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 460192 Add epoll_pwait2
37 469049 link failure on ppc64 (big endian) valgrind 3.20
38 469146 massif --ignore-fn does not ignore inlined functions
39 469768 Make it possible to install gdb scripts in a different location
40 470121 Can't run callgrind_control with valgrind 3.21.0 because of perl errors
41 470132 s390x: Assertion failure on VGM instruction
42 470520 Multiple realloc zero errors crash in MC_(eq_Error)
43 470713 Failure on the Yosys project: valgrind: m_libcfile.c:1802
44 (Bool vgPlain_realpath(const HChar *, HChar *)):
45 Assertion 'resolved' failed
46 470830 Don't print actions vgdb me ... continue for vgdb --multi mode
47 470978 s390x: Valgrind cannot start qemu-kvm when "sysctl vm.allocate_pgste=0"
48 471807 Add support for lazy reading and downloading of DWARF debuginfo
49 472219 Syscall param ppoll(ufds.events) points to uninitialised byte(s)
50 472963 Broken regular expression in configure.ac
52 To see details of a given bug, visit
53 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
54 where XXXXXX is the bug number as listed above.
56 Release 3.21.0 (28 Apr 2023)
57 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
59 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
60 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
61 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
62 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
63 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
64 AMD64/macOS 10.13 and nanoMIPS/Linux.
66 * ==================== CORE CHANGES ===================
68 * When GDB is used to debug a program running under valgrind using
69 the valgrind gdbserver, GDB will automatically load some
70 python code provided in valgrind defining GDB front end commands
71 corresponding to the valgrind monitor commands.
72 These GDB front end commands accept the same format as
73 the monitor commands directly sent to the Valgrind gdbserver.
74 These GDB front end commands provide a better integration
75 in the GDB command line interface, so as to use for example
76 GDB auto-completion, command specific help, searching for
77 a command or command help matching a regexp, ...
78 For relevant monitor commands, GDB will evaluate arguments
79 to make the use of monitor commands easier.
80 For example, instead of having to print the address of a variable
81 to pass it to a subsequent monitor command, the GDB front end
82 command will evaluate the address argument. It is for example
84 (gdb) memcheck who_points_at &some_struct sizeof(some_struct)
87 $2 = (some_struct_type *) 0x1130a0 <some_struct>
88 (gdb) p sizeof(some_struct)
90 (gdb) monitor who_point_at 0x1130a0 40
92 * The vgdb utility now supports extended-remote protocol when
93 invoked with --multi. In this mode the GDB run command is
94 supported. Which means you don't need to run gdb and valgrind
95 from different terminals. So for example to start your program
96 in gdb and run it under valgrind you can do:
98 (gdb) set remote exec-file prog
100 (gdb) target extended-remote | vgdb --multi
103 * The behaviour of realloc with a size of zero can now
104 be changed for tools that intercept malloc. Those
105 tools are memcheck, helgrind, drd, massif and dhat.
106 Realloc implementations generally do one of two things
107 - free the memory like free() and return NULL
108 (GNU libc and ptmalloc).
109 - either free the memory and then allocate a
110 minimum sized block or just return the
111 original pointer. Return NULL if the
112 allocation of the minimum sized block fails
113 (jemalloc, musl, snmalloc, Solaris, macOS).
114 When Valgrind is configured and built it will
115 try to match the OS and libc behaviour. However
116 if you are using a non-default library to replace
117 malloc and family (e.g., musl on a glibc Linux or
118 tcmalloc on FreeBSD) then you can use a command line
119 option to change the behaviour of Valgrind:
120 --realloc-zero-bytes-frees=yes|no [yes on Linux glibc, no otherwise]
122 * ================== PLATFORM CHANGES =================
124 * Make the address space limit on FreeBSD amd64 128Gbytes
125 (the same as Linux and Solaris, it was 32Gbytes)
127 * ==================== TOOL CHANGES ===================
130 - When doing a delta leak_search, it is now possible to only
131 output the new loss records compared to the previous leak search.
132 This is available in the memcheck monitor command 'leak_search'
133 by specifying the "new" keyword or in your program by using
134 the client request VALGRIND_DO_NEW_LEAK_CHECK.
135 Whenever a "delta" leak search is done (i.e. when specifying
136 "new" or "increased" or "changed" in the monitor command),
137 the new loss records have a "new" marker.
