request to DHAT to mark memory to be histogrammed
234 464476 Firefox fails to start under Valgrind
235 464609 Valgrind memcheck should support Linux pidfd_open
236 464680 Show issues caused by memory policies like selinux deny_execmem
237 464859 Build failures with GCC-13 (drd tsan_unittest)
238 464969 D language demangling
239 465435 m_libcfile.c:66 (vgPlain_safe_fd): Assertion 'newfd >= VG_(fd_hard_limit)' failed.
240 466104 aligned_alloc problems, part 1
241 467036 Add time cost statistics for Regtest
242 467482 Build failure on aarch64 Alpine
243 467714 fdleak_* and rlimit tests fail when parent process has more than
244 64 descriptors opened
245 467839 Gdbserver: Improve compatibility of library directory name
246 468401 [PATCH] Add a style file for clang-format
247 468556 Build failure for vgdb
248 468606 build: remove "Valgrind relies on GCC" check/output
249 469097 ppc64(be) doesn't support SCV syscall instruction
250 n-i-bz FreeBSD rfork syscall fail with EINVAL or ENOSYS rather than VG_(unimplemented)
252 To see details of a given bug, visit
253 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
254 where XXXXXX is the bug number as listed above.
256 * ==================== KNOWN ISSUES ===================
258 * configure --enable-lto=yes is know to not work in all setups.
259 See bug 469049. Workaround: Build without LTO.
261 (3.21.0.RC1: 14 Apr 2023)
262 (3.21.0.RC2: 21 Apr 2023)
264 Release 3.20.0 (24 Oct 2022)
265 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
267 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
268 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
269 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
270 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
271 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
272 AMD64/macOS 10.13 and nanoMIPS/Linux.
274 * ==================== CORE CHANGES ===================
276 * The option "--vgdb-stop-at=event1,event2,..." accepts the new value abexit.
277 This indicates to invoke gdbserver when your program exits abnormally
278 (i.e. with a non zero exit code).
279 * Fix Rust v0 name demangling.
280 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
281 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
282 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with
284 * The option --enable-debuginfod=<no|yes> [default: yes] has been added on
286 * More DWARF5 support as generated by clang14.
288 * ==================== FIXED BUGS ====================
290 The following bugs have been fixed or resolved. Note that "n-i-bz"
291 stands for "not in bugzilla" -- that is, a bug that was reported to us
292 but never got a bugzilla entry. We encourage you to file bugs in
293 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
294 than mailing the developers (or mailing lists) directly -- bugs that
295 are not entered into bugzilla tend to get forgotten about or ignored.
297 131186 writev reports error in (vector[...])
298 434764 iconv_open causes ld.so v2.28+ to use optimised strncmp
299 446754 Improve error codes from alloc functions under memcheck
300 452274 memcheck crashes with Assertion 'sci->status.what == SsIdle' failed
301 452779 Valgrind fails to build on FreeBSD 13.0 with llvm-devel (15.0.0)
302 453055 shared_timed_mutex drd test fails with "Lock shared failed" message
303 453602 Missing command line option to enable/disable debuginfod
304 452802 Handle lld 9+ split RW PT_LOAD segments correctly
305 454040 s390x: False-positive memcheck:cond in memmem on arch13 systems
306 456171 [PATCH] FreeBSD: Don't record address errors when accessing the 'kern.ps_strings' sysctl struct
307 n-i-bz Implement vgdb invoker on FreeBSD
308 458845 PowerPC: The L field for the dcbf and sync instruction should be
310 458915 Remove register cache to fix 458915 gdbserver causes wrong syscall return
311 459031 Documentation on --error-exitcode incomplete
312 459477 XERROR messages lacks ending '\n' in vgdb
313 462007 Implicit int in none/tests/faultstatus.c
315 To see details of a given bug, visit
316 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
317 where XXXXXX is the bug number as listed above.
319 (3.20.0.RC1: 20 Oct 2022)
322 Release 3.19.0 (11 Apr 2022)
323 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
325 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
326 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
327 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
328 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
329 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
330 AMD64/macOS 10.13 and nanoMIPS/Linux.
332 * ==================== CORE CHANGES ===================
334 * Fix Rust v0 name demangling.
335 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
336 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
337 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with HardenedBSD
339 * ================== PLATFORM CHANGES =================
342 - ignore the "v8.x" architecture levels, only look at actual CPU features
343 present. Fixes mismatch detected between RDMA and atomics features
344 preventing startup on some QEMU configurations.
345 - Implement LD{,A}XP and ST{,L}XP
346 - Fix incorrect code emitted for doubleword CAS.
349 - Fix sys_ipc semtimedop syscall
350 - Fix VFLRX and WFLRX instructions
351 - Fix EXRL instruction with negative offset
354 - Reimplement the vbpermq instruction support to generate less Iops and
355 avoid overflowing internal buffers.
356 - Fix checking for scv support to avoid "Facility 'SCV' unavailable (12),
357 exception" messages in dmsg.
358 - Fix setting condition code for Vector Compare quad word instructions.
359 - Fix fix lxsibzx, lxsihzx and lxsihzx instructions so they only load
360 their respective sized data.
361 - Fix the prefixed stq instruction in PC relative mode.
363 * ==================== TOOL CHANGES ===================
366 - Speed up --track-origins=yes for large (in the range of hundreds to
367 thousands of megabytes) mmap/munmaps.
369 - Several fixes for new versions of libstd++ using new posix try_lock
372 * ==================== FIXED BUGS ====================
374 The following bugs have been fixed or resolved. Note that "n-i-bz"
375 stands for "not in bugzilla" -- that is, a bug that was reported to us
376 but never got a bugzilla entry. We encourage you to file bugs in
377 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
378 than mailing the developers (or mailing lists) directly -- bugs that
379 are not entered into bugzilla tend to get forgotten about or ignored.
381 403802 leak_cpp_interior fails with some reachable blocks different than expected
382 435732 memcheck/tests/leak_cpp_interior fails with gcc11
383 444242 s390x: Valgrind crashes on EXRL with negative offset
384 444399 arm64: unhandled instruction 0xC87F2D89 (LD{,A}XP and ST{,L}XP).
386 444481 gdb_server test failures on s390x
387 444495 dhat/tests/copy fails on s390x
388 444552 memcheck/tests/sem fails on s390x with glibc 2.34
389 444571 PPC, fix the lxsibzx and lxsihzx so they only load their respective
391 444836 PPC, pstq instruction for R=1 is not storing to the correct address.
392 444925 fexecve syscall wrapper not properly implemented
393 445032 valgrind/memcheck crash with SIGSEGV when SIGVTALRM timer used and
395 445211 Fix out of tree builds
396 445300 [PATCH] Fix building tests with Musl
397 445011 SIGCHLD is sent when valgrind uses debuginfod-find
398 445354 arm64 backend: incorrect code emitted for doubleword CAS
399 445415 arm64 front end: alignment checks missing for atomic instructions
400 445504 Using C++ condition_variable results in bogus "mutex is locked simultaneously by two threads" warning
401 445607 Unhandled amd64-freebsd syscall: 247
402 445668 Inline stack frame generation is broken for Rust binaries
403 445916 Demangle Rust v0 symbols with .llvm suffix
404 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false positives
405 446138 DRD/Helgrind with std::timed_mutex::try_lock_until false positives
406 446281 Add a DRD suppression for fwrite
407 446103 Memcheck: `--track-origins=yes` causes extreme slowdowns for large mmap/munmap
408 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false
409 446251 TARGET_SIGNAL_THR added to enum target_signal
410 446823 FreeBSD - missing syscalls when using libzm4
411 447991 s390x: Valgrind indicates illegal instruction on wflrx
412 447995 Valgrind segfault on power10 due to hwcap checking code
413 449483 Powerpc: vcmpgtsq., vcmpgtuq,, vcmpequq. instructions not setting the
414 condition code correctly.
415 449672 ppc64 --track-origins=yes failures because of bad cmov addHRegUse
416 449838 sigsegv liburing the 'impossible' happened for io_uring_setup
417 450025 Powerc: ACC file not implemented as a logical overlay of the VSR
419 450437 Warn for execve syscall with argv or argv[0] being NULL
420 450536 Powerpc: valgrind throws 'facility scv unavailable exception'
421 451626 Syscall param bpf(attr->raw_tracepoint.name) points to unaddressable byte(s)
422 451827 [ppc64le] VEX temporary storage exhausted with several vbpermq instructions
423 451843 valgrind fails to start on a FreeBSD system which enforces W^X
425 To see details of a given bug, visit
426 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
427 where XXXXXX is the bug number as listed above.
429 (3.19.0.RC1: 02 Apr 2022)
430 (3.19.0.RC2: 08 Apr 2022)
433 Release 3.18.0 (15 Oct 2021)
434 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
436 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
437 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
438 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
439 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
440 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
441 AMD64/macOS 10.13 and nanoMIPS/Linux.
443 * ==================== CORE CHANGES ===================
445 * The libiberty demangler has been updated, which brings support for
446 Rust v0 name demangling. [Update: alas, due to a bug, this support
447 isn't working in 3.18.0.]
449 * __libc_freeres isn't called anymore after the program recieves a
450 fatal signal. Causing some internal glibc resources to hang around,
451 but preventing any crashes after the program has ended.
453 * The DWARF reader is now very much faster at startup when just
454 --read-inline-info=yes (the default in most cases) is given.
456 * glibc 2.34, which moved various functions from libpthread.so into
457 libc.so, is now supported.
459 * ================== PLATFORM CHANGES =================
463 - v8.2 scalar and vector FABD, FACGE, FACGT and FADD.
464 - v8.2 FP compare & conditional compare instructions.
465 - Zero variants of v8.2 FP compare instructions.
469 - Support the miscellaneous-instruction-extensions facility 3 and
470 the vector-enhancements facility 2. This enables programs
471 compiled with "-march=arch13" or "-march=z15" to be executed
476 - ISA 3.1 support is now complete
477 - ISA 3.0 support for the darn instruction added.
478 - ISA 3.0 support for the vector system call instruction scv added.
479 - ISA 3.0 support for the copy, paste and cpabort instructions added.
481 * Support for X86/FreeBSD and AMD64/FreeBSD has been added.
483 * ==================== OTHER CHANGES ====================
485 * Memcheck on amd64: minor fixes to remove some false positive
488 * ==================== FIXED BUGS ====================
490 The following bugs have been fixed or resolved. Note that "n-i-bz"
491 stands for "not in bugzilla" -- that is, a bug that was reported to us
492 but never got a bugzilla entry. We encourage you to file bugs in
493 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
494 than mailing the developers (or mailing lists) directly -- bugs that
495 are not entered into bugzilla tend to get forgotten about or ignored.
497 208531 [PATCH]: FreeBSD support for valgrind
498 368960 WARNING: unhandled amd64-linux syscall: 163 (acct)
499 407589 [Linux] Add support for C11 aligned_alloc() and GNU reallocarray()
500 423963 Error in child thread when CLONE_PIDFD is used
501 426148 crash with "impossible happened" when running BPF CO-RE programs
502 429375 PPC ISA 3.1 support is missing, part 9
503 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
504 431306 Update demangler to support Rust v0 name mangling
505 432387 s390x: z15 instructions support
506 433437 FreeBSD support, part 1
507 433438 FreeBSD support, part 2
508 433439 FreeBSD support, part 3
509 433469 FreeBSD support, part 4
510 433473 FreeBSD support, part 5
511 433477 FreeBSD support, part 6
512 433479 FreeBSD support, part 7
513 433504 FreeBSD support, part 8
514 433506 FreeBSD support, part 9
515 433507 FreeBSD support, part 10
516 433508 FreeBSD support, part 11
517 433510 FreeBSD support, part 12
518 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
519 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
520 434296 s390x: False-positive memcheck diagnostics from vector string
522 434840 PPC64 darn instruction not supported
523 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
524 435908 valgrind tries to fetch from deubginfod for files which already
525 have debug information
526 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2
527 439046 valgrind is unusably large when linked with lld
528 439090 Implement close_range(2)
529 439326 Valgrind 3.17.0 won't compile with Intel 2021 oneAPI compilers
530 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
531 440670 unhandled ppc64le-linux syscall: 252 statfs64 and 253 fstatfs64
532 440906 Fix impossible constraint issue in P10 testcase.
533 441512 Remove a unneeded / unnecessary prefix check.
534 441534 Update the expected output for test_isa_3_1_VRT.
535 442061 very slow execution under Fedora 34 (readdwarf3)
536 443031 Gcc -many change requires explicit .machine directives
537 443033 Add support for the ISA 3.0 mcrxrx instruction
538 443034 Sraw, srawi, srad, sradi, mfs
539 443178 Powerpc, test jm-mfspr expected output needs to be updated.
540 443179 Need new test for the lxvx and stxvx instructions on ISA 2.07 and
542 443180 The subnormal test and the ISA 3.0 test generate compiler warnings
543 443314 In the latest GIT version, Valgrind with "--trace-flags" crashes
545 443605 Don't call final_tidyup (__libc_freeres) on FatalSignal
547 To see details of a given bug, visit
548 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
549 where XXXXXX is the bug number as listed below.
551 (3.18.0.RC1: 12 Oct 2021)
552 (3.18.0: 15 Oct 2021)
556 Release 3.17.0 (19 Mar 2021)
557 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
559 3.17.0 fixes a number of bugs and adds some functional changes: support for
560 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
561 some new instructions for Arm64, S390 and POWER. There are also some tool
564 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
565 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
566 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
567 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
568 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
570 * ==================== CORE CHANGES ===================
572 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
575 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
576 debugging information. When a debuginfo file cannot be found locally,
577 Valgrind is able to query debuginfod servers for the file using its
578 build-id. See the user manual for more information about debuginfod support.
580 * ================== PLATFORM CHANGES =================
584 - Inaccuracies resulting from double-rounding in the simulation of
585 floating-point multiply-add/subtract instructions have been fixed. These
586 should now behave exactly as the hardware does.
588 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
589 ongoing. Support for the half-word variants of at least the following
590 instructions has been added:
592 FABS <Vd>.<T>, <Vn>.<T>
594 FNEG <Vd>.<T>, <Vn>.<T>
596 FSQRT <Vd>.<T>, <Vn>.<T>
601 - Implement the new instructions/features that were added to z/Architecture
602 with the vector-enhancements facility 1. Also cover the instructions from
603 the vector-packed-decimal facility that are defined outside the chapter
604 "Vector Decimal Instructions", but not the ones from that chapter itself.
606 For a detailed list of newly supported instructions see the updates to
607 `docs/internals/s390-opcodes.csv'.
609 Since the miscellaneous instruction extensions facility 2 was already
610 added in Valgrind 3.16.0, this completes the support necessary to run
611 general programs built with `--march=z14' under Valgrind. The
612 vector-packed-decimal facility is currently not exploited by the standard
613 toolchain and libraries.
617 - Various bug fixes. Fix for the sync field to limit setting just two of
618 the two bits in the L-field. Fix the write size for the stxsibx and
619 stxsihx instructions. Fix the modsw and modsd instructions.
621 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
622 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
623 conversion instructions are still missing.
625 * ==================== TOOL CHANGES ====================
627 * General tool changes
629 - All the tools and their vgpreload libraries are now installed under
630 libexec because they cannot be executed directly and should be run through
631 the valgrind executable. This should be an internal, not user visible,
632 change, but might impact valgrind packagers.
634 - The --track-fds option now respects -q, --quiet and won't output anything
635 if no file descriptors are leaked. It also won't report the standard stdin
636 (0), stdout (1) or stderr (2) descriptors as being leaked with
637 --trace-fds=yes anymore. To track whether the standard file descriptors
638 are still open at the end of the program run use --trace-fds=all.
642 - DHAT has been extended, with two new modes of operation. The new
643 --mode=copy flag triggers copy profiling, which records calls to memcpy,
644 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
645 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
646 the new dhat/dhat.h file. This is useful for learning more about hot code
647 paths. See the user manual for more information about the new modes.
649 - Because of these changes, DHAT's file format has changed. DHAT output
650 files produced with earlier versions of DHAT will not work with this
651 version of DHAT's viewer, and DHAT output files produced with this version
652 of DHAT will not work with earlier versions of DHAT's viewer.
654 * ==================== FIXED BUGS ====================
656 The following bugs have been fixed or resolved. Note that "n-i-bz"
657 stands for "not in bugzilla" -- that is, a bug that was reported to us
658 but never got a bugzilla entry. We encourage you to file bugs in
659 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
660 than mailing the developers (or mailing lists) directly -- bugs that
661 are not entered into bugzilla tend to get forgotten about or ignored.
663 To see details of a given bug, visit
664 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
665 where XXXXXX is the bug number as listed below.
667 140178 open("/proc/self/exe", ...); doesn't quite work
668 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
669 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
670 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
671 345077 linux syscall execveat support (linux 3.19)
672 361770 Missing F_ADD_SEALS
673 369029 handle linux syscalls sched_getattr and sched_setattr
674 384729 __libc_freeres inhibits cross-platform valgrind
675 388787 Support for C++17 new/delete
676 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
677 396656 Warnings while reading debug info
678 397605 ioctl FICLONE mishandled
679 401416 Compile failure with openmpi 4.0
680 408663 Suppression file for musl libc
681 404076 s390x: z14 vector instructions not implemented
682 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
683 (actually affected all x86 and nanomips regardless of host bitness)
684 413547 regression test does not check for Arm 64 features.
685 414268 Enable AArch64 feature detection and decoding for v8.x instructions
686 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
687 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
688 422261 platform selection fails for unqualified client name
689 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
690 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
691 423195 PPC ISA 3.1 support is missing, part 1
692 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
693 424012 crash with readv/writev having invalid but not NULL arg2 iovec
694 424298 amd64: Implement RDSEED
695 425232 PPC ISA 3.1 support is missing, part 2
696 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
697 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
698 426123 PPC ISA 3.1 support is missing, part 3
699 426144 Fix "condition variable has not been initialized" on Fedora 33.
700 427400 PPC ISA 3.1 support is missing, part 4
701 427401 PPC ISA 3.1 support is missing, part 5
702 427404 PPC ISA 3.1 support is missing, part 6
703 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
704 427787 Support new faccessat2 linux syscall (439)
705 427969 debuginfo section duplicates a section in the main ELF file
706 428035 drd: Unbreak the musl build
707 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
708 428716 cppcheck detects potential leak in VEX/useful/smchash.c
709 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
710 429352 PPC ISA 3.1 support is missing, part 7
711 429354 PPC ISA 3.1 support is missing, part 8
712 429692 unhandled ppc64le-linux syscall: 147 (getsid)
713 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
715 429952 Errors when building regtest with clang
716 430354 ppc stxsibx and stxsihx instructions write too much data
717 430429 valgrind.h doesn't compile on s390x with clang
718 430485 expr_is_guardable doesn't handle Iex_Qop
719 431556 Complete arm64 FADDP v8.2 instruction support
720 432102 Add support for DWARF5 as produced by GCC11
721 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
722 432381 drd: Process STACK_REGISTER client requests
723 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
724 432672 vg_regtest: test-specific environment variables not reset between tests
725 432809 VEX should support REX.W + POPF
726 432861 PPC modsw and modsd give incorrect results for 1 mod 12
727 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
728 432215 Add debuginfod functionality
729 433323 Use pkglibexecdir as vglibdir
730 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
731 433629 valgrind/README has type "abd" instead of "and"
732 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
733 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
734 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
735 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
736 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
738 (3.17.0.RC1: 13 Mar 2021)
739 (3.17.0.RC2: 17 Mar 2021)
740 (3.17.0: 19 Mar 2021)
744 Release 3.16.1 (22 June 2020)
745 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
747 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
748 fixes character encoding problems in the documentation HTML.
750 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
751 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
753 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
757 Release 3.16.0 (27 May 2020)
758 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
760 3.16.0 is a feature release with many improvements and the usual collection of
763 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
764 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
765 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
766 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
767 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
769 * ==================== CORE CHANGES ===================
771 * It is now possible to dynamically change the value of many command line
772 options while your program (or its children) are running under Valgrind.
774 To see the list of dynamically changeable options, run
775 "valgrind --help-dyn-options".
777 You can change the options from the shell by using vgdb to launch
778 the monitor command "v.clo <clo option>...".
779 The same monitor command can be used from a gdb connected
780 to the valgrind gdbserver.
781 Your program can also change the dynamically changeable options using
782 the client request VALGRIND_CLO_CHANGE(option).