138 - Valgrind now contains python code that defines GDB memcheck
139 front end monitor commands. See CORE CHANGES.
140 - Performs checks for the use of realloc with a size of zero.
141 This is non-portable and a source of errors. If memcheck
142 detects such a usage it will generate an error
143 realloc() with size 0
144 followed by the usual callstacks.
145 A switch has been added to allow this to be turned off:
146 --show-realloc-size-zero=yes|no [yes]
149 - The option ---history-backtrace-size=<number> allows to configure
150 the number of entries to record in the stack traces of "old"
151 accesses. Previously, this number was hardcoded to 8.
152 - Valgrind now contains python code that defines GDB helgrind
153 front end monitor commands. See CORE CHANGES.
156 - `--cache-sim=no` is now the default. The cache simulation is old and
157 unlikely to match any real modern machine. This means only the `Ir`
158 event are gathered by default, but that is by far the most useful
160 - `cg_annotate`, `cg_diff`, and `cg_merge` have been rewritten in
161 Python. As a result, they all have more flexible command line
162 argument handling, e.g. supporting `--show-percs` and
163 `--no-show-percs` forms as well as the existing `--show-percs=yes`
164 and `--show-percs=no`.
165 - `cg_annotate` has some functional changes.
166 - It's much faster, e.g. 3-4x on common cases.
167 - It now supports diffing (with `--diff`, `--mod-filename`, and
168 `--mod-funcname`) and merging (by passing multiple data files).
169 - It now provides more information at the file and function level.
170 There are now "File:function" and "Function:file" sections. These
171 are very useful for programs that use inlining a lot.
172 - Support for user-annotated files and the `-I`/`--include` option
173 has been removed, because it was of little use and blocked other
175 - The `--auto` option is renamed `--annotate`, though the old
176 `--auto=yes`/`--auto=no` forms are still supported.
177 - `cg_diff` and `cg_merge` are now deprecated, because `cg_annotate`
178 now does a better job of diffing and merging.
179 - The Cachegrind output file format has changed very slightly, but in
180 ways nobody is likely to notice.
183 - Valgrind now contains python code that defines GDB callgrind
184 front end monitor commands. See CORE CHANGES.
187 - Valgrind now contains python code that defines GDB massif
188 front end monitor commands. See CORE CHANGES.
191 - A new kind of user request has been added which allows you to
192 override the 1024 byte limit on access count histograms for blocks
193 of memory. The client request is DHAT_HISTOGRAM_MEMORY.
195 * ==================== FIXED BUGS ====================
197 The following bugs have been fixed or resolved. Note that "n-i-bz"
198 stands for "not in bugzilla" -- that is, a bug that was reported to us
199 but never got a bugzilla entry. We encourage you to file bugs in
200 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
201 than mailing the developers (or mailing lists) directly -- bugs that
202 are not entered into bugzilla tend to get forgotten about or ignored.
204 170510 Don't warn about ioctl of size 0 without direction hint
205 241072 List tools in --help output
206 327548 false positive while destroying mutex
207 382034 Testcases build fixes for musl
208 351857 confusing error message about valid command line option
209 374596 inconsistent RDTSCP support on x86_64
210 392331 Spurious lock not held error from inside pthread_cond_timedwait
211 397083 Likely false positive "uninitialised value(s)" for __wmemchr_avx2 and __wmemcmp_avx2_movbe
212 400793 pthread_rwlock_timedwrlock false positive
213 419054 Unhandled syscall getcpu on arm32
214 433873 openat2 syscall unimplemented on Linux
215 434057 Add stdio mode to valgrind's gdbserver
216 435441 valgrind fails to interpose malloc on musl 1.2.2 due to weak symbol name and no libc soname
217 436413 Warn about realloc of size zero
218 439685 compiler warning in callgrind/main.c
219 444110 priv/guest_ppc_toIR.c:36198:31: warning: duplicated 'if' condition.