784 * ================== PLATFORM CHANGES =================
786 * MIPS: preliminary support for nanoMIPS instruction set has been added.
788 * ==================== TOOL CHANGES ====================
792 - The implicit memcpy done by each call to realloc now counts towards the
793 read and write counts of resized heap blocks, making those counts higher
798 - cg_annotate's --auto and --show-percs options now default to 'yes', because
799 they are usually wanted.
803 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
804 because they are usually wanted.
806 - The command option --collect-systime has been enhanced to specify
807 the unit used to record the elapsed time spent during system calls.
808 The command option now accepts the values no|yes|msec|usec|nsec,
809 where yes is a synonym of msec. When giving the value nsec, the
810 system cpu time of system calls is also recorded.
814 - Several memcheck options are now dynamically changeable.
815 Use valgrind --help-dyn-options to list them.
817 - The release 3.15 introduced a backward incompatible change for
818 some suppression entries related to preadv and pwritev syscalls.
819 When reading a suppression entry using the unsupported 3.14 format,
820 valgrind will now produce a warning to say the suppression entry will not
821 work, and suggest the needed change.
823 - Significantly fewer false positive errors on optimised code generated by
824 Clang and GCC. In particular, Memcheck now deals better with the
825 situation where the compiler will transform C-level "A && B" into "B && A"
826 under certain circumstances (in which the transformation is valid).
827 Handling of integer equality/non-equality checks on partially defined
828 values is also improved on some architectures.
832 - The exprimental Stack and Global Array Checking tool has been removed.
833 It only ever worked on x86 and amd64, and even on those it had a
834 high false positive rate and was slow. An alternative for detecting
835 stack and global array overruns is using the AddressSanitizer (ASAN)
836 facility of the GCC and Clang compilers, which require you to rebuild
837 your code with -fsanitize=address.
839 * ==================== OTHER CHANGES ====================
841 * New and modified GDB server monitor features:
843 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
845 - The gdbserver monitor commands that require an address and an optional
846 length argument now accepts the alternate 'C like' syntax "address[length]".
847 For example, the memcheck command "monitor who_points_at 0x12345678 120"
848 can now also be given as "monitor who_points_at 0x12345678[120]".
850 * ==================== FIXED BUGS ====================
852 The following bugs have been fixed or resolved. Note that "n-i-bz"
853 stands for "not in bugzilla" -- that is, a bug that was reported to us
854 but never got a bugzilla entry. We encourage you to file bugs in
855 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
856 than mailing the developers (or mailing lists) directly -- bugs that
857 are not entered into bugzilla tend to get forgotten about or ignored.
859 To see details of a given bug, visit
860 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
861 where XXXXXX is the bug number as listed below.
863 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
864 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
865 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
866 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
867 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
868 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
869 400872 Add nanoMIPS support to Valgrind
870 403212 drd/tests/trylock hangs on FreeBSD
871 404406 s390x: z14 miscellaneous instructions not implemented
872 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
873 406561 mcinfcallWSRU gdbserver_test fails on ppc64
874 406824 Unsupported baseline
875 407218 Add support for the copy_file_range syscall
876 407307 Intercept stpcpy also in ld.so for arm64
877 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
879 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
880 408009 Expose rdrand and f16c even on avx if host cpu supports them
881 408091 Missing pkey syscalls
882 408414 Add support for missing for preadv2 and pwritev2 syscalls
883 409141 Valgrind hangs when SIGKILLed
884 409206 Support for Linux PPS and PTP ioctls
885 409367 exit_group() after signal to thread waiting in futex() causes hangs
886 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
887 409780 References to non-existent configure.in
888 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
889 410599 Non-deterministic behaviour of pth_self_kill_15_other test
890 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
891 411134 Allow the user to change a set of command line options during execution
892 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
893 412344 Problem setting mips flags with specific paths
894 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
895 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
896 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
897 413603 callgrind_annotate/cg_annotate truncate function names at '#'
898 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
899 415136 ARMv8.1 Compare-and-Swap instructions are not supported
900 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
901 416239 valgrind crashes when handling clock_adjtime
902 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
903 416286 DRD reports "conflicting load" error on std::mutex::lock()
904 416301 s390x: "compare and signal" not supported
905 416387 finit_module and bpf syscalls are unhandled on arm64
906 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
907 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
908 416753 new 32bit time syscalls for 2038+
909 417075 pwritev(vector[...]) suppression ignored
910 417075 is not fixed, but incompatible supp entries are detected
911 and a warning is produced for these.
912 417187 [MIPS] Conditional branch problem since 'grail' changes
913 417238 Test memcheck/tests/vbit-test fails on mips64 BE
914 417266 Make memcheck/tests/linux/sigqueue usable with musl
915 417281 s390x: /bin/true segfaults with "grail" enabled
916 417427 commit to fix vki_siginfo_t definition created numerous regression
918 417452 s390_insn_store_emit: dst->tag for HRcVec128
919 417578 Add suppressions for glibc DTV leaks
920 417906 clone with CLONE_VFORK and no CLONE_VM fails
921 418004 Grail code additions break ppc64.
922 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
923 418997 s390x: Support Iex_ITE for float and vector types
924 419503 s390x: Avoid modifying registers returned from isel functions
925 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
926 421570 std_mutex fails on Arm v8.1 h/w
927 434035 vgdb might crash if valgrind is killed
928 n-i-bz Fix minor one time leaks in dhat.
929 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
930 n-i-bz Add support for the Linux io_uring system calls
931 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
932 n-i-bz Fix non-glibc build of test suite with s390x_features
933 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
934 423195 PPC ISA 3.1 support is missing, part 1
936 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
937 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
938 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
942 Release 3.15.0 (12 April 2019)
943 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
945 3.15.0 is a feature release with many improvements and the usual collection of
948 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
949 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
950 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
951 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
952 support for X86/macOS 10.13 and AMD64/macOS 10.13.
954 * ==================== CORE CHANGES ===================
956 * The XTree Massif output format now makes use of the information obtained
957 when specifying --read-inline-info=yes.
959 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
961 * ==================== TOOL CHANGES ====================
965 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
966 it has been promoted from an experimental tool to a regular tool. Run it
967 with --tool=dhat instead of --tool=exp-dhat.
969 - DHAT now prints only minimal data when the program ends, instead writing
970 the bulk of the profiling data to a file. As a result, the --show-top-n
971 and --sort-by options have been removed.
973 - Profile results can be viewed with the new viewer, dh_view.html. When
974 a run ends, a short message is printed, explaining how to view the result.
976 - See the documentation for more details.
980 - cg_annotate has a new option, --show-percs, which prints percentages next
985 - callgrind_annotate has a new option, --show-percs, which prints percentages
986 next to all event counts.
988 - callgrind_annotate now inserts commas in call counts, and
989 sort the caller/callee lists in the call tree.
993 - The default value for --read-inline-info is now "yes" on
994 Linux/Android/Solaris. It is still "no" on other OS.
998 - The option --xtree-leak=yes (to output leak result in xtree format)
999 automatically activates the option --show-leak-kinds=all, as xtree
1000 visualisation tools such as kcachegrind can in any case select what kind
1001 of leak to visualise.
1003 - There has been further work to avoid false positives. In particular,
1004 integer equality on partially defined inputs (C == and !=) is now handled
1007 * ==================== OTHER CHANGES ====================
1009 * The new option --show-error-list=no|yes displays, at the end of the run, the
1010 list of detected errors and the used suppressions. Prior to this change,
1011 showing this information could only be done by specifying "-v -v", but that
1012 also produced a lot of other possibly-non-useful messages. The option -s is
1013 equivalent to --show-error-list=yes.
1015 * ==================== FIXED BUGS ====================
1017 The following bugs have been fixed or resolved. Note that "n-i-bz"
1018 stands for "not in bugzilla" -- that is, a bug that was reported to us
1019 but never got a bugzilla entry. We encourage you to file bugs in
1020 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1021 than mailing the developers (or mailing lists) directly -- bugs that
1022 are not entered into bugzilla tend to get forgotten about or ignored.
1024 To see details of a given bug, visit
1025 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1026 where XXXXXX is the bug number as listed below.
1028 385411 s390x: z13 vector floating-point instructions not implemented
1029 397187 z13 vector register support for vgdb gdbserver
1030 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
1031 398870 Please add support for instruction vcvtps2ph
1032 399287 amd64 front end: Illegal Instruction vcmptrueps
1033 399301 Use inlined frames in Massif XTree output.
1034 399322 Improve callgrind_annotate output
1035 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
1036 400164 helgrind test encounters mips x-compiler warnings and assembler error
1037 400490 s390x: VRs allocated as if separate from FPRs
1038 400491 s390x: Operand of LOCH treated as unsigned integer
1039 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
1040 options, which specify a mips64 processor
1041 401112 LLVM 5.0 generates comparison against partially initialized data
1042 401277 More bugs in z13 support
1043 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
1044 401578 drd: crashes sometimes on fork()
1045 401627 memcheck errors with glibc avx2 optimized wcsncmp
1046 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
1047 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
1048 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
1050 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
1051 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
1052 402123 invalid assembler opcodes for mips32r2
1053 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
1054 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
1055 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
1056 402351 mips64 libvexmultiarch_test fails on s390x
1057 402369 Overhaul DHAT
1058 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
1059 402480 Do not use %rsp in clobber list
1060 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
1061 402515 Implement new option --show-error-list=no|yes / -s
1062 402519 POWER 3.0 addex instruction incorrectly implemented
1063 402781 Redo the cache used to process indirect branch targets
1064 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
1065 403552 s390x: wrong facility bit checked for vector facility
1066 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
1067 404638 Add VG_(replaceIndexXA)
1068 404843 s390x: backtrace sometimes ends prematurely
1069 404888 autotools cleanup series
1070 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
1071 405182 Valgrind fails to build with Clang
1072 405205 filter_libc: remove the line holding the futex syscall error entirely
1073 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
1074 the upper and lower 32-bits of the 64-bit result
1075 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
1076 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
1077 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
1079 405403 s390x disassembler cannot be used on x86
1080 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
1081 405458 MIPS mkFormVEC arguments swapped?
1082 405716 drd: Fix an integer overflow in the stack margin calculation
1083 405722 Support arm64 core dump
1084 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
1085 of the 64-bit destination field.
1086 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
1087 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
1088 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
1090 406256 PPC64, vector floating point instructions don't handle subnormal
1091 according to VSCR[NJ] bit setting.
1092 406352 cachegrind/callgrind fails ann tests because of missing a.c
1093 406354 dhat is broken on x86 (32bit)
1094 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
1095 406357 gdbserver_tests fails because of gdb output change
1096 406360 memcheck/tests/libstdc++.supp needs more supression variants
1097 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
1098 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
1099 407340 PPC64, does not support the vlogefp, vexptefp instructions.
1100 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
1101 n-i-bz Fix callgrind_annotate non deterministic order for equal total
1102 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
1103 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
1104 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
1106 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
1107 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
1108 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
1112 Release 3.14.0 (9 October 2018)
1113 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1115 3.14.0 is a feature release with many improvements and the usual collection of
1118 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1119 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1120 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1121 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
1122 support for X86/macOS 10.13, AMD64/macOS 10.13.
1124 * ==================== CORE CHANGES ===================
1126 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
1127 debug info for unloaded code. This allows saved stack traces (e.g. for
1128 memory leaks) to include file/line info for code that has been dlclose'd (or
1129 similar). See the user manual for more information and known limitations.
1131 * Ability to specify suppressions based on source file name and line number.
1133 * Majorly overhauled register allocator. No end-user changes, but the JIT
1134 generates code a bit more quickly now.
1136 * ================== PLATFORM CHANGES =================
1138 * Preliminary support for macOS 10.13 has been added.
1140 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
1142 * mips: support for MIPS SIMD architecture (MSA) has been added.
1144 * mips: support for MIPS N32 ABI has been added.
1146 * s390: partial support for vector instructions (integer and string) has been
1149 * ==================== TOOL CHANGES ====================
1151 * Helgrind: Addition of a flag
1152 --delta-stacktrace=no|yes [yes on linux amd64/x86]
1153 which specifies how full history stack traces should be computed.
1154 Setting this to =yes can speed up Helgrind by 25% when using
1155 --history-level=full.
1157 * Memcheck: reduced false positive rate for optimised code created by Clang 6
1158 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
1159 blocks more carefully to determine where it can avoid expensive definedness
1160 checks without loss of precision. This is controlled by the flag
1161 --expensive-definedness-checks=no|auto|yes [auto].
1163 * ==================== OTHER CHANGES ====================
1165 * Valgrind is now buildable with link-time optimisation (LTO). A new
1166 configure option --enable-lto=yes allows building Valgrind with LTO. If the
1167 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
1168 Note that if you are doing Valgrind development, --enable-lto=yes massively
1169 slows down the build process.
1171 * ==================== FIXED BUGS ====================
1173 The following bugs have been fixed or resolved. Note that "n-i-bz"
1174 stands for "not in bugzilla" -- that is, a bug that was reported to us
1175 but never got a bugzilla entry. We encourage you to file bugs in
1176 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1177 than mailing the developers (or mailing lists) directly -- bugs that
1178 are not entered into bugzilla tend to get forgotten about or ignored.
1180 To see details of a given bug, visit
1181 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1182 where XXXXXX is the bug number as listed below.
1184 79362 Debug info is lost for .so files when they are dlclose'd
1185 208052 strlcpy error when n = 0
1186 255603 exp-sgcheck Assertion '!already_present' failed
1187 338252 building valgrind with -flto (link time optimisation) fails
1188 345763 MIPS N32 ABI support
1189 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
1190 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
1191 372347 Replacement problem of the additional c++14/c++17 new/delete operators
1192 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1193 376257 helgrind history full speed up using a cached stack
1194 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
1196 379748 Fix missing pselect syscall (OS X 10.11)
1197 379754 Fix missing syscall ulock_wait (OS X 10.12)
1198 380397 s390x: __GI_strcspn() replacemenet needed
1199 381162 possible array overrun in VEX register allocator
1200 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
1201 381274 powerpc too chatty even with --sigill-diagnostics=no
1202 381289 epoll_pwait can have a NULL sigmask
1203 381553 VEX register allocator v3
1204 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
1205 381769 Use ucontext_t instead of struct ucontext
1206 381805 arm32 needs ld.so index hardwire for new glibc security fixes
1207 382256 gz compiler flag test doesn't work for gold
1208 382407 vg_perf needs "--terse" command line option
1209 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
1210 382563 MIPS MSA ASE support
1211 382998 xml-socket doesn't work
1212 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
1213 383723 Fix missing kevent_qos syscall (macOS 10.11)
1214 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
1215 384096 Mention AddrCheck at Memcheck's command line option [..]
1216 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
1217 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
1218 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
1219 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
1220 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
1221 384337 performance improvements to VEX register allocator v2 and v3
1222 384526 reduce number of spill insns generated by VEX register allocator v3
1223 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
1224 384631 Sanitise client args as printed with -v
1225 384633 Add a simple progress-reporting facility
1226 384987 VEX regalloc: allocate caller-save registers for short lived vregs
1227 385055 PPC VEX temporary storage exhausted
1228 385182 PPC64 is missing support for the DSCR
1229 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
1230 385207 PPC64, generate_store_FPRF() generates too many Iops
1231 385208 PPC64, xxperm instruction exhausts temporary memory
1232 385210 PPC64, vpermr instruction could exhaust temporary memory
1233 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
1234 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
1235 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
1236 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
1237 385408 s390x: z13 vector "support" instructions not implemented
1238 385409 s390x: z13 vector integer instructions not implemented
1239 385410 s390x: z13 vector string instructions not implemented
1240 385412 s390x: new non-vector z13 instructions not implemented
1241 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
1242 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
1243 385939 Optionally exit on the first error
1244 386318 valgrind.org/info/tools.html is missing SGCheck
1245 386425 running valgrind + wine on armv7l gives illegal opcode
1246 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
1247 387410 MIPSr6 support
1248 387664 Memcheck: make expensive-definedness-checks be the default
1249 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
1250 387766 asm shifts cause false positive "Conditional jump or move depends
1251 on uninitialised value"
1252 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
1253 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
1254 388786 Support bpf syscall in amd64 Linux
1255 388862 Add replacements for wmemchr and wcsnlen on Linux
1256 389065 valgrind meets gcc flag -Wlogical-op
1257 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
1258 390471 suppression by specification of source-file line number
1259 390723 make xtree dump files world wide readable, similar to log files
1260 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
1261 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
1262 392118 unhandled amd64-linux syscall: 332 (statx)
1263 392449 callgrind not clearing the number of calls properly
1264 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
1265 lxssp, stxssp and stxvl instructions.
1266 393023 callgrind_control risks using the wrong vgdb
1267 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
1268 393099 posix_memalign() invalid write if alignment == 0
1269 393146 failing assert "is_DebugInfo_active(di)"
1270 395709 PPC64 is missing support for the xvnegsp instruction
1271 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
1273 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
1274 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
1275 396839 s390x: Trap instructions not implemented
1276 396887 arch_prctl should return EINVAL on unknown option
1277 == 397286 crash before launching binary (Unsupported arch_prctl option)
1278 == 397393 valgrind: the 'impossible' happened: (Archlinux)
1279 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
1280 396906 compile tests failure on mips32-linux: broken inline asm in tests on
1282 397012 glibc ld.so uses arch_prctl on i386
1283 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
1284 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
1285 397424 glibc 2.27 and gdb_server tests
1286 398028 Assertion `cfsi_fits` failing in simple C program
1287 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
1289 n-i-bz Fix missing workq_ops operations (macOS)
1290 n-i-bz fix bug in strspn replacement
1291 n-i-bz Add support for the Linux BLKFLSBUF ioctl
1292 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
1293 n-i-bz Fix possible stack trashing by semctl syscall wrapping
1294 n-i-bz Add support for the Linux membarrier() system call
1295 n-i-bz x86 front end: recognise and handle UD2 correctly
1296 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
1297 correctly aligned before entering the handler.
1299 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
1300 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
1301 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
1305 Release 3.13.0 (15 June 2017)
1306 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1308 3.13.0 is a feature release with many improvements and the usual collection of
1311 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1312 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1313 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1314 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
1316 * ==================== CORE CHANGES ===================
1318 * The translation cache size has been increased to keep up with the demands of
1319 large applications. The maximum number of sectors has increased from 24 to
1320 48. The default number of sectors has increased from 16 to 32 on all
1321 targets except Android, where the increase is from 6 to 12.
1323 * The amount of memory that Valgrind can use has been increased from 64GB to
1324 128GB. In particular this means your application can allocate up to about
1325 60GB when running on Memcheck.
1327 * Valgrind's default load address has been changed from 0x3800'0000 to
1328 0x5800'0000, so as to make it possible to load larger executables. This
1329 should make it possible to load executables of size at least 1200MB.
1331 * A massive spaceleak caused by reading compressed debuginfo files has been
1332 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
1335 * The C++ demangler has been updated.
1337 * Support for demangling Rust symbols has been added.
1339 * A new representation of stack traces, the "XTree", has been added. An XTree
1340 is a tree of stacktraces with data associated with the stacktraces. This is
1341 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
1342 consumption of your program. Reporting is controlled by the new options
1343 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
1345 A report can also be produced on demand using the gdbserver monitor command
1346 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
1347 format' and 'massif format. The existing visualisers for these formats (e.g.
1348 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
1349 analyse these reports.
1351 Memcheck can also produce XTree leak reports using the Callgrind file
1352 format. For more details, see the user manual.
1354 * ================== PLATFORM CHANGES =================
1356 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
1358 * amd64: fixes for JIT failure problems on long AVX2 code blocks
1360 * amd64 and x86: support for CET prefixes has been added
1362 * arm32: a few missing ARMv8 instructions have been implemented
1364 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
1365 Store-Conditional instructions has been added. This is to deal with
1366 processor implementations that implement the LL/SC specifications strictly
1367 and as a result cause Valgrind to hang in certain situations. The
1368 alternative implementation is automatically enabled at startup, as required.
1369 You can use the option --sim-hints=fallback-llsc to force-enable it if you
1372 * Support for OSX 10.12 has been improved.
1374 * On Linux, clone handling has been improved to honour CLONE_VFORK that
1375 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
1376 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
1377 CLONE_VM exact semantics will (still) not work.
1379 * The TileGX/Linux port has been removed because it appears to be both unused
1382 * ==================== TOOL CHANGES ====================
1386 - Memcheck should give fewer false positives when running optimised
1387 Clang/LLVM generated code.
1389 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1391 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
1392 to produce the end of execution leak report in a xtree callgrind format
1395 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
1396 the leak report in an xtree file.
1400 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1402 - For some workloads (typically, for big applications), Massif memory
1403 consumption and CPU consumption has decreased significantly.
1407 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1409 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
1410 for Ada gnat compiled applications.
1412 * ==================== OTHER CHANGES ====================
1414 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
1415 append the inner guest stacktrace to the inner host stacktrace. This helps
1416 to investigate the errors reported by the outer, when they are caused by the
1417 inner guest program (such as an inner regtest). See README_DEVELOPERS for
1420 * To allow fast detection of callgrind files by desktop environments and file
1421 managers, the format was extended to have an optional first line that
1422 uniquely identifies the format ("# callgrind format"). Callgrind creates
1423 this line now, as does the new xtree functionality.
1425 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
1426 have a new %n format letter that is replaced by a sequence number.
1428 * "--version -v" now shows the SVN revision numbers from which Valgrind was
1431 * ==================== FIXED BUGS ====================
1433 The following bugs have been fixed or resolved. Note that "n-i-bz"
1434 stands for "not in bugzilla" -- that is, a bug that was reported to us
1435 but never got a bugzilla entry. We encourage you to file bugs in
1436 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1437 than mailing the developers (or mailing lists) directly -- bugs that
1438 are not entered into bugzilla tend to get forgotten about or ignored.
1440 To see details of a given bug, visit
1441 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1442 where XXXXXX is the bug number as listed below.
1444 162848 --log-file output isn't split when a program forks
1445 340777 Illegal instruction on mips (ar71xx)
1446 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
1447 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
1448 to a different stack.
1449 344139 x86 stack-seg overrides, needed by the Wine people
1450 344524 store conditional of guest applications always fail - observed on
1452 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
1453 352395 Please provide SVN revision info in --version -v
1454 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
1455 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
1456 INVALID_POSIX_THREADID' failed
1457 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
1458 barrier implementation
1459 358697 valgrind.h: Some code remains even when defining NVALGRIND
1460 359202 Add musl libc configure/compile
1461 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
1462 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
1463 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
1464 362223 assertion failed when .valgrindrc is a directory instead of a file
1465 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
1466 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
1467 368507 can't malloc chunks larger than about 34GB
1468 368529 Android arm target link error, missing atexit and pthread_atfork
1469 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
1470 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
1471 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
1472 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
1473 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
1474 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
1475 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
1476 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
1477 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
1478 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
1479 370028 Reduce the number of compiler warnings on MIPS platforms
1480 370635 arm64 missing syscall getcpu
1481 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
1482 371227 Clean AArch64 syscall table
1483 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
1484 371471 Valgrind complains about non legit memory leaks on placement new (C++)
1485 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
1486 371503 disInstr(arm64): unhandled instruction 0xF89F0000
1487 371869 support '%' in symbol Z-encoding
1488 371916 execution tree xtree concept
1489 372120 c++ demangler demangles symbols which are not c++
1490 372185 Support of valgrind on ARMv8 with 32 bit executable
1491 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1492 372195 Power PC, xxsel instruction is not always recognized.
1493 372504 Hanging on exit_group
1494 372600 process loops forever when fatal signals are arriving quickly
1495 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1496 373046 Stacks registered by core are never deregistered
1497 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1498 373086 Implement additional Xen hypercalls
1499 373192 Calling posix_spawn in glibc 2.24 completely broken
1500 373488 Support for fanotify API on ARM64 architecture
1501 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1502 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1503 373938 const IRExpr arguments for matchIRExpr()
1504 374719 some spelling fixes
1505 374963 increase valgrind's load address to prevent mmap failure
1506 375514 valgrind_get_tls_addr() does not work in case of static TLS
1507 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1508 for ML_(find_rx_mapping)()
1509 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1510 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1511 == 377159 "vex: the `impossible' happened" still present
1512 == 375150 Assertion 'tres.status == VexTransOK' failed
1513 == 378068 valgrind crashes on AVX2 function in FFmpeg
1514 376142 Segfaults on MIPS Cavium Octeon boards
1515 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1516 376455 Solaris: unhandled syscall lgrpsys(180)
1517 376518 Solaris: unhandled fast trap getlgrp(6)
1518 376611 ppc64 and arm64 don't know about prlimit64 syscall
1519 376729 PPC64, remove R2 from the clobber list
1521 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1522 to be wrongly marked as addressable
1523 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1524 PIE enabled by default
1525 377376 memcheck/tests/linux/getregset fails with glibc2.24
1526 377427 PPC64, lxv instruction failing on odd destination register
1527 377478 PPC64: ISA 3.0 setup fixes
1528 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1529 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1530 and 2 args for FUTEX_TRYLOCK_PI
1531 377717 Fix massive space leak when reading compressed debuginfo sections
1532 377891 Update Xen 4.6 domctl wrappers
1533 377930 fcntl syscall wrapper is missing flock structure check
1534 378524 libvexmultiarch_test regression on s390x and ppc64
1535 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1536 378673 Update libiberty demangler
1537 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1538 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1539 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1540 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1541 (task_register_dyld_image_infos)
1542 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1543 (task_register_dyld_shared_cache_image_info)
1544 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1545 379473 MIPS: add support for rdhwr cycle counter register
1546 379504 remove TileGX/Linux port
1547 379525 Support more x86 nop opcodes
1548 379838 disAMode(x86): not an addr!
1549 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1550 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1551 379895 clock_gettime does not execute POST syscall wrapper
1552 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1553 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1554 380200 xtree generated callgrind files refer to files without directory name
1555 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1556 380397 s390x: __GI_strcspn() replacement needed
1557 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1559 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1560 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1561 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1565 Release 3.12.0 (20 October 2016)
1566 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1568 3.12.0 is a feature release with many improvements and the usual
1569 collection of bug fixes.
1571 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1572 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1573 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1574 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1575 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1576 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1578 * ================== PLATFORM CHANGES =================
1580 * POWER: Support for ISA 3.0 has been added
1582 * mips: support for O32 FPXX ABI has been added.
1583 * mips: improved recognition of different processors
1584 * mips: determination of page size now done at run time
1586 * amd64: Partial support for AMD FMA4 instructions.
1588 * arm, arm64: Support for v8 crypto and CRC instructions.
1590 * Improvements and robustification of the Solaris port.
1592 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1594 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1595 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1596 where possible. Valgrind's support for 32-bit x86 has stagnated in
1597 recent years and has fallen far behind that for 64-bit x86
1598 instructions. By contrast 64-bit x86 is well supported, up to and
1601 * ==================== TOOL CHANGES ====================
1605 - Added meta mempool support for describing a custom allocator which:
1606 - Auto-frees all chunks assuming that destroying a pool destroys all
1608 - Uses itself to allocate other memory blocks
1610 - New flag --ignore-range-below-sp to ignore memory accesses below
1611 the stack pointer, if you really have to. The related flag
1612 --workaround-gcc296-bugs=yes is now deprecated. Use
1613 --ignore-range-below-sp=1024-1 as a replacement.
1617 - Improved thread startup time significantly on non-Linux platforms.
1621 - Added collection of the metric "tot-blocks-allocd"
1623 * ==================== OTHER CHANGES ====================
1625 * Replacement/wrapping of malloc/new related functions is now done not just
1626 for system libraries by default, but for any globally defined malloc/new
1627 related function (both in shared libraries and statically linked alternative
1628 malloc implementations). The dynamic (runtime) linker is excluded, though.
1629 To only intercept malloc/new related functions in
1630 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1631 "nouserintercepts" can be any non-existing library name).
1632 This new functionality is not implemented for MacOS X.
1634 * The maximum number of callers in a suppression entry is now equal to
1635 the maximum size for --num-callers (500).
1636 Note that --gen-suppressions=yes|all similarly generates suppressions
1637 containing up to --num-callers frames.
1639 * New and modified GDB server monitor features:
1641 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1642 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1645 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1646 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1649 * Valgrind is able to read compressed debuginfo sections in two formats:
1650 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1651 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1653 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1654 for the most common use case (x86_64-linux, Memcheck) has been
1657 * Improved performance for programs that do a lot of discarding of
1658 instruction address ranges of 8KB or less.
1660 * The C++ symbol demangler has been updated.
1662 * More robustness against invalid syscall parameters on Linux.
1664 * ==================== FIXED BUGS ====================
1666 The following bugs have been fixed or resolved. Note that "n-i-bz"
1667 stands for "not in bugzilla" -- that is, a bug that was reported to us
1668 but never got a bugzilla entry. We encourage you to file bugs in
1669 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1670 than mailing the developers (or mailing lists) directly -- bugs that
1671 are not entered into bugzilla tend to get forgotten about or ignored.
1673 To see details of a given bug, visit
1674 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1675 where XXXXXX is the bug number as listed below.
1677 191069 Exiting due to signal not reported in XML output
1678 199468 Suppressions: stack size limited to 25
1679 while --num-callers allows more frames
1680 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1681 278744 cvtps2pd with redundant RexW
1682 303877 valgrind doesn't support compressed debuginfo sections.
1683 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1684 348345 Assertion fails for negative lineno
1685 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1686 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1687 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1688 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1689 352197 mips: mmap2() not wrapped correctly for page size > 4096
1690 353083 arm64 doesn't implement various xattr system calls
1691 353084 arm64 doesn't support sigpending system call
1692 353137 www: update info for Supported Platforms
1693 353138 www: update "The Valgrind Developers" page
1694 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1697 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1698 353398 WARNING: unhandled amd64-solaris syscall: 207
1699 353660 XML in auxwhat tag not escaping reserved symbols properly
1700 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1701 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1702 353802 ELF debug info reader confused with multiple .rodata sections
1703 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1704 353917 unhandled amd64-solaris syscall fchdir(120)
1705 353920 unhandled amd64-solaris syscall: 170
1706 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1707 354392 unhandled amd64-solaris syscall: 171
1708 354797 Vbit test does not include Iops for Power 8 instruction support
1709 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1713 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1714 355188 valgrind should intercept all malloc related global functions
1715 355454 do not intercept malloc related symbols from the runtime linker
1716 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1717 356044 Dwarf line info reader misinterprets is_stmt register
1718 356112 mips: replace addi with addiu
1719 356393 valgrind (vex) crashes because isZeroU happened
1722 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1723 356678 arm64-linux: unhandled syscall 232 (mincore)
1724 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1725 356823 Unsupported ARM instruction: stlex
1726 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1727 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1728 357673 crash if I try to run valgrind with a binary link with libcurl
1729 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1730 357871 pthread_spin_destroy not properly wrapped
1731 357887 Calls to VG_(fclose) do not close the file descriptor
1732 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1733 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1734 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1735 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1736 359181 Buffer Overflow during Demangling
1737 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1738 359289 s390x: popcnt (B9E1) not implemented
1739 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1740 359503 Add missing syscalls for aarch64 (arm64)
1741 359645 "You need libc6-dbg" help message could be more helpful
1742 359703 s390: wire up separate socketcalls system calls
1743 359724 getsockname might crash - deref_UInt should call safe_to_deref
1744 359733 amd64 implement ld.so strchr/index override like x86
1745 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1746 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1748 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1749 359871 Incorrect mask handling in ppoll
1750 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1751 360008 Contents of Power vr registers contents is not printed correctly when
1752 the --vgdb-shadow-registers=yes option is used
1753 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1754 non-zero shadow bits
1755 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1756 360425 arm64 unsupported instruction ldpsw
1758 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1759 360571 Error about the Android Runtime reading below the stack pointer on ARM
1760 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1761 360749 kludge for multiple .rodata sections on Solaris no longer needed
1762 360752 raise the number of reserved fds in m_main.c from 10 to 12
1763 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1764 361226 s390x: risbgn (EC59) not implemented
1765 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1766 361354 ppc64[le]: wire up separate socketcalls system calls
1767 361615 Inconsistent termination for multithreaded process terminated by signal
1768 361926 Unhandled Solaris syscall: sysfs(84)
1769 362009 V dumps core on unimplemented functionality before threads are created
1770 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1771 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1772 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1773 362953 Request for an update to the Valgrind Developers page
1774 363680 add renameat2() support
1775 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1776 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1777 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1778 364058 clarify in manual limitations of array overruns detections
1779 364413 pselect sycallwrapper mishandles NULL sigmask
1780 364728 Power PC, missing support for several HW registers in
1781 get_otrack_shadow_offset_wrk()
1782 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1783 365273 Invalid write to stack location reported after signal handler runs
1784 365912 ppc64BE segfault during jm-insns test (RELRO)
1785 366079 FPXX Support for MIPS32 Valgrind
1786 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1787 366344 Multiple unhandled instruction for Aarch64
1788 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1789 367995 Integration of memcheck with custom memory allocator
1790 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1791 368412 False positive result for altivec capability check
1792 368416 Add tc06_two_races_xml.exp output for ppc64
1793 368419 Perf Events ioctls not implemented
1794 368461 mmapunmap test fails on ppc64
1795 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1796 369000 AMD64 fma4 instructions unsupported.
1797 369169 ppc64 fails jm_int_isa_2_07 test
1798 369175 jm_vec_isa_2_07 test crashes on ppc64
1799 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1800 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1801 369359 msghdr_foreachfield can crash when handling bad iovec
1802 369360 Bad sigprocmask old or new sets can crash valgrind
1803 369361 vmsplice syscall wrapper crashes on bad iovec
1804 369362 Bad sigaction arguments crash valgrind
1805 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1806 369402 Bad set/get_thread_area pointer crashes valgrind
1807 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1808 369446 valgrind crashes on unknown fcntl command
1809 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1810 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1811 370265 ISA 3.0 HW cap stuff needs updating
1812 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1813 372195 Power PC, xxsel instruction is not always recognized
1815 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1816 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1817 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1818 n-i-bz Fix ppoll_alarm exclusion on OS X
1819 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1820 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1821 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1822 n-i-bz amd64: memcheck false positive with shr %edx
1823 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1824 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1825 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1826 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1828 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1829 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1830 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1834 Release 3.11.0 (22 September 2015)
1835 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1837 3.11.0 is a feature release with many improvements and the usual
1838 collection of bug fixes.
1840 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1841 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1842 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1843 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1844 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1845 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1847 * ================== PLATFORM CHANGES =================
1849 * Support for Solaris/x86 and Solaris/amd64 has been added.
1851 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1853 * Preliminary support for the Tilera TileGX architecture has been added.
1855 * s390x: It is now required for the host to have the "long displacement"
1856 facility. The oldest supported machine model is z990.
1858 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1859 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1861 * The JIT's register allocator is significantly faster, making the JIT
1862 as a whole somewhat faster, so JIT-intensive activities, for example
1863 program startup, are modestly faster, around 5%.
1865 * There have been changes to the default settings of several command
1866 line flags, as detailed below.
1868 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1869 capable hosts, the simulated CPUID will now indicate AVX2 support.
1871 * ==================== TOOL CHANGES ====================
1875 - The default value for --leak-check-heuristics has been changed from
1876 "none" to "all". This helps to reduce the number of possibly
1877 lost blocks, in particular for C++ applications.
1879 - The default value for --keep-stacktraces has been changed from
1880 "malloc-then-free" to "malloc-and-free". This has a small cost in
1881 memory (one word per malloc-ed block) but allows Memcheck to show the
1882 3 stacktraces of a dangling reference: where the block was allocated,
1883 where it was freed, and where it is acccessed after being freed.
1885 - The default value for --partial-loads-ok has been changed from "no" to
1886 "yes", so as to avoid false positive errors resulting from some kinds
1887 of vectorised loops.
1889 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1890 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1891 than get_vbits when you need to associate byte data value with
1892 their corresponding validity bits.
1894 - The 'block_list' monitor command has been enhanced:
1895 o it can print a range of loss records
1896 o it now accepts an optional argument 'limited <max_blocks>'
1897 to control the number of blocks printed.
1898 o if a block has been found using a heuristic, then
1899 'block_list' now shows the heuristic after the block size.
1900 o the loss records/blocks to print can be limited to the blocks
1901 found via specified heuristics.
1903 - The C helper functions used to instrument loads on
1904 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1905 replaced by handwritten assembly sequences. This gives speedups
1906 in the region of 0% to 7% for those targets only.
1908 - A new command line option, --expensive-definedness-checks=yes|no,
1909 has been added. This is useful for avoiding occasional invalid
1910 uninitialised-value errors in optimised code. Watch out for
1911 runtime degradation, as this can be up to 25%. As always, though,
1912 the slowdown is highly application specific. The default setting
1917 - A new monitor command 'all_snapshots <filename>' dumps all
1918 snapshots taken so far.
1922 - Significant memory reduction and moderate speedups for
1923 --history-level=full for applications accessing a lot of memory
1924 with many different stacktraces.
1926 - The default value for --conflict-cache-size=N has been doubled to
1927 2000000. Users that were not using the default value should
1928 preferably also double the value they give.
1930 The default was changed due to the changes in the "full history"
1931 implementation. Doubling the value gives on average a slightly more
1932 complete history and uses similar memory (or significantly less memory
1933 in the worst case) than the previous implementation.
1935 - The Helgrind monitor command 'info locks' now accepts an optional
1936 argument 'lock_addr', which shows information about the lock at the
1939 - When using --history-level=full, the new Helgrind monitor command
1940 'accesshistory <addr> [<len>]' will show the recorded accesses for
1941 <len> (or 1) bytes at <addr>.
1943 * ==================== OTHER CHANGES ====================
1945 * The default value for the --smc-check option has been changed from
1946 "stack" to "all-non-file" on targets that provide automatic D-I
1947 cache coherence (x86, amd64 and s390x). The result is to provide,
1948 by default, transparent support for JIT generated and self-modifying
1949 code on all targets.
1951 * Mac OS X only: the default value for the --dsymutil option has been
1952 changed from "no" to "yes", since any serious usage on Mac OS X
1953 always required it to be "yes".
1955 * The command line options --db-attach and --db-command have been removed.
1956 They were deprecated in 3.10.0.
1958 * When a process dies due to a signal, Valgrind now shows the signal
1959 and the stacktrace at default verbosity (i.e. verbosity 1).
1961 * The address description logic used by Memcheck and Helgrind now
1962 describes addresses in anonymous segments, file mmap-ed segments,
1963 shared memory segments and the brk data segment.
1965 * The new option --error-markers=<begin>,<end> can be used to mark the
1966 begin/end of errors in textual output mode, to facilitate
1967 searching/extracting errors in output files that mix valgrind errors
1968 with program output.
1970 * The new option --max-threads=<number> can be used to change the number
1971 of threads valgrind can handle. The default is 500 threads which
1972 should be more than enough for most applications.
1974 * The new option --valgrind-stacksize=<number> can be used to change the
1975 size of the private thread stacks used by Valgrind. This is useful
1976 for reducing memory use or increasing the stack size if Valgrind
1977 segfaults due to stack overflow.
1979 * The new option --avg-transtab-entry-size=<number> can be used to specify
1980 the expected instrumented block size, either to reduce memory use or
1981 to avoid excessive retranslation.
1983 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1985 * New and modified GDB server monitor features:
1987 - When a signal is reported in GDB, you can now use the GDB convenience
1988 variable $_siginfo to examine detailed signal information.
1990 - Valgrind's gdbserver now allows the user to change the signal
1991 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1992 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1993 continue without passing the signal to the process.
1995 - With GDB >= 7.10, the command 'target remote'
1996 will automatically load the executable file of the process running
1997 under Valgrind. This means you do not need to specify the executable
1998 file yourself, GDB will discover it itself. See GDB documentation about
1999 'qXfer:exec-file:read' packet for more info.
2001 * ==================== FIXED BUGS ====================
2003 The following bugs have been fixed or resolved. Note that "n-i-bz"
2004 stands for "not in bugzilla" -- that is, a bug that was reported to us
2005 but never got a bugzilla entry. We encourage you to file bugs in
2006 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2007 than mailing the developers (or mailing lists) directly -- bugs that
2008 are not entered into bugzilla tend to get forgotten about or ignored.
2010 To see details of a given bug, visit
2011 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2012 where XXXXXX is the bug number as listed below.