220 444487 hginfo test detects an extra lock inside data symbol "_rtld_local"
221 444488 Use glibc.pthread.stack_cache_size tunable
222 444568 drd/tests/pth_barrier_thr_cr fails on Fedora 38
223 445743 "The impossible happened: mutex is locked simultaneously by two threads"
224 while using mutexes with priority inheritance and signals
225 449309 Missing loopback device ioctl(s)
226 459476 vgdb: allow address reuse to avoid "address already in use" errorsuse" errors
227 460356 s390: Sqrt32Fx4 -- cannot reduce tree
228 462830 WARNING: unhandled amd64-freebsd syscall: 474
229 463027 broken check for MPX instruction support in assembler
230 464103 Enhancement: add a client request to DHAT to mark memory to be histogrammed
231 464476 Firefox fails to start under Valgrind
232 464609 Valgrind memcheck should support Linux pidfd_open
233 464680 Show issues caused by memory policies like selinux deny_execmem
234 464859 Build failures with GCC-13 (drd tsan_unittest)
235 464969 D language demangling
236 465435 m_libcfile.c:66 (vgPlain_safe_fd): Assertion 'newfd >= VG_(fd_hard_limit)' failed.
237 466104 aligned_alloc problems, part 1
238 467036 Add time cost statistics for Regtest
239 467482 Build failure on aarch64 Alpine
240 467714 fdleak_* and rlimit tests fail when parent process has more than
241 64 descriptors opened
242 467839 Gdbserver: Improve compatibility of library directory name
243 468401 [PATCH] Add a style file for clang-format
244 468556 Build failure for vgdb
245 468606 build: remove "Valgrind relies on GCC" check/output
246 469097 ppc64(be) doesn't support SCV syscall instruction
247 n-i-bz FreeBSD rfork syscall fail with EINVAL or ENOSYS rather than VG_(unimplemented)
249 To see details of a given bug, visit
250 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
251 where XXXXXX is the bug number as listed above.
253 * ==================== KNOWN ISSUES ===================
255 * configure --enable-lto=yes is know to not work in all setups.
256 See bug 469049. Workaround: Build without LTO.
258 (3.21.0.RC1: 14 Apr 2023)
259 (3.21.0.RC2: 21 Apr 2023)
261 Release 3.20.0 (24 Oct 2022)
262 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
264 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
265 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
266 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
267 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
268 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
269 AMD64/macOS 10.13 and nanoMIPS/Linux.
271 * ==================== CORE CHANGES ===================
273 * The option "--vgdb-stop-at=event1,event2,..." accepts the new value abexit.
274 This indicates to invoke gdbserver when your program exits abnormally
275 (i.e. with a non zero exit code).
276 * Fix Rust v0 name demangling.
277 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
278 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
279 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with
281 * The option --enable-debuginfod=<no|yes> [default: yes] has been added on
283 * More DWARF5 support as generated by clang14.
285 * ==================== FIXED BUGS ====================
287 The following bugs have been fixed or resolved. Note that "n-i-bz"
288 stands for "not in bugzilla" -- that is, a bug that was reported to us
289 but never got a bugzilla entry. We encourage you to file bugs in
290 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
291 than mailing the developers (or mailing lists) directly -- bugs that
292 are not entered into bugzilla tend to get forgotten about or ignored.
294 131186 writev reports error in (vector[...])
295 434764 iconv_open causes ld.so v2.28+ to use optimised strncmp
296 446754 Improve error codes from alloc functions under memcheck
297 452274 memcheck crashes with Assertion 'sci->status.what == SsIdle' failed
298 452779 Valgrind fails to build on FreeBSD 13.0 with llvm-devel (15.0.0)
299 453055 shared_timed_mutex drd test fails with "Lock shared failed" message
300 453602 Missing command line option to enable/disable debuginfod
301 452802 Handle lld 9+ split RW PT_LOAD segments correctly
302 454040 s390x: False-positive memcheck:cond in memmem on arch13 systems
303 456171 [PATCH] FreeBSD: Don't record address errors when accessing the 'kern.ps_strings' sysctl struct
304 n-i-bz Implement vgdb invoker on FreeBSD
305 458845 PowerPC: The L field for the dcbf and sync instruction should be
307 458915 Remove register cache to fix 458915 gdbserver causes wrong syscall return
308 459031 Documentation on --error-exitcode incomplete
309 459477 XERROR messages lacks ending '\n' in vgdb
310 462007 Implicit int in none/tests/faultstatus.c
312 To see details of a given bug, visit
313 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
314 where XXXXXX is the bug number as listed above.