2014 116002 VG_(printf): Problems with justification of strings and integers
2015 155125 avoid cutting away file:lineno after long function name
2016 197259 Unsupported arch_prtctl PR_SET_GS option
2017 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
2018 201216 Fix Valgrind does not support pthread_sigmask() on OS X
2019 201435 Fix Darwin: -v does not show kernel version
2020 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
2021 211256 Fixed an outdated comment regarding the default platform.
2022 211529 Incomplete call stacks for code compiled by newer versions of MSVC
2023 211926 Avoid compilation warnings in valgrind.h with -pedantic
2024 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
2026 226609 Crediting upstream authors in man page
2027 231257 Valgrind omits path when executing script from shebang line
2028 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
2029 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
2030 269360 s390x: Fix addressing mode selection for compare-and-swap
2031 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
2033 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
2034 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
2035 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
2036 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
2037 330147 libmpiwrap PMPI_Get_count returns undefined value
2038 333051 mmap of huge pages fails due to incorrect alignment
2040 334802 valgrind does not always explain why a given option is bad
2041 335618 mov.w rN, pc/sp (ARM32)
2042 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
2047 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
2048 338602 AVX2 bit in CPUID missing
2049 338606 Strange message for scripts with invalid interpreter
2050 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
2051 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
2052 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
2054 339156 gdbsrv not called for fatal signal
2055 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
2056 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
2057 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
2058 339442 Fix testsuite build failure on OS X 10.9
2059 339542 Enable compilation with Intel's ICC compiler
2060 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
2061 339688 Mac-specific ASM does not support .version directive (cpuid,
2062 tronical and pushfpopf tests)
2063 339745 Valgrind crash when check Marmalade app (partial fix)
2064 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
2065 339778 Linux/TileGx platform support for Valgrind
2066 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
2067 339789 Fix none/tests/execve test on Mac OS X 10.9
2068 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
2069 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
2070 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
2071 340392 Allow user to select more accurate definedness checking in memcheck
2072 to avoid invalid complaints on optimised code
2073 340430 Fix some grammatical weirdness in the manual.
2074 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
2075 341419 Signal handler ucontext_t not filled out correctly on OS X
2076 341539 VG_(describe_addr) should not describe address as belonging to client
2077 segment if it is past the heap end
2078 341613 Enable building of manythreads and thread-exits tests on Mac OS X
2079 341615 Fix none/tests/darwin/access_extended test on Mac OS X
2080 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
2081 341789 aarch64: shmat fails with valgrind on ARMv8
2082 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
2083 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
2084 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
2085 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
2086 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
2087 342221 socket connect false positive uninit memory for unknown af family
2088 342353 Allow dumping full massif output while valgrind is still running
2089 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
2093 342603 Add I2C_SMBUS ioctl support
2094 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
2095 342683 Mark memory past the initial brk limit as unaddressable
2096 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
2097 342795 Internal glibc __GI_mempcpy call should be intercepted
2098 342841 s390x: Support instructions fiebr(a) and fidbr(a)
2099 343012 Unhandled syscall 319 (memfd_create)
2100 343069 Patch updating v4l2 API support
2101 343173 helgrind crash during stack unwind
2102 343219 fix GET_STARTREGS for arm
2103 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
2104 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
2105 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
2106 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
2107 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
2108 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
2109 343597 ppc64le: incorrect use of offseof macro
2110 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
2111 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
2112 343732 Unhandled syscall 144 (setgid) on aarch64
2113 343733 Unhandled syscall 187 (msgctl and related) on aarch64
2114 343802 s390x: False positive "conditional jump or move depends on [..]
2115 343902 --vgdb=yes doesn't break when --xml=yes is used
2116 343967 Don't warn about setuid/setgid/setcap executable for directories
2117 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
2118 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
2119 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
2120 344054 www - update info for Solaris/illumos
2121 344416 'make regtest' does not work cleanly on OS X
2122 344235 Remove duplicate include of pub_core_aspacemgr.h
2123 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
2124 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
2125 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
2126 344314 callgrind_annotate ... warnings about commands containing newlines
2127 344318 socketcall should wrap recvmmsg and sendmmsg
2128 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
2129 344416 Fix 'make regtest' does not work cleanly on OS X
2130 344499 Fix compilation for Linux kernel >= 4.0.0
2131 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
2132 unix:349 (__pthread_fchdir)
2133 344559 Garbage collection of unused segment names in address space manager
2134 344560 Fix stack traces missing penultimate frame on OS X
2135 344621 Fix memcheck/tests/err_disable4 test on OS X
2136 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
2137 344702 Fix missing libobjc suppressions on OS X 10.10
2139 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
2140 344939 Fix memcheck/tests/xml1 on OS X 10.10
2141 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
2142 345079 Fix build problems in VEX/useful/test_main.c
2143 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
2144 345177 arm64: prfm (reg) not implemented
2145 345215 Performance improvements for the register allocator
2146 345248 add support for Solaris OS in valgrind
2147 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
2148 345394 Fix memcheck/tests/strchr on OS X
2149 345637 Fix memcheck/tests/sendmsg on OS X
2150 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
2151 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
2152 345887 Fix an assertion in the address space manager
2153 345928 amd64: callstack only contains current function for small stacks
2154 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
2155 345987 MIPS64: Implement cavium LHX instruction
2156 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
2157 346185 Fix typo saving altivec register v24
2158 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
2159 and LibVEX_GuestPPC64_get_CR()
2160 346270 Regression tests none/tests/jm_vec/isa_2_07 and
2161 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
2162 346307 fuse filesystem syscall deadlocks
2163 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
2164 346411 MIPS: SysRes::_valEx handling is incorrect
2165 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
2166 346474 PPC64 Power 8, spr TEXASRU register not supported
2167 346487 Compiler generates "note" about a future ABI change for PPC64
2168 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
2169 and causing spurious "invalid read of size 8" warnings
2170 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
2171 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
2172 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
2173 347322 Power PC regression test cleanup
2174 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
2176 347389 unhandled syscall: 373 (Linux ARM syncfs)
2177 347686 Patch set to cleanup PPC64 regtests
2178 347978 Remove bash dependencies where not needed
2179 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
2180 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
2182 348102 Patch updating v4l2 API support
2183 348247 amd64 front end: jno jumps wrongly when overflow is not set
2184 348269 Improve mmap MAP_HUGETLB support.
2185 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
2186 348345 Assertion fails for negative lineno
2187 348377 Unsupported ARM instruction: yield
2188 348565 Fix detection of command line option availability for clang
2189 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
2190 348728 Fix broken check for VIDIOC_G_ENC_INDEX
2191 348748 Fix redundant condition
2192 348890 Fix clang warning about unsupported --param inline-unit-growth=900
2193 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
2194 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
2195 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
2196 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
2197 349626 Implemented additional Xen hypercalls
2198 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
2199 349790 Clean up of the hardware capability checking utilities.
2200 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
2201 349874 Fix typos in source code
2202 349879 memcheck: add handwritten assembly for helperc_LOADV*
2203 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
2204 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
2205 350202 Add limited param to 'monitor block_list'
2206 350290 s390x: Support instructions fixbr(a)
2207 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
2208 350809 Fix none/tests/async-sigs for Solaris
2209 350811 Remove reference to --db-attach which has been removed.
2210 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
2211 350854 hard-to-understand code in VG_(load_ELF)()
2212 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2213 351386 Solaris: Cannot run ld.so.1 under Valgrind
2214 351474 Fix VG_(iseqsigset) as obvious
2215 351531 Typo in /include/vki/vki-xen-physdev.h header guard
2216 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
2217 351858 ldsoexec support on Solaris
2218 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
2219 352130 helgrind reports false races for printfs using mempcpy on FILE* state
2220 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
2221 352320 arm64 crash on none/tests/nestedfs
2222 352765 Vbit test fails on Power 6
2223 352768 The mbar instruction is missing from the Power PC support
2224 352769 Power PC program priority register (PPR) is not supported
2225 n-i-bz Provide implementations of certain compiler builtins to support
2226 compilers that may not provide those
2227 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
2228 n-i-bz Fix compilation on distros with glibc < 2.5
2229 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
2230 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
2231 n-i-bz mremap did not work properly on shared memory
2232 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
2233 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
2235 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
2236 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
2237 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
2241 Release 3.10.1 (25 November 2014)
2242 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2243 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
2244 and backports fixes for all reported missing AArch64 ARMv8 instructions
2245 and syscalls from the trunk. If you package or deliver 3.10.0 for others
2246 to use, you might want to consider upgrading to 3.10.1 instead.
2248 The following bugs have been fixed or resolved. Note that "n-i-bz"
2249 stands for "not in bugzilla" -- that is, a bug that was reported to us
2250 but never got a bugzilla entry. We encourage you to file bugs in
2251 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2252 than mailing the developers (or mailing lists) directly -- bugs that
2253 are not entered into bugzilla tend to get forgotten about or ignored.
2255 To see details of a given bug, visit
2256 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2257 where XXXXXX is the bug number as listed below.
2259 335440 arm64: ld1 (single structure) is not implemented
2260 335713 arm64: unhanded instruction: prfm (immediate)
2261 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
2262 339182 ppc64: AvSplat ought to load destination vector register with [..]
2263 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
2264 339433 ppc64 lxvw4x instruction uses four 32-byte loads
2265 339645 Use correct tag names in sys_getdents/64 wrappers
2266 339706 Fix false positive for ioctl(TIOCSIG) on linux
2267 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
2268 339853 arm64 times syscall unknown
2269 339855 arm64 unhandled getsid/setsid syscalls
2270 339858 arm64 dmb sy not implemented
2271 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
2272 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
2273 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
2275 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
2276 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
2277 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
2278 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
2279 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
2280 340509 arm64: unhandled instruction fcvtas
2281 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
2282 340632 arm64: unhandled instruction fcvtas
2283 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
2284 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
2285 340788 warning: unhandled syscall: 318 (getrandom)
2286 340807 disInstr(arm): unhandled instruction: 0xEE989B20
2287 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
2288 340922 arm64: unhandled getgroups/setgroups syscalls
2289 350251 Fix typo in VEX utility program (test_main.c).
2290 350407 arm64: unhandled instruction ucvtf (vector, integer)
2291 350809 none/tests/async-sigs breaks when run under cron on Solaris
2292 350811 update README.solaris after r15445
2293 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
2294 350854 strange code in VG_(load_ELF)()
2295 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2296 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
2297 n-i-bz Add missing ]] to terminate CDATA.
2298 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
2299 n-i-bz Enable sys_fadvise64_64 on arm32.
2300 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
2301 n-i-bz Add test cases for all known arm64 load/store instructions.
2302 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
2303 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
2304 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
2305 n-i-bz arm64: Implement "fcvtpu w, s".
2306 n-i-bz arm64: implement ADDP and various others
2307 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
2308 n-i-bz arm64: enable FCVT{A,N}S X,S.
2310 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
2314 Release 3.10.0 (10 September 2014)
2315 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2317 3.10.0 is a feature release with many improvements and the usual
2318 collection of bug fixes.
2320 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
2321 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
2322 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
2323 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
2324 significantly improved relative to the 3.9.0 release.
2326 * ================== PLATFORM CHANGES =================
2328 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
2329 is mostly complete, and is usable, but some SIMD instructions are as
2332 * Support for little-endian variant of the 64-bit POWER architecture.
2334 * Support for Android on MIPS32.
2336 * Support for 64bit FPU on MIPS32 platforms.
2338 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
2340 * Configuration for and running on Android targets has changed.
2341 See README.android in the source tree for details.
2343 * ================== DEPRECATED FEATURES =================
2345 * --db-attach is now deprecated and will be removed in the next
2346 valgrind feature release. The built-in GDB server capabilities are
2347 superior and should be used instead. Learn more here:
2348 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
2350 * ==================== TOOL CHANGES ====================
2354 - Client code can now selectively disable and re-enable reporting of
2355 invalid address errors in specific ranges using the new client
2356 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
2357 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
2359 - Leak checker: there is a new leak check heuristic called
2360 "length64". This is used to detect interior pointers pointing 8
2361 bytes inside a block, on the assumption that the first 8 bytes
2362 holds the value "block size - 8". This is used by
2363 sqlite3MemMalloc, for example.
2365 - Checking of system call parameters: if a syscall parameter
2366 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
2367 several fields not initialised, an error is now reported for each
2368 field. Previously, an error was reported only for the first
2369 uninitialised field.
2371 - Mismatched alloc/free checking: a new flag
2372 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
2373 such checks if necessary.
2377 - Improvements to error messages:
2379 o Race condition error message involving heap allocated blocks also
2380 show the thread number that allocated the raced-on block.
2382 o All locks referenced by an error message are now announced.
2383 Previously, some error messages only showed the lock addresses.
2385 o The message indicating where a lock was first observed now also
2386 describes the address/location of the lock.
2388 - Helgrind now understands the Ada task termination rules and
2389 creates a happens-before relationship between a terminated task
2390 and its master. This avoids some false positives and avoids a big
2391 memory leak when a lot of Ada tasks are created and terminated.
2392 The interceptions are only activated with forthcoming releases of
2393 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
2395 - A new GDB server monitor command "info locks" giving the list of
2396 locks, their location, and their status.
2400 - callgrind_control now supports the --vgdb-prefix argument,
2401 which is needed if valgrind was started with this same argument.
2403 * ==================== OTHER CHANGES ====================
2405 * Unwinding through inlined function calls. Stack unwinding can now
2406 make use of Dwarf3 inlined-unwind information if it is available.
2407 The practical effect is that inlined calls become visible in stack
2408 traces. The suppression matching machinery has been adjusted
2409 accordingly. This is controlled by the new option
2410 --read-inline-info=yes|no. Currently this is enabled by default
2411 only on Linux and Android targets and only for the tools Memcheck,
2414 * Valgrind can now read EXIDX unwind information on 32-bit ARM
2415 targets. If an object contains both CFI and EXIDX unwind
2416 information, Valgrind will prefer the CFI over the EXIDX. This
2417 facilitates unwinding through system libraries on arm-android
2420 * Address description logic has been improved and is now common
2421 between Memcheck and Helgrind, resulting in better address
2422 descriptions for some kinds of error messages.
2424 * Error messages about dubious arguments (eg, to malloc or calloc) are
2425 output like other errors. This means that they can be suppressed
2426 and they have a stack trace.
2428 * The C++ demangler has been updated for better C++11 support.
2430 * New and modified GDB server monitor features:
2432 - Thread local variables/storage (__thread) can now be displayed.
2434 - The GDB server monitor command "v.info location <address>"
2435 displays information about an address. The information produced
2436 depends on the tool and on the options given to valgrind.
2437 Possibly, the following are described: global variables, local
2438 (stack) variables, allocated or freed blocks, ...
2440 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
2441 ask the GDB server to stop at the start of program execution, at
2442 the end of the program execution and on Valgrind internal errors.
2444 - A new monitor command "v.info stats" shows various Valgrind core
2445 and tool statistics.
2447 - A new monitor command "v.set hostvisibility" allows the GDB server
2448 to provide access to Valgrind internal host status/memory.
2450 * A new option "--aspace-minaddr=<address>" can in some situations
2451 allow the use of more memory by decreasing the address above which
2452 Valgrind maps memory. It can also be used to solve address
2453 conflicts with system libraries by increasing the default value.
2454 See user manual for details.
2456 * The amount of memory used by Valgrind to store debug info (unwind
2457 info, line number information and symbol data) has been
2458 significantly reduced, even though Valgrind now reads more
2459 information in order to support unwinding of inlined function calls.
2461 * Dwarf3 handling with --read-var-info=yes has been improved:
2463 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
2465 - Code compiled with
2466 -ffunction-sections -fdata-sections -Wl,--gc-sections
2467 no longer causes assertion failures.
2469 * Improved checking for the --sim-hints= and --kernel-variant=
2470 options. Unknown strings are now detected and reported to the user
2473 * The semantics of stack start/end boundaries in the valgrind.h
2474 VALGRIND_STACK_REGISTER client request has been clarified and
2475 documented. The convention is that start and end are respectively
2476 the lowest and highest addressable bytes of the stack.
2478 * ==================== FIXED BUGS ====================
2480 The following bugs have been fixed or resolved. Note that "n-i-bz"
2481 stands for "not in bugzilla" -- that is, a bug that was reported to us
2482 but never got a bugzilla entry. We encourage you to file bugs in
2483 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2484 than mailing the developers (or mailing lists) directly -- bugs that
2485 are not entered into bugzilla tend to get forgotten about or ignored.
2487 To see details of a given bug, visit
2488 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2489 where XXXXXX is the bug number as listed below.
2491 175819 Support for ipv6 socket reporting with --track-fds
2492 232510 make distcheck fails
2493 249435 Analyzing wine programs with callgrind triggers a crash
2494 278972 support for inlined function calls in stacktraces and suppression
2496 291310 FXSAVE instruction marks memory as undefined on amd64
2497 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2498 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2499 315199 vgcore file for threaded app does not show which thread crashed
2500 315952 tun/tap ioctls are not supported
2501 323178 Unhandled instruction: PLDW register (ARM)
2502 323179 Unhandled instruction: PLDW immediate (ARM)
2503 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2504 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2505 325124 [MIPSEL] Compilation error
2506 325477 Phase 4 support for IBM Power ISA 2.07
2507 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2508 325628 Phase 5 support for IBM Power ISA 2.07
2509 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2510 325751 Missing the two privileged Power PC Transactional Memory Instructions
2511 325816 Phase 6 support for IBM Power ISA 2.07
2512 325856 Make SGCheck fail gracefully on unsupported platforms
2513 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2514 326436 DRD: False positive in libstdc++ std::list::push_back
2515 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2516 326462 Refactor vgdb to isolate invoker stuff into separate module
2517 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2518 326623 DRD: false positive conflict report in a field assignment
2519 326724 Valgrind does not compile on OSX 1.9 Mavericks
2520 326816 Intercept for __strncpy_sse2_unaligned missing?
2521 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2522 326983 Clear direction flag after tests on amd64.
2523 327212 Do not prepend the current directory to absolute path names.
2524 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2525 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2526 327284 s390x: Fix translation of the risbg instruction
2527 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2528 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2529 327916 DW_TAG_typedef may have no name
2530 327943 s390x: add a redirection for the 'index' function
2531 328100 XABORT not implemented
2532 328205 Implement additional Xen hypercalls
2533 328454 add support Backtraces with ARM unwind tables (EXIDX)
2534 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2535 328711 valgrind.1 manpage "memcheck options" section is badly generated
2536 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2537 329612 Incorrect handling of AT_BASE for image execution
2538 329694 clang warns about using uninitialized variable
2539 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2540 330228 mmap must align to VKI_SHMLBA on mips32
2541 330257 LLVM does not support `-mno-dynamic-no-pic` option
2542 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2543 330459 --track-fds=yes doesn't track eventfds
2544 330469 Add clock_adjtime syscall support
2545 330594 Missing sysalls on PowerPC / uClibc
2546 330622 Add test to regression suite for POWER instruction: dcbzl
2547 330939 Support for AMD's syscall instruction on x86
2549 330941 Typo in PRE(poll) syscall wrapper
2550 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2551 331254 Fix expected output for memcheck/tests/dw4
2552 331255 Fix race condition in test none/tests/coolo_sigaction
2553 331257 Fix type of jump buffer in test none/tests/faultstatus
2554 331305 configure uses bash specific syntax
2555 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2556 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2557 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2558 331829 Unexpected ioctl opcode sign extension
2559 331830 ppc64: WARNING: unhandled syscall: 96/97
2560 331839 drd/tests/sem_open specifies invalid semaphore name
2561 331847 outcome of drd/tests/thread_name is nondeterministic
2562 332037 Valgrind cannot handle Thumb "add pc, reg"
2563 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2564 consistency checks enabled
2565 332263 intercepts for pthread_rwlock_timedrdlock and
2566 pthread_rwlock_timedwrlock are incorrect
2567 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2569 332276 Implement additional Xen hypercalls
2570 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2571 332765 Fix ms_print to create temporary files in a proper directory
2572 333072 drd: Add semaphore annotations
2573 333145 Tests for missaligned PC+#imm access for arm
2574 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2575 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2576 333248 WARNING: unhandled syscall: unix:443
2577 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2578 333501 cachegrind: assertion: Cache set count is not a power of two.
2581 333666 Recognize MPX instructions and bnd prefix.
2582 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2583 333817 Valgrind reports the memory areas written to by the SG_IO
2585 334049 lzcnt fails silently (x86_32)
2586 334384 Valgrind does not have support Little Endian support for
2588 334585 recvmmsg unhandled (+patch) (arm)
2589 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2590 334727 Build fails with -Werror=format-security
2591 334788 clarify doc about --log-file initial program directory
2592 334834 PPC64 Little Endian support, patch 2
2593 334836 PPC64 Little Endian support, patch 3 testcase fixes
2594 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2595 335034 Unhandled ioctl: HCIGETDEVLIST
2596 335155 vgdb, fix error print statement.
2597 335262 arm64: movi 8bit version is not supported
2598 335263 arm64: dmb instruction is not implemented
2599 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2600 335496 arm64: sbc/abc instructions are not implemented
2601 335554 arm64: unhandled instruction: abs
2602 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2603 335735 arm64: unhandled instruction: cnt
2604 335736 arm64: unhandled instruction: uaddlv
2605 335848 arm64: unhandled instruction: {s,u}cvtf
2606 335902 arm64: unhandled instruction: sli
2607 335903 arm64: unhandled instruction: umull (vector)
2608 336055 arm64: unhandled instruction: mov (element)
2609 336062 arm64: unhandled instruction: shrn{,2}
2610 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2611 336189 arm64: unhandled Instruction: mvn
2612 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2613 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2614 336772 Make moans about unknown ioctls more informative
2615 336957 Add a section about the Solaris/illumos port on the webpage
2616 337094 ifunc wrapper is broken on ppc64
2617 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2618 337528 leak check heuristic for block prefixed by length as 64bit number
2619 337740 Implement additional Xen hypercalls
2620 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2621 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2622 337871 deprecate --db-attach
2623 338023 Add support for all V4L2/media ioctls
2624 338024 inlined functions are not shown if DW_AT_ranges is used
2625 338106 Add support for 'kcmp' syscall
2626 338115 DRD: computed conflict set differs from actual after fork
2627 338160 implement display of thread local storage in gdbsrv
2628 338205 configure.ac and check for -Wno-tautological-compare
2629 338300 coredumps are missing one byte of every segment
2630 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2631 338499 --sim-hints parsing broken due to wrong order in tokens
2632 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2633 338681 Unable to unwind through clone thread created on i386-linux
2634 338698 race condition between gdbsrv and vgdb on startup
2635 338703 helgrind on arm-linux gets false positives in dynamic loader
2636 338791 alt dwz files can be relative of debug/main file
2637 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2638 338932 build V-trunk with gcc-trunk
2639 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2640 345079 Fix build problems in VEX/useful/test_main.c
2641 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2642 n-i-bz s390x: Fix memory corruption for multithreaded applications
2643 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2644 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2645 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2646 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2647 n-i-bz Add command-line option --default-suppressions
2648 n-i-bz Add support for BLKDISCARDZEROES ioctl
2649 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2650 n-i-bz Add support for sys_pivot_root and sys_unshare
2652 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2653 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2654 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2658 Release 3.9.0 (31 October 2013)
2659 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2660 3.9.0 is a feature release with many improvements and the usual
2661 collection of bug fixes.
2663 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2664 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2665 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2666 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2668 * ================== PLATFORM CHANGES =================
2670 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2671 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2673 * Support for MIPS DSP ASE on MIPS32 platforms.
2675 * Support for s390x Decimal Floating Point instructions on hosts that
2676 have the DFP facility installed.
2678 * Support for POWER8 (Power ISA 2.07) instructions
2680 * Support for Intel AVX2 instructions. This is available only on 64
2683 * Initial support for Intel Transactional Synchronization Extensions,
2686 * Initial support for Hardware Transactional Memory on POWER.
2688 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2689 run large GUI apps tolerably well.
2691 * ==================== TOOL CHANGES ====================
2695 - Improvements in handling of vectorised code, leading to
2696 significantly fewer false error reports. You need to use the flag
2697 --partial-loads-ok=yes to get the benefits of these changes.
2699 - Better control over the leak checker. It is now possible to
2700 specify which leak kinds (definite/indirect/possible/reachable)
2701 should be displayed, which should be regarded as errors, and which
2702 should be suppressed by a given leak suppression. This is done
2703 using the options --show-leak-kinds=kind1,kind2,..,
2704 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2705 "match-leak-kinds:" line in suppression entries, respectively.
2707 Note that generated leak suppressions contain this new line and
2708 are therefore more specific than in previous releases. To get the
2709 same behaviour as previous releases, remove the "match-leak-kinds:"
2710 line from generated suppressions before using them.
2712 - Reduced "possible leak" reports from the leak checker by the use
2713 of better heuristics. The available heuristics provide detection
2714 of valid interior pointers to std::stdstring, to new[] allocated
2715 arrays with elements having destructors and to interior pointers
2716 pointing to an inner part of a C++ object using multiple
2717 inheritance. They can be selected individually using the
2718 option --leak-check-heuristics=heur1,heur2,...
2720 - Better control of stacktrace acquisition for heap-allocated
2721 blocks. Using the --keep-stacktraces option, it is possible to
2722 control independently whether a stack trace is acquired for each
2723 allocation and deallocation. This can be used to create better
2724 "use after free" errors or to decrease Valgrind's resource
2725 consumption by recording less information.
2727 - Better reporting of leak suppression usage. The list of used
2728 suppressions (shown when the -v option is given) now shows, for
2729 each leak suppressions, how many blocks and bytes it suppressed
2730 during the last leak search.
2734 - False errors resulting from the use of statically initialised
2735 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2738 - False errors resulting from the use of pthread_cond_waits that
2739 timeout, have been removed.
2741 * ==================== OTHER CHANGES ====================
2743 * Some attempt to tune Valgrind's space requirements to the expected
2744 capabilities of the target:
2746 - The default size of the translation cache has been reduced from 8
2747 sectors to 6 on Android platforms, since each sector occupies
2748 about 40MB when using Memcheck.
2750 - The default size of the translation cache has been increased to 16
2751 sectors on all other platforms, reflecting the fact that large
2752 applications require instrumentation and storage of huge amounts
2753 of code. For similar reasons, the number of memory mapped
2754 segments that can be tracked has been increased by a factor of 6.
2756 - In all cases, the maximum number of sectors in the translation
2757 cache can be controlled by the new flag --num-transtab-sectors.
2759 * Changes in how debug info (line numbers, etc) is read:
2761 - Valgrind no longer temporarily mmaps the entire object to read
2762 from it. Instead, reading is done through a small fixed sized
2763 buffer. This avoids virtual memory usage spikes when Valgrind
2764 reads debuginfo from large shared objects.
2766 - A new experimental remote debug info server. Valgrind can read
2767 debug info from a different machine (typically, a build host)
2768 where debuginfo objects are stored. This can save a lot of time
2769 and hassle when running Valgrind on resource-constrained targets
2770 (phones, tablets) when the full debuginfo objects are stored
2771 somewhere else. This is enabled by the --debuginfo-server=
2774 - Consistency checking between main and debug objects can be
2775 disabled using the --allow-mismatched-debuginfo option.
2777 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2778 scanning can recover stack traces in some cases when the normal
2779 unwind mechanisms fail. Stack scanning is best described as "a
2780 nasty, dangerous and misleading hack" and so is disabled by default.
2781 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2784 * Detection and merging of recursive stack frame cycles. When your
2785 program has recursive algorithms, this limits the memory used by
2786 Valgrind for recorded stack traces and avoids recording
2787 uninteresting repeated calls. This is controlled by the command
2788 line option --merge-recursive-frame and by the monitor command
2789 "v.set merge-recursive-frames".
2791 * File name and line numbers for used suppressions. The list of used
2792 suppressions (shown when the -v option is given) now shows, for each
2793 used suppression, the file name and line number where the suppression
2796 * New and modified GDB server monitor features:
2798 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2799 that can be used to execute gdbserver monitor commands from the
2802 - A new monitor command, "v.info open_fds", that gives the list of
2803 open file descriptors and additional details.
2805 - An optional message in the "v.info n_errs_found" monitor command,
2806 for example "v.info n_errs_found test 1234 finished", allowing a
2807 comment string to be added to the process output, perhaps for the
2808 purpose of separating errors of different tests or test phases.
2810 - A new monitor command "v.info execontext" that shows information
2811 about the stack traces recorded by Valgrind.
2813 - A new monitor command "v.do expensive_sanity_check_general" to run
2814 some internal consistency checks.
2816 * New flag --sigill-diagnostics to control whether a diagnostic
2817 message is printed when the JIT encounters an instruction it can't
2818 translate. The actual behavior -- delivery of SIGILL to the
2819 application -- is unchanged.
2821 * The maximum amount of memory that Valgrind can use on 64 bit targets
2822 has been increased from 32GB to 64GB. This should make it possible
2823 to run applications on Memcheck that natively require up to about 35GB.
2825 * ==================== FIXED BUGS ====================
2827 The following bugs have been fixed or resolved. Note that "n-i-bz"
2828 stands for "not in bugzilla" -- that is, a bug that was reported to us
2829 but never got a bugzilla entry. We encourage you to file bugs in
2830 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2831 than mailing the developers (or mailing lists) directly -- bugs that
2832 are not entered into bugzilla tend to get forgotten about or ignored.
2834 To see details of a given bug, visit
2835 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2836 where XXXXXX is the bug number as listed below.
2838 123837 system call: 4th argument is optional, depending on cmd
2839 135425 memcheck should tell you where Freed blocks were Mallocd
2840 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2841 207815 Adds some of the drm ioctls to syswrap-linux.c
2842 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2843 252955 Impossible to compile with ccache
2844 253519 Memcheck reports auxv pointer accesses as invalid reads.
2845 263034 Crash when loading some PPC64 binaries
2846 269599 Increase deepest backtrace
2847 274695 s390x: Support "compare to/from logical" instructions (z196)
2848 275800 s390x: Autodetect cache info (part 2)
2849 280271 Valgrind reports possible memory leaks on still-reachable std::string
2850 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2851 289578 Backtraces with ARM unwind tables (stack scan flags)
2852 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2853 304832 ppc32: build failure
2854 305431 Use find_buildid shdr fallback for separate .debug files
2855 305728 Add support for AVX2 instructions
2856 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2857 306035 s390x: Fix IR generation for LAAG and friends
2858 306054 s390x: Condition code computation for convert-to-int/logical
2859 306098 s390x: alternate opcode form for convert to/from fixed
2860 306587 Fix cache line detection from auxiliary vector for PPC.
2861 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2862 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2863 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2864 307101 sys_capget second argument can be NULL
2865 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2866 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2867 307113 s390x: DFP support
2868 307141 valgrind does't work in mips-linux system
2869 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2870 307285 x86_amd64 feature test for avx in test suite is wrong
2871 307290 memcheck overlap testcase needs memcpy version filter
2872 307463 Please add "&limit=0" to the "all open bugs" link
2873 307465 --show-possibly-lost=no should reduce the error count / exit code
2874 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2875 307729 pkgconfig support broken valgrind.pc
2876 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2877 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2878 308089 Unhandled syscall on ppc64: prctl
2879 308135 PPC32 MPC8xx has 16 bytes cache size
2880 308321 testsuite memcheck filter interferes with gdb_filter
2882 308341 vgdb should report process exit (or fatal signal)
2883 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2884 308495 Remove build dependency on installed Xen headers
2885 308573 Internal error on 64-bit instruction executed in 32-bit mode
2887 308627 pmovmskb validity bit propagation is imprecise
2888 308644 vgdb command for having the info for the track-fds option
2889 308711 give more info about aspacemgr and arenas in out_of_memory
2890 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2891 308718 ARM implement SMLALBB family of instructions
2892 308886 Missing support for PTRACE_SET/GETREGSET
2893 308930 syscall name_to_handle_at (303 on amd64) not handled
2894 309229 V-bit tester does not report number of tests generated
2895 309323 print unrecognized instuction on MIPS
2896 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2897 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2898 309430 Self hosting ppc64 encounters a vassert error on operand type
2899 309600 valgrind is a bit confused about 0-sized sections
2900 309823 Generate errors for still reachable blocks
2901 309921 PCMPISTRI validity bit propagation is imprecise
2902 309922 none/tests/ppc64/test_dfp5 sometimes fails
2903 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2904 310424 --read-var-info does not properly describe static variables
2905 310792 search additional path for debug symbols
2906 310931 s390x: Message-security assist (MSA) instruction extension [..]
2907 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2908 311318 ARM: "128-bit constant is not implemented" error message
2909 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2910 311690 V crashes because it redirects branches inside of a redirected function
2911 311880 x86_64: make regtest hangs at shell_valid1
2912 311922 WARNING: unhandled syscall: 170
2914 312171 ppc: insn selection for DFP
2915 312571 Rounding mode call wrong for the DFP Iops [..]
2916 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2917 312913 Dangling pointers error should also report the alloc stack trace
2918 312980 Building on Mountain Lion generates some compiler warnings
2919 313267 Adding MIPS64/Linux port to Valgrind
2922 313811 Buffer overflow in assert_fail
2923 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2924 314269 ppc: dead code in insn selection
2925 314718 ARM: implement integer divide instruction (sdiv and udiv)
2926 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2927 315441 sendmsg syscall should ignore unset msghdr msg_flags
2928 315534 msgrcv inside a thread causes valgrind to hang (block)
2929 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2930 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2931 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2932 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2933 316144 valgrind.1 manpage contains unknown ??? strings [..]
2934 316145 callgrind command line options in manpage reference (unknown) [..]
2935 316145 callgrind command line options in manpage reference [..]
2936 316181 drd: Fixed a 4x slowdown for certain applications
2937 316503 Valgrind does not support SSE4 "movntdqa" instruction
2938 316535 Use of |signed int| instead of |size_t| in valgrind messages
2939 316696 fluidanimate program of parsec 2.1 stuck
2940 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2941 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2942 317186 "Impossible happens" when occurs VCVT instruction on ARM
2943 317318 Support for Threading Building Blocks "scalable_malloc"
2944 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2945 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2946 317463 bmi testcase IR SANITY CHECK FAILURE
2947 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2948 318050 libmpiwrap fails to compile with out-of-source build
2949 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2950 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2951 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2952 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2953 318932 Add missing PPC64 and PPC32 system call support
2954 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2955 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2956 319494 VEX Makefile-gcc standalone build update after r2702
2957 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2958 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2959 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2960 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2961 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2962 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2963 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2964 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2965 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2966 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2967 320895 add fanotify support (patch included)
2968 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2969 321065 Valgrind updates for Xen 4.3
2970 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2971 321363 Unhandled instruction: SSAX (ARM + Thumb)
2972 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2973 321466 Unhandled instruction: SHASX (ARM + Thumb)
2974 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2975 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2976 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2977 321620 Unhandled instruction: UASX (ARM + Thumb)
2978 321621 Unhandled instruction: USAX (ARM + Thumb)
2979 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2980 321693 Unhandled instruction: LDRSBT (Thumb)
2981 321694 Unhandled instruction: UQASX (ARM + Thumb)
2982 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2983 321697 Unhandled instruction: UHASX (ARM + Thumb)
2984 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2985 321704 Unhandled instruction: REVSH (ARM + Thumb)
2986 321730 Add cg_diff and cg_merge man pages
2987 321738 Add vgdb and valgrind-listener man pages
2989 321891 Unhandled instruction: LDRHT (Thumb)
2990 321960 pthread_create() then alloca() causing invalid stack write errors
2991 321969 ppc32 and ppc64 don't support [lf]setxattr
2992 322254 Show threadname together with tid if set by application
2993 322294 Add initial support for IBM Power ISA 2.07
2994 322368 Assertion failure in wqthread_hijack under OS X 10.8
2995 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2996 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2997 322851 0bXXX binary literal syntax is not standard
2998 323035 Unhandled instruction: LDRSHT(Thumb)
2999 323036 Unhandled instruction: SMMLS (ARM and Thumb)
3000 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
3001 323175 Unhandled instruction: SMLALD (ARM + Thumb)
3002 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
3003 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
3004 323437 Phase 2 support for IBM Power ISA 2.07
3005 323713 Support mmxext (integer sse) subset on i386 (athlon)
3006 323803 Transactional memory instructions are not supported for Power
3007 323893 SSE3 not available on amd cpus in valgrind
3008 323905 Probable false positive from Valgrind/drd on close()
3009 323912 valgrind.h header isn't compatible for mingw64
3010 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
3011 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
3012 324181 mmap does not handle MAP_32BIT
3013 324227 memcheck false positive leak when a thread calls exit+block [..]
3014 324421 Support for fanotify API on ARM architecture
3015 324514 gdbserver monitor cmd output behaviour consistency [..]
3016 324518 ppc64: Emulation of dcbt instructions does not handle [..]
3017 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
3018 324582 When access is made to freed memory, report both allocation [..]
3019 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
3020 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
3021 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
3022 324834 Unhandled instructions in Microsoft C run-time for x86_64
3023 324894 Phase 3 support for IBM Power ISA 2.07
3024 326091 drd: Avoid false race reports from optimized strlen() impls
3025 326113 valgrind libvex hwcaps error on AMD64
3026 n-i-bz Some wrong command line options could be ignored
3027 n-i-bz patch to allow fair-sched on android
3028 n-i-bz report error for vgdb snapshot requested before execution
3029 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
3031 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
3035 Release 3.8.1 (19 September 2012)
3036 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3037 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
3038 that occur moderately frequently in real use cases, adds support for
3039 some missing instructions on ARM, and fixes a deadlock condition on
3040 MacOSX. If you package or deliver 3.8.0 for others to use, you might
3041 want to consider upgrading to 3.8.1 instead.
3043 The following bugs have been fixed or resolved. Note that "n-i-bz"
3044 stands for "not in bugzilla" -- that is, a bug that was reported to us
3045 but never got a bugzilla entry. We encourage you to file bugs in
3046 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3047 than mailing the developers (or mailing lists) directly -- bugs that
3048 are not entered into bugzilla tend to get forgotten about or ignored.
3050 To see details of a given bug, visit
3051 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3052 where XXXXXX is the bug number as listed below.
3055 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
3056 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
3057 298281 wcslen causes false(?) uninitialised value warnings
3058 301281 valgrind hangs on OS X when the process calls system()
3059 304035 disInstr(arm): unhandled instruction 0xE1023053
3060 304867 implement MOVBE instruction in x86 mode
3061 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
3062 305042 amd64: implement 0F 7F encoding of movq between two registers
3063 305199 ARM: implement QDADD and QDSUB
3064 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
3065 305513 killed by fatal signal: SIGSEGV
3066 305690 DRD reporting invalid semaphore when sem_trywait fails
3067 305926 Invalid alignment checks for some AVX instructions
3068 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
3069 306310 3.8.0 release tarball missing some files
3070 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
3071 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
3072 n-i-bz shmat of a segment > 4Gb does not work
3073 n-i-bz simulate_control_c script wrong USR1 signal number on mips
3074 n-i-bz vgdb ptrace calls wrong on mips [...]
3075 n-i-bz Fixes for more MPI false positives
3076 n-i-bz exp-sgcheck's memcpy causes programs to segfault
3077 n-i-bz OSX build w/ clang: asserts at startup
3078 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
3079 n-i-bz fix a couple of union tag-vs-field mixups
3080 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
3082 The following bugs were fixed in 3.8.0 but not listed in this NEWS
3085 254088 Valgrind should know about UD2 instruction
3088 304754 NEWS blows TeX's little mind
3090 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
3094 Release 3.8.0 (10 August 2012)
3095 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3096 3.8.0 is a feature release with many improvements and the usual
3097 collection of bug fixes.
3099 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3100 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
3101 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
3102 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
3103 There is initial support for MacOSX 10.8, but it is not usable for
3104 serious work at present.
3106 * ================== PLATFORM CHANGES =================
3108 * Support for MIPS32 platforms running Linux. Valgrind has been
3109 tested on MIPS32 and MIPS32r2 platforms running different Debian
3110 Squeeze and MeeGo distributions. Both little-endian and big-endian
3111 cores are supported. The tools Memcheck, Massif and Lackey have
3112 been tested and are known to work. See README.mips for more details.
3114 * Preliminary support for Android running on x86.
3116 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
3118 * Support for Intel AVX instructions and for AES instructions. This
3119 support is available only for 64 bit code.
3121 * Support for POWER Decimal Floating Point instructions.