316 (3.20.0.RC1: 20 Oct 2022)
319 Release 3.19.0 (11 Apr 2022)
320 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
322 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
323 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
324 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
325 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
326 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
327 AMD64/macOS 10.13 and nanoMIPS/Linux.
329 * ==================== CORE CHANGES ===================
331 * Fix Rust v0 name demangling.
332 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
333 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
334 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with HardenedBSD
336 * ================== PLATFORM CHANGES =================
339 - ignore the "v8.x" architecture levels, only look at actual CPU features
340 present. Fixes mismatch detected between RDMA and atomics features
341 preventing startup on some QEMU configurations.
342 - Implement LD{,A}XP and ST{,L}XP
343 - Fix incorrect code emitted for doubleword CAS.
346 - Fix sys_ipc semtimedop syscall
347 - Fix VFLRX and WFLRX instructions
348 - Fix EXRL instruction with negative offset
351 - Reimplement the vbpermq instruction support to generate less Iops and
352 avoid overflowing internal buffers.
353 - Fix checking for scv support to avoid "Facility 'SCV' unavailable (12),
354 exception" messages in dmsg.
355 - Fix setting condition code for Vector Compare quad word instructions.
356 - Fix fix lxsibzx, lxsihzx and lxsihzx instructions so they only load
357 their respective sized data.
358 - Fix the prefixed stq instruction in PC relative mode.
360 * ==================== TOOL CHANGES ===================
363 - Speed up --track-origins=yes for large (in the range of hundreds to
364 thousands of megabytes) mmap/munmaps.
366 - Several fixes for new versions of libstd++ using new posix try_lock
369 * ==================== FIXED BUGS ====================
371 The following bugs have been fixed or resolved. Note that "n-i-bz"
372 stands for "not in bugzilla" -- that is, a bug that was reported to us
373 but never got a bugzilla entry. We encourage you to file bugs in
374 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
375 than mailing the developers (or mailing lists) directly -- bugs that
376 are not entered into bugzilla tend to get forgotten about or ignored.
378 403802 leak_cpp_interior fails with some reachable blocks different than expected
379 435732 memcheck/tests/leak_cpp_interior fails with gcc11
380 444242 s390x: Valgrind crashes on EXRL with negative offset
381 444399 arm64: unhandled instruction 0xC87F2D89 (LD{,A}XP and ST{,L}XP).
383 444481 gdb_server test failures on s390x
384 444495 dhat/tests/copy fails on s390x
385 444552 memcheck/tests/sem fails on s390x with glibc 2.34
386 444571 PPC, fix the lxsibzx and lxsihzx so they only load their respective
388 444836 PPC, pstq instruction for R=1 is not storing to the correct address.
389 444925 fexecve syscall wrapper not properly implemented
390 445032 valgrind/memcheck crash with SIGSEGV when SIGVTALRM timer used and
392 445211 Fix out of tree builds
393 445300 [PATCH] Fix building tests with Musl
394 445011 SIGCHLD is sent when valgrind uses debuginfod-find
395 445354 arm64 backend: incorrect code emitted for doubleword CAS
396 445415 arm64 front end: alignment checks missing for atomic instructions
397 445504 Using C++ condition_variable results in bogus "mutex is locked simultaneously by two threads" warning
398 445607 Unhandled amd64-freebsd syscall: 247
399 445668 Inline stack frame generation is broken for Rust binaries
400 445916 Demangle Rust v0 symbols with .llvm suffix
401 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false positives
402 446138 DRD/Helgrind with std::timed_mutex::try_lock_until false positives
403 446281 Add a DRD suppression for fwrite
404 446103 Memcheck: `--track-origins=yes` causes extreme slowdowns for large mmap/munmap
405 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false
406 446251 TARGET_SIGNAL_THR added to enum target_signal
407 446823 FreeBSD - missing syscalls when using libzm4
408 447991 s390x: Valgrind indicates illegal instruction on wflrx
409 447995 Valgrind segfault on power10 due to hwcap checking code
410 449483 Powerpc: vcmpgtsq., vcmpgtuq,, vcmpequq. instructions not setting the
411 condition code correctly.