3123 * ==================== TOOL CHANGES ====================
3125 * Non-libc malloc implementations are now supported. This is useful
3126 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
3127 Using the new option --soname-synonyms, such tools can be informed
3128 that the malloc implementation is either linked statically into the
3129 executable, or is present in some other shared library different
3130 from libc.so. This makes it possible to process statically linked
3131 programs, and programs using other malloc libraries, for example
3132 TCMalloc or JEMalloc.
3134 * For tools that provide their own replacement for malloc et al, the
3135 option --redzone-size=<number> allows users to specify the size of
3136 the padding blocks (redzones) added before and after each client
3137 allocated block. Smaller redzones decrease the memory needed by
3138 Valgrind. Bigger redzones increase the chance to detect blocks
3139 overrun or underrun. Prior to this change, the redzone size was
3140 hardwired to 16 bytes in Memcheck.
3144 - The leak_check GDB server monitor command now can
3145 control the maximum nr of loss records to output.
3147 - Reduction of memory use for applications allocating
3148 many blocks and/or having many partially defined bytes.
3150 - Addition of GDB server monitor command 'block_list' that lists
3151 the addresses/sizes of the blocks of a leak search loss record.
3153 - Addition of GDB server monitor command 'who_points_at' that lists
3154 the locations pointing at a block.
3156 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
3157 detect an invalid access of these redzones, by marking them
3158 noaccess. Similarly, if a redzone size is given for a memory
3159 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
3160 This still allows to find some bugs if the user has forgotten to
3161 mark the pool superblock noaccess.
3163 - Performance of memory leak check has been improved, especially in
3164 cases where there are many leaked blocks and/or many suppression
3165 rules used to suppress leak reports.
3167 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
3168 more precise analysis, which is important for LLVM/Clang
3169 generated code. This is at the cost of somewhat reduced
3170 performance. Note there is no change to analysis precision or
3171 costs on Linux targets.
3175 - Added even more facilities that can help finding the cause of a data
3176 race, namely the command-line option --ptrace-addr and the macro
3177 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
3179 - Fixed a subtle bug that could cause false positive data race reports.
3181 * ==================== OTHER CHANGES ====================
3183 * The C++ demangler has been updated so as to work well with C++
3184 compiled by up to at least g++ 4.6.
3186 * Tool developers can make replacement/wrapping more flexible thanks
3187 to the new option --soname-synonyms. This was reported above, but
3188 in fact is very general and applies to all function
3189 replacement/wrapping, not just to malloc-family functions.
3191 * Round-robin scheduling of threads can be selected, using the new
3192 option --fair-sched= yes. Prior to this change, the pipe-based
3193 thread serialisation mechanism (which is still the default) could
3194 give very unfair scheduling. --fair-sched=yes improves
3195 responsiveness of interactive multithreaded applications, and
3196 improves repeatability of results from the thread checkers Helgrind
3199 * For tool developers: support to run Valgrind on Valgrind has been
3200 improved. We can now routinely Valgrind on Helgrind or Memcheck.
3202 * gdbserver now shows the float shadow registers as integer
3203 rather than float values, as the shadow values are mostly
3204 used as bit patterns.
3206 * Increased limit for the --num-callers command line flag to 500.
3208 * Performance improvements for error matching when there are many
3209 suppression records in use.
3211 * Improved support for DWARF4 debugging information (bug 284184).
3213 * Initial support for DWZ compressed Dwarf debug info.
3215 * Improved control over the IR optimiser's handling of the tradeoff
3216 between performance and precision of exceptions. Specifically,
3217 --vex-iropt-precise-memory-exns has been removed and replaced by
3218 --vex-iropt-register-updates, with extended functionality. This
3219 allows the Valgrind gdbserver to always show up to date register
3222 * Modest performance gains through the use of translation chaining for
3225 * ==================== FIXED BUGS ====================
3227 The following bugs have been fixed or resolved. Note that "n-i-bz"
3228 stands for "not in bugzilla" -- that is, a bug that was reported to us
3229 but never got a bugzilla entry. We encourage you to file bugs in
3230 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3231 than mailing the developers (or mailing lists) directly -- bugs that
3232 are not entered into bugzilla tend to get forgotten about or ignored.
3234 To see details of a given bug, visit
3235 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3236 where XXXXXX is the bug number as listed below.
3238 197914 Building valgrind from svn now requires automake-1.10
3239 203877 increase to 16Mb maximum allowed alignment for memalign et al
3240 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
3241 247386 make perf does not run all performance tests
3242 270006 Valgrind scheduler unfair
3243 270777 Adding MIPS/Linux port to Valgrind
3244 270796 s390x: Removed broken support for the TS insn
3245 271438 Fix configure for proper SSE4.2 detection
3246 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
3247 273475 Add support for AVX instructions
3248 274078 improved configure logic for mpicc
3249 276993 fix mremap 'no thrash checks'
3250 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
3251 281482 memcheck incorrect byte allocation count in realloc() for silly argument
3252 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
3253 283413 Fix wrong sanity check
3254 283671 Robustize alignment computation in LibVEX_Alloc
3255 283961 Adding support for some HCI IOCTLs
3256 284124 parse_type_DIE: confused by: DWARF 4
3257 284864 == 273475 (Add support for AVX instructions)
3258 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
3259 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
3260 285725 == 273475 (Add support for AVX instructions)
3261 286261 add wrapper for linux I2C_RDWR ioctl
3262 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
3263 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
3264 286384 configure fails "checking for a supported version of gcc"
3265 286497 == 273475 (Add support for AVX instructions)
3266 286596 == 273475 (Add support for AVX instructions)
3267 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
3268 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
3269 287260 Incorrect conditional jump or move depends on uninitialised value(s)
3270 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
3271 287307 == 273475 (Add support for AVX instructions)
3272 287858 VG_(strerror): unknown error
3273 288298 (MacOSX) unhandled syscall shm_unlink
3274 288995 == 273475 (Add support for AVX instructions)
3275 289470 Loading of large Mach-O thin binaries fails.
3276 289656 == 273475 (Add support for AVX instructions)
3277 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
3278 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
3279 289839 s390x: Provide support for unicode conversion instructions
3280 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
3281 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
3282 290655 Add support for AESKEYGENASSIST instruction
3283 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
3284 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
3285 291253 ES register not initialised in valgrind simulation
3286 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
3287 291865 s390x: Support the "Compare Double and Swap" family of instructions
3288 292300 == 273475 (Add support for AVX instructions)
3289 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
3290 292493 == 273475 (Add support for AVX instructions)
3291 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
3292 292627 Missing support for some SCSI ioctls
3293 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
3294 292841 == 273475 (Add support for AVX instructions)
3295 292993 implement the getcpu syscall on amd64-linux
3296 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
3297 293088 Add some VEX sanity checks for ppc64 unhandled instructions
3298 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
3299 293754 PCMPxSTRx not implemented for 16-bit characters
3300 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
3301 293808 CLFLUSH not supported by latest VEX for amd64
3302 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
3303 294048 MPSADBW instruction not implemented
3304 294055 regtest none/tests/shell fails when locale is not set to C
3305 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
3306 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
3307 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
3308 294260 disInstr_AMD64: disInstr miscalculated next %rip
3309 294523 --partial-loads-ok=yes causes false negatives
3310 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
3311 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
3312 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
3313 295089 can not annotate source for both helgrind and drd
3314 295221 POWER Processor decimal floating point instruction support missing
3315 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
3316 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
3317 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
3318 295617 ARM - Add some missing syscalls
3319 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
3320 296229 Linux user input device ioctls missing wrappers
3321 296318 ELF Debug info improvements (more than one rx/rw mapping)
3322 296422 Add translation chaining support
3323 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
3324 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
3325 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
3326 297078 gdbserver signal handling problems [..]
3327 297147 drd false positives on newly allocated memory
3328 297329 disallow decoding of IBM Power DFP insns on some machines
3329 297497 POWER Processor decimal floating point instruction support missing
3330 297701 Another alias for strncasecmp_l in libc-2.13.so
3331 297911 'invalid write' not reported when using APIs for custom mem allocators.
3332 297976 s390x: revisit EX implementation
3333 297991 Valgrind interferes with mmap()+ftell()
3334 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
3335 297993 Fix compilation of valgrind with gcc -g3.
3336 298080 POWER Processor DFP support missing, part 3
3337 298227 == 273475 (Add support for AVX instructions)
3338 298335 == 273475 (Add support for AVX instructions)
3339 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
3340 298394 s390x: Don't bail out on an unknown machine model. [..]
3341 298421 accept4() syscall (366) support is missing for ARM
3342 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
3343 298732 valgrind installation problem in ubuntu with kernel version 3.x
3344 298862 POWER Processor DFP instruction support missing, part 4
3345 298864 DWARF reader mis-parses DW_FORM_ref_addr
3346 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
3347 299053 Support DWARF4 DW_AT_high_pc constant form
3348 299104 == 273475 (Add support for AVX instructions)
3349 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
3350 299629 dup3() syscall (358) support is missing for ARM
3351 299694 POWER Processor DFP instruction support missing, part 5
3352 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
3353 299803 == 273475 (Add support for AVX instructions)
3354 299804 == 273475 (Add support for AVX instructions)
3355 299805 == 273475 (Add support for AVX instructions)
3356 300140 ARM - Missing (T1) SMMUL
3357 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
3358 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
3359 300414 FCOM and FCOMP unimplemented for amd64 guest
3360 301204 infinite loop in canonicaliseSymtab with ifunc symbol
3361 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
3362 301265 add x86 support to Android build
3363 301984 configure script doesn't detect certain versions of clang
3364 302205 Fix compiler warnings for POWER VEX code and POWER test cases
3365 302287 Unhandled movbe instruction on Atom processors
3366 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
3367 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
3368 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
3369 302656 == 273475 (Add support for AVX instructions)
3370 302709 valgrind for ARM needs extra tls support for android emulator [..]
3371 302827 add wrapper for CDROM_GET_CAPABILITY
3372 302901 Valgrind crashes with dwz optimized debuginfo
3373 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
3374 303116 Add support for the POWER instruction popcntb
3375 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
3376 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
3377 303466 == 273475 (Add support for AVX instructions)
3378 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
3379 303963 strstr() function produces wrong results under valgrind callgrind
3380 304054 CALL_FN_xx macros need to enforce stack alignment
3381 304561 tee system call not supported
3382 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
3383 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
3384 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
3385 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
3386 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
3387 n-i-bz s390x: Shadow registers can now be examined using vgdb
3389 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
3390 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
3394 Release 3.7.0 (5 November 2011)
3395 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3396 3.7.0 is a feature release with many significant improvements and the
3397 usual collection of bug fixes.
3399 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3400 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
3401 Support for recent distros and toolchain components (glibc 2.14, gcc
3402 4.6, MacOSX 10.7) has been added.
3404 * ================== PLATFORM CHANGES =================
3406 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
3407 analyse 64-bit programs running on z/Architecture. Most user space
3408 instructions up to and including z10 are supported. Valgrind has
3409 been tested extensively on z9, z10, and z196 machines running SLES
3410 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
3411 known to work well. Callgrind, Helgrind, and DRD work reasonably
3412 well on z9 and later models. See README.s390 for more details.
3414 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
3415 64-bit processes are supported. Some complex threaded applications
3416 (Firefox) are observed to hang when run as 32 bit applications,
3417 whereas 64-bit versions run OK. The cause is unknown. Memcheck
3418 will likely report some false errors. In general, expect some rough
3419 spots. This release also supports MacOSX 10.6, but drops support
3422 * Preliminary support for Android (on ARM). Valgrind can now run
3423 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
3424 README.android for more details, plus instructions on how to get
3427 * Support for the IBM Power ISA 2.06 (Power7 instructions)
3429 * General correctness and performance improvements for ARM/Linux, and,
3430 by extension, ARM/Android.
3432 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
3433 instruction set support is under development but is not available in
3436 * Support for AIX5 has been removed.
3438 * ==================== TOOL CHANGES ====================
3440 * Memcheck: some incremental changes:
3442 - reduction of memory use in some circumstances
3444 - improved handling of freed memory, which in some circumstances
3445 can cause detection of use-after-free that would previously have
3448 - fix of a longstanding bug that could cause false negatives (missed
3449 errors) in programs doing vector saturated narrowing instructions.
3451 * Helgrind: performance improvements and major memory use reductions,
3452 particularly for large, long running applications which perform many
3453 synchronisation (lock, unlock, etc) events. Plus many smaller
3456 - display of locksets for both threads involved in a race
3458 - general improvements in formatting/clarity of error messages
3460 - addition of facilities and documentation regarding annotation
3461 of thread safe reference counted C++ classes
3463 - new flag --check-stack-refs=no|yes [yes], to disable race checking
3464 on thread stacks (a performance hack)
3466 - new flag --free-is-write=no|yes [no], to enable detection of races
3467 where one thread accesses heap memory but another one frees it,
3468 without any coordinating synchronisation event
3470 * DRD: enabled XML output; added support for delayed thread deletion
3471 in order to detect races that occur close to the end of a thread
3472 (--join-list-vol); fixed a memory leak triggered by repeated client
3473 memory allocatation and deallocation; improved Darwin support.
3475 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
3477 * exp-sgcheck: this tool has been reduced in scope so as to improve
3478 performance and remove checking that Memcheck does better.
3479 Specifically, the ability to check for overruns for stack and global
3480 arrays is unchanged, but the ability to check for overruns of heap
3481 blocks has been removed. The tool has accordingly been renamed to
3482 exp-sgcheck ("Stack and Global Array Checking").
3484 * ==================== OTHER CHANGES ====================
3486 * GDB server: Valgrind now has an embedded GDB server. That means it
3487 is possible to control a Valgrind run from GDB, doing all the usual
3488 things that GDB can do (single stepping, breakpoints, examining
3489 data, etc). Tool-specific functionality is also available. For
3490 example, it is possible to query the definedness state of variables
3491 or memory from within GDB when running Memcheck; arbitrarily large
3492 memory watchpoints are supported, etc. To use the GDB server, start
3493 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3496 * Improved support for unfriendly self-modifying code: a new option
3497 --smc-check=all-non-file is available. This adds the relevant
3498 consistency checks only to code that originates in non-file-backed
3499 mappings. In effect this confines the consistency checking only to
3500 code that is or might be JIT generated, and avoids checks on code
3501 that must have been compiled ahead of time. This significantly
3502 improves performance on applications that generate code at run time.
3504 * It is now possible to build a working Valgrind using Clang-2.9 on
3507 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3508 These enable and disable error reporting on a per-thread, and
3509 nestable, basis. This is useful for hiding errors in particularly
3510 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3511 now uses this facility.
3513 * Added the --mod-funcname option to cg_diff.
3515 * ==================== FIXED BUGS ====================
3517 The following bugs have been fixed or resolved. Note that "n-i-bz"
3518 stands for "not in bugzilla" -- that is, a bug that was reported to us
3519 but never got a bugzilla entry. We encourage you to file bugs in
3520 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3521 mailing the developers (or mailing lists) directly -- bugs that are
3522 not entered into bugzilla tend to get forgotten about or ignored.
3524 To see details of a given bug, visit
3525 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3526 where XXXXXX is the bug number as listed below.
3528 79311 malloc silly arg warning does not give stack trace
3529 210935 port valgrind.h (not valgrind) to win32 to support client requests
3530 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3531 243404 Port to zSeries
3532 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3533 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3534 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3535 253206 Some fixes for the faultstatus testcase
3536 255223 capget testcase fails when running as root
3537 256703 xlc_dbl_u32.c testcase broken
3538 256726 Helgrind tests have broken inline asm
3539 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3540 264800 testcase compile failure on zseries
3541 265762 make public VEX headers compilable by G++ 3.x
3542 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3543 266753 configure script does not give the user the option to not use QtCore
3544 266931 gen_insn_test.pl is broken
3545 266961 ld-linux.so.2 i?86-linux strlen issues
3546 266990 setns instruction causes false positive
3547 267020 Make directory for temporary files configurable at run-time.
3548 267342 == 267997 (segmentation fault on Mac OS 10.6)
3549 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3550 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3551 267488 regtest: darwin support for 64-bit build
3552 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3553 267630 Add support for IBM Power ISA 2.06 -- stage 1
3554 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3555 267819 Add client request for informing the core about reallocation
3556 267925 laog data structure quadratic for a single sequence of lock
3557 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3558 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3559 268513 missed optimizations in fold_Expr
3560 268619 s390x: fpr - gpr transfer facility
3561 268620 s390x: reconsider "long displacement" requirement
3562 268621 s390x: improve IR generation for XC
3563 268715 s390x: FLOGR is not universally available
3564 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3565 268930 s390x: MHY is not universally available
3566 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3567 269079 Support ptrace system call on ARM
3568 269144 missing "Bad option" error message
3569 269209 conditional load and store facility (z196)
3570 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3571 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3572 269736 s390x: minor code generation tweaks
3573 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3574 269863 s390x: remove unused function parameters
3575 269864 s390x: tweak s390_emit_load_cc
3576 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3577 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3578 270115 s390x: rewrite some testcases
3579 270309 == 267997 (valgrind crash on startup)
3580 270320 add support for Linux FIOQSIZE ioctl() call
3581 270326 segfault while trying to sanitize the environment passed to execle
3582 270794 IBM POWER7 support patch causes regression in none/tests
3583 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3584 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3585 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3586 270959 s390x: invalid use of R0 as base register
3587 271042 VSX configure check fails when it should not
3588 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3589 271259 s390x: fix code confusion
3590 271337 == 267997 (Valgrind segfaults on MacOS X)
3591 271385 s390x: Implement Ist_MBE
3592 271501 s390x: misc cleanups
3593 271504 s390x: promote likely and unlikely
3594 271579 ppc: using wrong enum type
3595 271615 unhandled instruction "popcnt" (arch=amd10h)
3596 271730 Fix bug when checking ioctls: duplicate check
3597 271776 s390x: provide STFLE instruction support
3598 271779 s390x: provide clock instructions like STCK
3599 271799 Darwin: ioctls without an arg report a memory error
3600 271820 arm: fix type confusion
3601 271917 pthread_cond_timedwait failure leads to not-locked false positive
3602 272067 s390x: fix DISP20 macro
3603 272615 A typo in debug output in mc_leakcheck.c
3604 272661 callgrind_annotate chokes when run from paths containing regex chars
3605 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3606 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3607 272967 make documentation build-system more robust
3608 272986 Fix gcc-4.6 warnings with valgrind.h
3609 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3610 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3611 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3612 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3613 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3614 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3615 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3616 273778 exp-ptrcheck: unhandled sysno == 259
3617 274089 exp-ptrcheck: unhandled sysno == 208
3618 274378 s390x: Various dispatcher tweaks
3619 274447 WARNING: unhandled syscall: 340
3620 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3621 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3622 274926 valgrind does not build against linux-3
3623 275148 configure FAIL with glibc-2.14
3624 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3625 275168 Make Valgrind work for MacOSX 10.7 Lion
3626 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3627 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3628 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3629 275308 Fix implementation for ppc64 fres instruc
3630 275339 s390x: fix testcase compile warnings
3631 275517 s390x: Provide support for CKSM instruction
3632 275710 s390x: get rid of redundant address mode calculation
3633 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3634 275852 == 250101 (valgrind uses all swap space and is killed)
3635 276784 Add support for IBM Power ISA 2.06 -- stage 3
3636 276987 gdbsrv: fix tests following recent commits
3637 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3638 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3639 277471 Unhandled syscall: 340
3640 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3641 277653 ARM: support Thumb2 PLD instruction
3642 277663 ARM: NEON float VMUL by scalar incorrect
3643 277689 ARM: tests for VSTn with register post-index are broken
3644 277694 ARM: BLX LR instruction broken in ARM mode
3645 277780 ARM: VMOV.F32 (immediate) instruction is broken
3646 278057 fuse filesystem syscall deadlocks
3647 278078 Unimplemented syscall 280 on ppc32
3648 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3649 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3650 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3651 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3652 279027 Support for MVCL and CLCL instruction
3653 279027 s390x: Provide support for CLCL and MVCL instructions
3654 279062 Remove a redundant check in the insn selector for ppc.
3655 279071 JDK creates PTEST with redundant REX.W prefix
3656 279212 gdbsrv: add monitor cmd v.info scheduler.
3657 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3658 279698 memcheck discards valid-bits for packuswb
3659 279795 memcheck reports uninitialised values for mincore on amd64
3660 279994 Add support for IBM Power ISA 2.06 -- stage 3
3661 280083 mempolicy syscall check errors
3662 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3663 280710 s390x: config files for nightly builds
3664 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3665 280965 Valgrind breaks fcntl locks when program does mmap
3666 281138 WARNING: unhandled syscall: 340
3667 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3668 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3669 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3670 281468 s390x: handle do_clone and gcc clones in call traces
3671 281488 ARM: VFP register corruption
3672 281828 == 275284 (false memmove warning: "Source and destination overlap")
3673 281883 s390x: Fix system call wrapper for "clone".
3674 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3675 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3676 282238 SLES10: make check fails
3677 282979 strcasestr needs replacement with recent(>=2.12) glibc
3678 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3679 283243 Regression in ppc64 memcheck tests
3680 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3681 283427 re-connect epoll_pwait syscall on ARM linux
3682 283600 gdbsrv: android: port vgdb.c
3683 283709 none/tests/faultstatus needs to account for page size
3684 284305 filter_gdb needs enhancement to work on ppc64
3685 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3686 284472 Thumb2 ROR.W encoding T2 not implemented
3687 284621 XML-escape process command line in XML output
3688 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3689 that have non-power-of-2 sizes (also AMDs)
3690 n-i-bz don't be spooked by libraries mashed by elfhack
3691 n-i-bz don't be spooked by libxul.so linked with gold
3692 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3694 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3695 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3696 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3700 Release 3.6.1 (16 February 2011)
3701 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3702 3.6.1 is a bug fix release. It adds support for some SSE4
3703 instructions that were omitted in 3.6.0 due to lack of time. Initial
3704 support for glibc-2.13 has been added. A number of bugs causing
3705 crashing or assertion failures have been fixed.
3707 The following bugs have been fixed or resolved. Note that "n-i-bz"
3708 stands for "not in bugzilla" -- that is, a bug that was reported to us
3709 but never got a bugzilla entry. We encourage you to file bugs in
3710 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3711 mailing the developers (or mailing lists) directly -- bugs that are
3712 not entered into bugzilla tend to get forgotten about or ignored.
3714 To see details of a given bug, visit
3715 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3716 where XXXXXX is the bug number as listed below.
3718 188572 Valgrind on Mac should suppress setenv() mem leak
3719 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3720 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3721 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3722 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3723 254420 memory pool tracking broken
3724 254957 Test code failing to compile due to changes in memcheck.h
3725 255009 helgrind/drd: crash on chmod with invalid parameter
3726 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3727 255355 helgrind/drd: crash on threaded programs doing fork
3729 255418 (SSE4.x) rint call compiled with ICC
3730 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3731 255888 closing valgrindoutput tag outputted to log-stream on error
3732 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3733 255966 Slowness when using mempool annotations
3734 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3735 256600 super-optimized strcasecmp() false positive
3736 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3737 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3738 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3739 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3740 257276 Missing case in memcheck --track-origins=yes
3741 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3742 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3743 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3744 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3745 263099 callgrind_annotate counts Ir improperly [...]
3746 263877 undefined coprocessor instruction on ARMv7
3747 265964 configure FAIL with glibc-2.13
3748 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3749 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3750 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3751 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3752 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3754 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3758 Release 3.6.0 (21 October 2010)
3759 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3760 3.6.0 is a feature release with many significant improvements and the
3761 usual collection of bug fixes.
3763 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3764 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3765 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3767 -------------------------
3769 Here are some highlights. Details are shown further down:
3771 * Support for ARM/Linux.
3773 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3775 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3777 * Support for the SSE4.2 instruction set.
3779 * Enhancements to the Callgrind profiler, including the ability to
3780 handle CPUs with three levels of cache.
3782 * A new experimental heap profiler, DHAT.
3784 * A huge number of bug fixes and small enhancements.
3786 -------------------------
3788 Here are details of the above changes, together with descriptions of
3789 many other changes, and a list of fixed bugs.
3791 * ================== PLATFORM CHANGES =================
3793 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3794 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3795 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3797 This requires a CPU capable of running the ARMv7-A instruction set
3798 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3799 of the user space instruction set, including ARM and Thumb integer
3800 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3801 Cachegrind and Massif tools work properly; other tools work to
3804 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3805 with support for recent releases of the underlying toolchain
3806 components, notably gcc-4.5 and glibc-2.12.
3808 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3809 support also works much better on OS X 10.5, and is as solid as
3812 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3813 64-bit mode. In 32-bit mode, support is only available up to and
3814 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3815 supported in 64-bit mode, and 32-bit mode does in fact support the
3816 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3817 10.6 on 32-bit targets.
3819 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3820 and including version 2.05 is supported.
3822 * ==================== TOOL CHANGES ====================
3824 * Cachegrind has a new processing script, cg_diff, which finds the
3825 difference between two profiles. It's very useful for evaluating
3826 the performance effects of a change in a program.
3828 Related to this change, the meaning of cg_annotate's (rarely-used)
3829 --threshold option has changed; this is unlikely to affect many
3830 people, if you do use it please see the user manual for details.
3832 * Callgrind now can do branch prediction simulation, similar to
3833 Cachegrind. In addition, it optionally can count the number of
3834 executed global bus events. Both can be used for a better
3835 approximation of a "Cycle Estimation" as derived event (you need to
3836 update the event formula in KCachegrind yourself).
3838 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3839 rather than the L2 cache. This is to accommodate machines with
3840 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3841 cache configuration of such a machine it will run the simulation as
3842 if the L2 cache isn't present. This means the results are less
3843 likely to match the true result for the machine, but
3844 Cachegrind/Callgrind's results are already only approximate, and
3845 should not be considered authoritative. The results are still
3846 useful for giving a general idea about a program's locality.
3848 * Massif has a new option, --pages-as-heap, which is disabled by
3849 default. When enabled, instead of tracking allocations at the level
3850 of heap blocks (as allocated with malloc/new/new[]), it instead
3851 tracks memory allocations at the level of memory pages (as mapped by
3852 mmap, brk, etc). Each mapped page is treated as its own block.
3853 Interpreting the page-level output is harder than the heap-level
3854 output, but this option is useful if you want to account for every
3855 byte of memory used by a program.
3857 * DRD has two new command-line options: --free-is-write and
3858 --trace-alloc. The former allows to detect reading from already freed
3859 memory, and the latter allows tracing of all memory allocations and
3862 * DRD has several new annotations. Custom barrier implementations can
3863 now be annotated, as well as benign races on static variables.
3865 * DRD's happens before / happens after annotations have been made more
3866 powerful, so that they can now also be used to annotate e.g. a smart
3867 pointer implementation.
3869 * Helgrind's annotation set has also been drastically improved, so as
3870 to provide to users a general set of annotations to describe locks,
3871 semaphores, barriers and condition variables. Annotations to
3872 describe thread-safe reference counted heap objects have also been
3875 * Memcheck has a new command-line option, --show-possibly-lost, which
3876 is enabled by default. When disabled, the leak detector will not
3877 show possibly-lost blocks.
3879 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3880 has been added. DHAT keeps track of allocated heap blocks, and also
3881 inspects every memory reference to see which block (if any) is being
3882 accessed. This gives a lot of insight into block lifetimes,
3883 utilisation, turnover, liveness, and the location of hot and cold
3884 fields. You can use DHAT to do hot-field profiling.
3886 * ==================== OTHER CHANGES ====================
3888 * Improved support for unfriendly self-modifying code: the extra
3889 overhead incurred by --smc-check=all has been reduced by
3890 approximately a factor of 5 as compared with 3.5.0.
3892 * Ability to show directory names for source files in error messages.
3893 This is combined with a flexible mechanism for specifying which
3894 parts of the paths should be shown. This is enabled by the new flag
3897 * A new flag, --require-text-symbol, which will stop the run if a
3898 specified symbol is not found it a given shared object when it is
3899 loaded into the process. This makes advanced working with function
3900 intercepting and wrapping safer and more reliable.
3902 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3903 and control of Valgrind is now available for the tools Memcheck and
3904 Helgrind. XML output from Valgrind is available for Memcheck,
3905 Helgrind and exp-Ptrcheck.
3907 * More reliable stack unwinding on amd64-linux, particularly in the
3908 presence of function wrappers, and with gcc-4.5 compiled code.
3910 * Modest scalability (performance improvements) for massive
3911 long-running applications, particularly for those with huge amounts
3914 * Support for analyzing programs running under Wine with has been
3915 improved. The header files <valgrind/valgrind.h>,
3916 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3917 Windows-programs compiled with MinGW or one of the Microsoft Visual
3920 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3921 The 32-bit simulator was not affected. This did not occur often,
3922 but when it did would usually crash the program under test.
3925 * A large number of bugs were fixed. These are shown below.
3927 * A number of bugs were investigated, and were candidates for fixing,
3928 but are not fixed in 3.6.0, due to lack of developer time. They may
3929 get fixed in later releases. They are:
3931 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3932 212419 false positive "lock order violated" (A+B vs A)
3933 213685 Undefined value propagates past dependency breaking instruction
3934 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3935 237920 valgrind segfault on fork failure
3936 242137 support for code compiled by LLVM-2.8
3937 242423 Another unknown Intel cache config value
3938 243232 Inconsistent Lock Orderings report with trylock
3939 243483 ppc: callgrind triggers VEX assertion failure
3940 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3941 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3943 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3944 249435 Analyzing wine programs with callgrind triggers a crash
3945 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3946 250065 Handling large allocations
3947 250101 huge "free" memory usage due to m_mallocfree.c
3948 "superblocks fragmentation"
3949 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3950 252091 Callgrind on ARM does not detect function returns correctly
3951 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3952 254420 memory pool tracking broken
3953 n-i-bz support for adding symbols for JIT generated code
3956 The following bugs have been fixed or resolved. Note that "n-i-bz"
3957 stands for "not in bugzilla" -- that is, a bug that was reported to us
3958 but never got a bugzilla entry. We encourage you to file bugs in
3959 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3960 mailing the developers (or mailing lists) directly -- bugs that are
3961 not entered into bugzilla tend to get forgotten about or ignored.
3963 To see details of a given bug, visit
3964 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3965 where XXXXXX is the bug number as listed below.
3967 135264 dcbzl instruction missing
3969 153699 Valgrind should report unaligned reads with movdqa
3971 190429 Valgrind reports lost of errors in ld.so
3972 with x86_64 2.9.90 glibc
3973 197266 valgrind appears to choke on the xmms instruction
3975 197988 Crash when demangling very large symbol names
3976 202315 unhandled syscall: 332 (inotify_init1)
3977 203256 Add page-level profiling to Massif
3978 205093 dsymutil=yes needs quotes, locking (partial fix)
3979 205241 Snow Leopard 10.6 support (partial fix)
3980 206600 Leak checker fails to upgrade indirect blocks when their
3981 parent becomes reachable
3982 210935 port valgrind.h (not valgrind) to win32 so apps run under
3983 wine can make client requests
3984 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3985 within Linux ip-stack checksum functions
3986 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3988 213685 Undefined value propagates past dependency breaking instruction
3990 215914 Valgrind inserts bogus empty environment variable
3992 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3993 222545 shmat fails under valgind on some arm targets
3994 222560 ARM NEON support
3997 232509 Docs build fails with formatting inside <title></title> elements
3999 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
4000 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
4001 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
4002 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
4003 237485 symlink (syscall 57) is not supported on Mac OS
4004 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
4006 238208 is_just_below_ESP doesn't take into account red-zone
4007 238345 valgrind passes wrong $0 when executing a shell script
4008 238679 mq_timedreceive syscall doesn't flag the reception buffer
4010 238696 fcntl command F_DUPFD_CLOEXEC not supported
4011 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4012 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4013 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
4015 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
4021 242606 unhandled syscall: setegid (in Ptrcheck)
4022 242814 Helgrind "Impossible has happened" during
4023 QApplication::initInstance();
4024 243064 Valgrind attempting to read debug information from iso
4025 243270 Make stack unwinding in Valgrind wrappers more reliable
4026 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
4027 sysno = 277 (mq_open)
4028 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
4029 244493 ARM VFP d16-d31 registers support
4030 244670 add support for audit_session_self syscall on Mac OS 10.6
4031 244921 The xml report of helgrind tool is not well format
4032 244923 In the xml report file, the <preamble> not escape the
4033 xml char, eg '<','&','>'
4034 245535 print full path names in plain text reports
4035 245925 x86-64 red zone handling problem
4036 246258 Valgrind not catching integer underruns + new [] s
4037 246311 reg/reg cmpxchg doesn't work on amd64
4038 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
4039 246888 Improve Makefile.vex.am
4040 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
4041 to [f]chmod_extended
4042 247526 IBM POWER6 (ISA 2.05) support is incomplete
4043 247561 Some leak testcases fails due to reachable addresses in
4045 247875 sizeofIRType to handle Ity_I128
4046 247894 [PATCH] unhandled syscall sys_readahead
4047 247980 Doesn't honor CFLAGS passed to configure
4048 248373 darwin10.supp is empty in the trunk
4049 248822 Linux FIBMAP ioctl has int parameter instead of long
4050 248893 [PATCH] make readdwarf.c big endianess safe to enable
4051 unwinding on big endian systems
4052 249224 Syscall 336 not supported (SYS_proc_info)
4054 249775 Incorrect scheme for detecting NEON capabilities of host CPU
4055 249943 jni JVM init fails when using valgrind
4056 249991 Valgrind incorrectly declares AESKEYGENASSIST support
4058 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
4059 250799 frexp$fenv_access_off function generates SIGILL
4060 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
4061 251251 support pclmulqdq insn
4062 251362 valgrind: ARM: attach to debugger either fails or provokes
4064 251674 Unhandled syscall 294
4067 254257 Add support for debugfiles found by build-id
4068 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
4069 254646 Wrapped functions cause stack misalignment on OS X
4070 (and possibly Linux)
4071 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
4073 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
4077 Release 3.5.0 (19 August 2009)
4078 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4079 3.5.0 is a feature release with many significant improvements and the
4080 usual collection of bug fixes. The main improvement is that Valgrind
4081 now works on Mac OS X.
4083 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
4084 and X86/Darwin. Support for recent distros and toolchain components
4085 (glibc 2.10, gcc 4.5) has been added.
4087 -------------------------
4089 Here is a short summary of the changes. Details are shown further
4092 * Support for Mac OS X (10.5.x).
4094 * Improvements and simplifications to Memcheck's leak checker.
4096 * Clarification and simplifications in various aspects of Valgrind's
4099 * XML output for Helgrind and Ptrcheck.
4101 * Performance and stability improvements for Helgrind and DRD.
4103 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
4105 * A new experimental tool, BBV, useful for computer architecture
4108 * Improved Wine support, including ability to read Windows PDB
4111 -------------------------
4113 Here are details of the above changes, followed by descriptions of
4114 many other minor changes, and a list of fixed bugs.
4117 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
4118 called "Darwin" because that is the name of the OS core, which is the
4119 level that Valgrind works at.)
4123 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
4124 because it would require work and 10.4 is only becoming less common.
4126 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
4127 fairly well. For 10.5.x, 32-bit programs are the default even on
4128 64-bit machines, so it handles most current programs.
4130 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
4131 officially supported, but simple programs at least will probably work.
4132 However, start-up is slow.
4134 - PowerPC machines are not supported.
4136 Things that don't work:
4138 - The Ptrcheck tool.
4140 - Objective-C garbage collection.
4144 - If you have Rogue Amoeba's "Instant Hijack" program installed,
4145 Valgrind will fail with a SIGTRAP at start-up. See
4146 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
4151 - You will likely find --dsymutil=yes a useful option, as error
4152 messages may be imprecise without it.
4154 - Mac OS X support is new and therefore will be less robust than the
4155 Linux support. Please report any bugs you find.
4157 - Threaded programs may run more slowly than on Linux.
4159 Many thanks to Greg Parker for developing this port over several years.
4162 * Memcheck's leak checker has been improved.
4164 - The results for --leak-check=summary now match the summary results
4165 for --leak-check=full. Previously they could differ because
4166 --leak-check=summary counted "indirectly lost" blocks and
4167 "suppressed" blocks as "definitely lost".
4169 - Blocks that are only reachable via at least one interior-pointer,
4170 but are directly pointed to by a start-pointer, were previously
4171 marked as "still reachable". They are now correctly marked as
4174 - The default value for the --leak-resolution option has been
4175 changed from "low" to "high". In general, this means that more
4176 leak reports will be produced, but each leak report will describe
4177 fewer leaked blocks.
4179 - With --leak-check=full, "definitely lost" and "possibly lost"
4180 leaks are now considered as proper errors, ie. they are counted
4181 for the "ERROR SUMMARY" and affect the behaviour of
4182 --error-exitcode. These leaks are not counted as errors if
4183 --leak-check=summary is specified, however.
4185 - Documentation for the leak checker has been improved.
4188 * Various aspects of Valgrind's text output have changed.
4190 - Valgrind's start-up message has changed. It is shorter but also
4191 includes the command being run, which makes it easier to use
4192 --trace-children=yes. An example:
4194 - Valgrind's shut-down messages have also changed. This is most
4195 noticeable with Memcheck, where the leak summary now occurs before
4196 the error summary. This change was necessary to allow leaks to be
4197 counted as proper errors (see the description of the leak checker
4198 changes above for more details). This was also necessary to fix a
4199 longstanding bug in which uses of suppressions against leaks were
4200 not "counted", leading to difficulties in maintaining suppression
4201 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
4203 - Behavior of -v has changed. In previous versions, -v printed out
4204 a mixture of marginally-user-useful information, and tool/core
4205 statistics. The statistics printing has now been moved to its own
4206 flag, --stats=yes. This means -v is less verbose and more likely
4207 to convey useful end-user information.
4209 - The format of some (non-XML) stack trace entries has changed a
4210 little. Previously there were six possible forms:
4212 0x80483BF: really (a.c:20)
4213 0x80483BF: really (in /foo/a.out)
4215 0x80483BF: (within /foo/a.out)
4216 0x80483BF: ??? (a.c:20)
4219 The third and fourth of these forms have been made more consistent
4220 with the others. The six possible forms are now:
4222 0x80483BF: really (a.c:20)
4223 0x80483BF: really (in /foo/a.out)
4224 0x80483BF: really (in ???)
4225 0x80483BF: ??? (in /foo/a.out)
4226 0x80483BF: ??? (a.c:20)
4229 Stack traces produced when --xml=yes is specified are different
4233 * Helgrind and Ptrcheck now support XML output, so they can be used
4234 from GUI tools. Also, the XML output mechanism has been
4237 - The XML format has been overhauled and generalised, so it is more
4238 suitable for error reporting tools in general. The Memcheck
4239 specific aspects of it have been removed. The new format, which
4240 is an evolution of the old format, is described in
4241 docs/internals/xml-output-protocol4.txt.
4243 - Memcheck has been updated to use the new format.
4245 - Helgrind and Ptrcheck are now able to emit output in this format.
4247 - The XML output mechanism has been overhauled. XML is now output
4248 to its own file descriptor, which means that:
4250 * Valgrind can output text and XML independently.
4252 * The longstanding problem of XML output being corrupted by
4253 unexpected un-tagged text messages is solved.
4255 As before, the destination for text output is specified using
4256 --log-file=, --log-fd= or --log-socket=.
4258 As before, XML output for a tool is enabled using --xml=yes.
4260 Because there's a new XML output channel, the XML output
4261 destination is now specified by --xml-file=, --xml-fd= or
4264 Initial feedback has shown this causes some confusion. To
4265 clarify, the two envisaged usage scenarios are:
4267 (1) Normal text output. In this case, do not specify --xml=yes
4268 nor any of --xml-file=, --xml-fd= or --xml-socket=.
4270 (2) XML output. In this case, specify --xml=yes, and one of
4271 --xml-file=, --xml-fd= or --xml-socket= to select the XML
4272 destination, one of --log-file=, --log-fd= or --log-socket=
4273 to select the destination for any remaining text messages,
4274 and, importantly, -q.
4276 -q makes Valgrind completely silent on the text channel,
4277 except in the case of critical failures, such as Valgrind
4278 itself segfaulting, or failing to read debugging information.
4279 Hence, in this scenario, it suffices to check whether or not
4280 any output appeared on the text channel. If yes, then it is
4281 likely to be a critical error which should be brought to the
4282 attention of the user. If no (the text channel produced no
4283 output) then it can be assumed that the run was successful.
4285 This allows GUIs to make the critical distinction they need to
4286 make (did the run fail or not?) without having to search or
4287 filter the text output channel in any way.
4289 It is also recommended to use --child-silent-after-fork=yes in
4293 * Improvements and changes in Helgrind:
4295 - XML output, as described above
4297 - Checks for consistent association between pthread condition
4298 variables and their associated mutexes are now performed.
4300 - pthread_spinlock functions are supported.
4302 - Modest performance improvements.
4304 - Initial (skeletal) support for describing the behaviour of
4305 non-POSIX synchronisation objects through ThreadSanitizer
4306 compatible ANNOTATE_* macros.
4308 - More controllable tradeoffs between performance and the level of
4309 detail of "previous" accesses in a race. There are now three
4312 * --history-level=full. This is the default, and was also the
4313 default in 3.4.x. It shows both stacks involved in a race, but
4314 requires a lot of memory and can be very slow in programs that
4315 do many inter-thread synchronisation events.
4317 * --history-level=none. This only shows the later stack involved
4318 in a race. This can be much faster than --history-level=full,
4319 but makes it much more difficult to find the other access
4320 involved in the race.
4322 The new intermediate setting is
4324 * --history-level=approx
4326 For the earlier (other) access, two stacks are presented. The
4327 earlier access is guaranteed to be somewhere in between the two
4328 program points denoted by those stacks. This is not as useful
4329 as showing the exact stack for the previous access (as per
4330 --history-level=full), but it is better than nothing, and it's
4331 almost as fast as --history-level=none.
4334 * New features and improvements in DRD:
4336 - The error messages printed by DRD are now easier to interpret.
4337 Instead of using two different numbers to identify each thread
4338 (Valgrind thread ID and DRD thread ID), DRD does now identify
4339 threads via a single number (the DRD thread ID). Furthermore
4340 "first observed at" information is now printed for all error
4341 messages related to synchronization objects.
4343 - Added support for named semaphores (sem_open() and sem_close()).
4345 - Race conditions between pthread_barrier_wait() and
4346 pthread_barrier_destroy() calls are now reported.
4348 - Added support for custom allocators through the macros
4349 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
4350 in <valgrind/valgrind.h>). An alternative for these two macros is
4351 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
4354 - Added support for annotating non-POSIX synchronization objects
4355 through several new ANNOTATE_*() macros.
4357 - OpenMP: added support for the OpenMP runtime (libgomp) included
4358 with gcc versions 4.3.0 and 4.4.0.
4362 - Added two new command-line options (--first-race-only and
4363 --segment-merging-interval).
4366 * Genuinely atomic support for x86/amd64/ppc atomic instructions
4368 Valgrind will now preserve (memory-access) atomicity of LOCK-
4369 prefixed x86/amd64 instructions, and any others implying a global
4370 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
4372 This means that Valgrinded processes will "play nicely" in
4373 situations where communication with other processes, or the kernel,
4374 is done through shared memory and coordinated with such atomic
4375 instructions. Prior to this change, such arrangements usually
4376 resulted in hangs, races or other synchronisation failures, because
4377 Valgrind did not honour atomicity of such instructions.
4380 * A new experimental tool, BBV, has been added. BBV generates basic
4381 block vectors for use with the SimPoint analysis tool, which allows
4382 a program's overall behaviour to be approximated by running only a
4383 fraction of it. This is useful for computer architecture
4384 researchers. You can run BBV by specifying --tool=exp-bbv (the
4385 "exp-" prefix is short for "experimental"). BBV was written by
4389 * Modestly improved support for running Windows applications under
4390 Wine. In particular, initial support for reading Windows .PDB debug
4391 information has been added.
4394 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
4395 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
4399 * The Valgrind client requests VALGRIND_PRINTF and
4400 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
4401 the string was always printed immediately on its own line. Now, the
4402 string will be added to a buffer but not printed until a newline is
4403 encountered, or other Valgrind output is printed (note that for
4404 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
4405 "other Valgrind output"). This allows you to use multiple
4406 VALGRIND_PRINTF calls to build up a single output line, and also to
4407 print multiple output lines with a single request (by embedding
4408 multiple newlines in the string).
4411 * The graphs drawn by Massif's ms_print program have changed slightly:
4413 - The half-height chars '.' and ',' are no longer drawn, because
4414 they are confusing. The --y option can be used if the default
4415 y-resolution is not high enough.
4417 - Horizontal lines are now drawn after the top of a snapshot if
4418 there is a gap until the next snapshot. This makes it clear that
4419 the memory usage has not dropped to zero between snapshots.
4422 * Something that happened in 3.4.0, but wasn't clearly announced: the
4423 option --read-var-info=yes can be used by some tools (Memcheck,
4424 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
4425 variable type and location information. This makes those tools
4426 start up more slowly and increases memory consumption, but
4427 descriptions of data addresses in error messages become more
4431 * exp-Omega, an experimental instantaneous leak-detecting tool, was
4432 disabled in 3.4.0 due to a lack of interest and maintenance,
4433 although the source code was still in the distribution. The source
4434 code has now been removed from the distribution. For anyone
4435 interested, the removal occurred in SVN revision r10247.
4438 * Some changes have been made to the build system.
4440 - VEX/ is now integrated properly into the build system. This means
4441 that dependency tracking within VEX/ now works properly, "make
4442 install" will work without requiring "make" before it, and
4443 parallel builds (ie. 'make -j') now work (previously a
4444 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
4445 was effectively ignored).
4447 - The --with-vex configure option has been removed. It was of
4448 little use and removing it simplified the build system.
4450 - The location of some install files has changed. This should not
4451 affect most users. Those who might be affected:
4453 * For people who use Valgrind with MPI programs, the installed
4454 libmpiwrap.so library has moved from
4455 $(INSTALL)/<platform>/libmpiwrap.so to
4456 $(INSTALL)/libmpiwrap-<platform>.so.
4458 * For people who distribute standalone Valgrind tools, the
4459 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
4460 have moved to $(INSTALL)/libcoregrind-<platform>.a.
4462 These changes simplify the build system.
4464 - Previously, all the distributed suppression (*.supp) files were
4465 installed. Now, only default.supp is installed. This should not
4466 affect users as the other installed suppression files were not
4467 read; the fact that they were installed was a mistake.
4470 * KNOWN LIMITATIONS:
4472 - Memcheck is unusable with the Intel compiler suite version 11.1,
4473 when it generates code for SSE2-and-above capable targets. This
4474 is because of icc's use of highly optimised inlined strlen
4475 implementations. It causes Memcheck to report huge numbers of
4476 false errors even in simple programs. Helgrind and DRD may also
4479 Versions 11.0 and earlier may be OK, but this has not been
4483 The following bugs have been fixed or resolved. Note that "n-i-bz"
4484 stands for "not in bugzilla" -- that is, a bug that was reported to us
4485 but never got a bugzilla entry. We encourage you to file bugs in
4486 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4487 mailing the developers (or mailing lists) directly -- bugs that are
4488 not entered into bugzilla tend to get forgotten about or ignored.
4490 To see details of a given bug, visit
4491 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4492 where XXXXXX is the bug number as listed below.
4494 84303 How about a LockCheck tool?
4495 91633 dereference of null ptr in vgPlain_st_basetype
4496 97452 Valgrind doesn't report any pthreads problems
4497 100628 leak-check gets assertion failure when using
4498 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4499 108528 NPTL pthread cleanup handlers not called
4500 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4501 110128 mallinfo is not implemented...
4502 110770 VEX: Generated files not always updated when making valgrind
4503 111102 Memcheck: problems with large (memory footprint) applications
4504 115673 Vex's decoder should never assert
4505 117564 False positive: Syscall param clone(child_tidptr) contains
4506 uninitialised byte(s)
4507 119404 executing ssh from inside valgrind fails
4508 133679 Callgrind does not write path names to sources with dwarf debug
4510 135847 configure.in problem with non gnu compilers (and possible fix)
4511 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4512 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4513 136230 memcheck reports "possibly lost", should be "still reachable"
4514 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4515 137904 Valgrind reports a memory leak when using POSIX threads,
4517 139076 valgrind VT_GETSTATE error
4518 142228 complaint of elf_dynamic_do_rela in trivial usage
4519 145347 spurious warning with USBDEVFS_REAPURB
4520 148441 (wine) can't find memory leak in Wine, win32 binary
4522 148742 Leak-check fails assert on exit
4523 149878 add (proper) check for calloc integer overflow
4524 150606 Call graph is broken when using callgrind control
4525 152393 leak errors produce an exit code of 0. I need some way to
4526 cause leak errors to result in a nonzero exit code.
4527 157154 documentation (leak-resolution doc speaks about num-callers
4528 def=4) + what is a loss record
4529 159501 incorrect handling of ALSA ioctls
4530 162020 Valgrinding an empty/zero-byte file crashes valgrind
4531 162482 ppc: Valgrind crashes while reading stabs information
4532 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4533 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4534 163560 VEX/test_main.c is missing from valgrind-3.3.1
4535 164353 malloc_usable_size() doesn't return a usable size
4536 165468 Inconsistent formatting in memcheck manual -- please fix
4537 169505 main.c:286 (endOfInstr):
4538 Assertion 'ii->cost_offset == *cost_offset' failed
4539 177206 Generate default.supp during compile instead of configure
4540 177209 Configure valt_load_address based on arch+os
4541 177305 eventfd / syscall 323 patch lost
4542 179731 Tests fail to build because of inlining of non-local asm labels
4543 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4544 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4545 181594 Bogus warning for empty text segment
4546 181707 dwarf doesn't require enumerations to have name
4547 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4548 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4549 Assertion '!already_present' failed.
4550 185359 exp-ptrcheck: unhandled syscall getresuid()
4551 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4552 185816 Valgrind is unable to handle debug info for files with split
4553 debug info that are prelinked afterwards
4554 185980 [darwin] unhandled syscall: sem_open
4555 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4556 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4557 186790 Suppression pattern used for leaks are not reported
4558 186796 Symbols with length>200 in suppression files are ignored
4559 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4560 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4561 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4562 188046 bashisms in the configure script
4563 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4564 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4565 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4566 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4567 assertion fail "!lock->heldBy"
4568 188427 Add support for epoll_create1 (with patch)
4569 188530 Support for SIOCGSTAMPNS
4570 188560 Include valgrind.spec in the tarball
4571 188572 Valgrind on Mac should suppress setenv() mem leak
4572 189054 Valgrind fails to build because of duplicate non-local asm labels
4573 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4574 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4575 189763 drd assertion failure: s_threadinfo[tid].is_recording
4576 190219 unhandled syscall: 328 (x86-linux)
4577 190391 dup of 181394; see above
4578 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4579 190820 No debug information on powerpc-linux
4580 191095 PATCH: Improve usbdevfs ioctl handling
4581 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4583 191189 --xml=yes should obey --gen-suppressions=all
4584 191192 syslog() needs a suppression on macosx
4585 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4586 191761 getrlimit on MacOSX
4587 191992 multiple --fn-skip only works sometimes; dependent on order
4588 192634 V. reports "aspacem sync_check_mapping_callback:
4589 segment mismatch" on Darwin
4590 192954 __extension__ missing on 2 client requests
4591 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4592 194474 "INSTALL" file has different build instructions than "README"
4593 194671 Unhandled syscall (sem_wait?) from mac valgrind
4594 195069 memcheck: reports leak (memory still reachable) for
4596 195169 drd: (vgDrd_barrier_post_wait):
4597 Assertion 'r->sg[p->post_iteration]' failed.
4598 195268 valgrind --log-file doesn't accept ~/...
4599 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4600 195860 WARNING: unhandled syscall: unix:223
4601 196528 need a error suppression for pthread_rwlock_init under os x?
4602 197227 Support aio_* syscalls on Darwin
4603 197456 valgrind should reject --suppressions=(directory)
4604 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4605 197591 unhandled syscall 27 (mincore)
4606 197793 Merge DCAS branch to the trunk == 85756, 142103
4607 197794 Avoid duplicate filenames in Vex
4608 197898 make check fails on current SVN
4609 197901 make check fails also under exp-ptrcheck in current SVN
4610 197929 Make --leak-resolution=high the default
4611 197930 Reduce spacing between leak reports
4612 197933 Print command line of client at start-up, and shorten preamble
4613 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4614 198395 add BBV to the distribution as an experimental tool
4615 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4616 198649 callgrind_annotate doesn't cumulate counters
4617 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4618 199977 Valgrind complains about an unrecognized instruction in the
4619 atomic_incs test program
4620 200029 valgrind isn't able to read Fedora 12 debuginfo
4621 200760 darwin unhandled syscall: unix:284
4622 200827 DRD doesn't work on Mac OS X
4623 200990 VG_(read_millisecond_timer)() does not work correctly
4624 201016 Valgrind does not support pthread_kill() on Mac OS
4625 201169 Document --read-var-info
4626 201323 Pre-3.5.0 performance sanity checking
4627 201384 Review user manual for the 3.5.0 release
4628 201585 mfpvr not implemented on ppc
4629 201708 tests failing because x86 direction flag is left set
4630 201757 Valgrind doesn't handle any recent sys_futex additions
4631 204377 64-bit valgrind can not start a shell script
4632 (with #!/path/to/shell) if the shell is a 32-bit executable
4633 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4634 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4635 about memory allocation events with memory access tracing enabled
4636 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4638 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4642 Release 3.4.1 (28 February 2009)
4643 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4644 3.4.1 is a bug-fix release that fixes some regressions and assertion
4645 failures in debug info reading in 3.4.0, most notably incorrect stack
4646 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4647 other debug info problems are also fixed. A number of bugs in the
4648 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4650 In view of the fact that 3.4.0 contains user-visible regressions
4651 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4652 encouraged to ship 3.4.1 in preference to 3.4.0.
4654 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4655 bugzilla" -- that is, a bug that was reported to us but never got a
4656 bugzilla entry. We encourage you to file bugs in bugzilla
4657 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4658 developers (or mailing lists) directly -- bugs that are not entered
4659 into bugzilla tend to get forgotten about or ignored.
4661 n-i-bz Fix various bugs reading icc-11 generated debug info
4662 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4663 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4664 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4665 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4666 179618 exp-ptrcheck crashed / exit prematurely
4667 179624 helgrind: false positive races with pthread_create and
4668 recv/open/close/read
4669 134207 pkg-config output contains @VG_PLATFORM@
4670 176926 floating point exception at valgrind startup with PPC 440EPX
4671 181594 Bogus warning for empty text segment
4672 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4673 181707 Dwarf3 doesn't require enumerations to have name
4674 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4675 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4676 Assertion '!already_present' failed.
4677 185359 exp-ptrcheck unhandled syscall getresuid()
4679 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4680 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4684 Release 3.4.0 (2 January 2009)
4685 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4686 3.4.0 is a feature release with many significant improvements and the
4687 usual collection of bug fixes. This release supports X86/Linux,
4688 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4689 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4691 3.4.0 brings some significant tool improvements. Memcheck can now
4692 report the origin of uninitialised values, the thread checkers
4693 Helgrind and DRD are much improved, and we have a new experimental
4694 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4695 global arrays. In detail:
4697 * Memcheck is now able to track the origin of uninitialised values.
4698 When it reports an uninitialised value error, it will try to show
4699 the origin of the value, as either a heap or stack allocation.
4700 Origin tracking is expensive and so is not enabled by default. To
4701 use it, specify --track-origins=yes. Memcheck's speed will be
4702 essentially halved, and memory usage will be significantly
4703 increased. Nevertheless it can drastically reduce the effort
4704 required to identify the root cause of uninitialised value errors,
4705 and so is often a programmer productivity win, despite running more
4708 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4709 3.4.0, will be released shortly.
4711 * Helgrind's race detection algorithm has been completely redesigned
4712 and reimplemented, to address usability and scalability concerns:
4714 - The new algorithm has a lower false-error rate: it is much less
4715 likely to report races that do not really exist.
4717 - Helgrind will display full call stacks for both accesses involved
4718 in a race. This makes it easier to identify the root causes of
4721 - Limitations on the size of program that can run have been removed.
4723 - Performance has been modestly improved, although that is very
4726 - Direct support for Qt4 threading has been added.
4728 - pthread_barriers are now directly supported.
4730 - Helgrind works well on all supported Linux targets.
4732 * The DRD thread debugging tool has seen major improvements:
4734 - Greatly improved performance and significantly reduced memory
4737 - Support for several major threading libraries (Boost.Thread, Qt4,
4738 glib, OpenMP) has been added.
4740 - Support for atomic instructions, POSIX semaphores, barriers and
4741 reader-writer locks has been added.
4743 - Works now on PowerPC CPUs too.
4745 - Added support for printing thread stack usage at thread exit time.
4747 - Added support for debugging lock contention.
4749 - Added a manual for Drd.
4751 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4752 checks for misuses of pointers. In that sense it is a bit like
4753 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4754 detect overruns of stack and global arrays, it can detect
4755 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4756 detect accesses heap blocks that have been freed a very long time
4757 ago (millions of blocks in the past).
4759 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4760 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4761 of the main Valgrind documentation. As this is an experimental
4762 tool, we would be particularly interested in hearing about your
4763 experiences with it.
4765 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4766 longer built by default, although the code remains in the repository
4767 and the tarball. This is due to three factors: a perceived lack of
4768 users, a lack of maintenance, and concerns that it may not be
4769 possible to achieve reliable operation using the existing design.
4771 * As usual, support for the latest Linux distros and toolchain
4772 components has been added. It should work well on Fedora Core 10,
4773 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4774 state) is supported, as is glibc-2.9. The C++ demangler has been
4775 updated so as to work well with C++ compiled by even the most recent
4778 * You can now use frame-level wildcards in suppressions. This was a
4779 frequently-requested enhancement. A line "..." in a suppression now
4780 matches zero or more frames. This makes it easier to write
4781 suppressions which are precise yet insensitive to changes in
4784 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4786 * Very basic support for IBM Power6 has been added (64-bit processes only).
4788 * Valgrind is now cross-compilable. For example, it is possible to
4789 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4790 on a ppc32/64-linux target.
4792 * You can set the main thread's stack size at startup using the
4793 new --main-stacksize= flag (subject of course to ulimit settings).
4794 This is useful for running apps that need a lot of stack space.
4796 * The limitation that you can't use --trace-children=yes together
4797 with --db-attach=yes has been removed.
4799 * The following bugs have been fixed. Note that "n-i-bz" stands for
4800 "not in bugzilla" -- that is, a bug that was reported to us but
4801 never got a bugzilla entry. We encourage you to file bugs in
4802 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4803 mailing the developers (or mailing lists) directly.
4805 n-i-bz Make return types for some client requests 64-bit clean
4806 n-i-bz glibc 2.9 support
4807 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4808 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4809 n-i-bz Building in an env without gdb gives bogus gdb attach
4810 92456 Tracing the origin of uninitialised memory
4811 106497 Valgrind does not demangle some C++ template symbols
4813 151612 Suppression with "..." (frame-level wildcards in .supp files)
4814 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4815 159285 unhandled syscall:25 (stime, on x86-linux)
4816 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4817 160954 ppc build of valgrind crashes with illegal instruction (isel)
4818 160956 mallinfo implementation, w/ patch
4819 162092 Valgrind fails to start gnome-system-monitor
4820 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4821 163794 assertion failure with "--track-origins=yes"
4822 163933 sigcontext.err and .trapno must be set together
4823 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4824 164476 Missing kernel module loading system calls
4825 164669 SVN regression: mmap() drops posix file locks
4826 166581 Callgrind output corruption when program forks
4827 167288 Patch file for missing system calls on Cell BE
4828 168943 unsupported scas instruction pentium
4829 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4830 172417 x86->IR: 0x82 ...
4831 172563 amd64->IR: 0xD9 0xF5 - fprem1
4832 173099 .lds linker script generation error
4833 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4834 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4836 174908 --log-file value not expanded correctly for core file
4837 175044 Add lookup_dcookie for amd64
4838 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4840 Developer-visible changes:
4842 * Valgrind's debug-info reading machinery has been majorly overhauled.
4843 It can now correctly establish the addresses for ELF data symbols,
4844 which is something that has never worked properly before now.
4846 Also, Valgrind can now read DWARF3 type and location information for
4847 stack and global variables. This makes it possible to use the
4848 framework to build tools that rely on knowing the type and locations
4849 of stack and global variables, for example exp-Ptrcheck.
4851 Reading of such information is disabled by default, because most
4852 tools don't need it, and because it is expensive in space and time.
4853 However, you can force Valgrind to read it, using the
4854 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4855 make use of such information, if present, to provide source-level
4856 descriptions of data addresses in the error messages they create.
4858 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4859 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).