412 449672 ppc64 --track-origins=yes failures because of bad cmov addHRegUse
413 449838 sigsegv liburing the 'impossible' happened for io_uring_setup
414 450025 Powerc: ACC file not implemented as a logical overlay of the VSR
416 450437 Warn for execve syscall with argv or argv[0] being NULL
417 450536 Powerpc: valgrind throws 'facility scv unavailable exception'
418 451626 Syscall param bpf(attr->raw_tracepoint.name) points to unaddressable byte(s)
419 451827 [ppc64le] VEX temporary storage exhausted with several vbpermq instructions
420 451843 valgrind fails to start on a FreeBSD system which enforces W^X
422 To see details of a given bug, visit
423 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
424 where XXXXXX is the bug number as listed above.
426 (3.19.0.RC1: 02 Apr 2022)
427 (3.19.0.RC2: 08 Apr 2022)
430 Release 3.18.0 (15 Oct 2021)
431 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
433 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
434 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
435 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
436 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
437 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
438 AMD64/macOS 10.13 and nanoMIPS/Linux.
440 * ==================== CORE CHANGES ===================
442 * The libiberty demangler has been updated, which brings support for
443 Rust v0 name demangling. [Update: alas, due to a bug, this support
444 isn't working in 3.18.0.]
446 * __libc_freeres isn't called anymore after the program recieves a
447 fatal signal. Causing some internal glibc resources to hang around,
448 but preventing any crashes after the program has ended.
450 * The DWARF reader is now very much faster at startup when just
451 --read-inline-info=yes (the default in most cases) is given.
453 * glibc 2.34, which moved various functions from libpthread.so into
454 libc.so, is now supported.
456 * ================== PLATFORM CHANGES =================
460 - v8.2 scalar and vector FABD, FACGE, FACGT and FADD.
461 - v8.2 FP compare & conditional compare instructions.
462 - Zero variants of v8.2 FP compare instructions.
466 - Support the miscellaneous-instruction-extensions facility 3 and
467 the vector-enhancements facility 2. This enables programs
468 compiled with "-march=arch13" or "-march=z15" to be executed
473 - ISA 3.1 support is now complete
474 - ISA 3.0 support for the darn instruction added.
475 - ISA 3.0 support for the vector system call instruction scv added.
476 - ISA 3.0 support for the copy, paste and cpabort instructions added.
478 * Support for X86/FreeBSD and AMD64/FreeBSD has been added.
480 * ==================== OTHER CHANGES ====================
482 * Memcheck on amd64: minor fixes to remove some false positive
485 * ==================== FIXED BUGS ====================
487 The following bugs have been fixed or resolved. Note that "n-i-bz"
488 stands for "not in bugzilla" -- that is, a bug that was reported to us
489 but never got a bugzilla entry. We encourage you to file bugs in
490 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
491 than mailing the developers (or mailing lists) directly -- bugs that
492 are not entered into bugzilla tend to get forgotten about or ignored.
494 208531 [PATCH]: FreeBSD support for valgrind
495 368960 WARNING: unhandled amd64-linux syscall: 163 (acct)
496 407589 [Linux] Add support for C11 aligned_alloc() and GNU reallocarray()
497 423963 Error in child thread when CLONE_PIDFD is used
498 426148 crash with "impossible happened" when running BPF CO-RE programs
499 429375 PPC ISA 3.1 support is missing, part 9
500 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
501 431306 Update demangler to support Rust v0 name mangling
502 432387 s390x: z15 instructions support
503 433437 FreeBSD support, part 1
504 433438 FreeBSD support, part 2
505 433439 FreeBSD support, part 3
506 433469 FreeBSD support, part 4
507 433473 FreeBSD support, part 5
508 433477 FreeBSD support, part 6
509 433479 FreeBSD support, part 7
510 433504 FreeBSD support, part 8
511 433506 FreeBSD support, part 9
512 433507 FreeBSD support, part 10
513 433508 FreeBSD support, part 11
514 433510 FreeBSD support, part 12
515 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
516 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
517 434296 s390x: False-positive memcheck diagnostics from vector string
519 434840 PPC64 darn instruction not supported
520 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
521 435908 valgrind tries to fetch from deubginfod for files which already
522 have debug information
523 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2
524 439046 valgrind is unusably large when linked with lld
525 439090 Implement close_range(2)
526 439326 Valgrind 3.17.0 won't compile with Intel 2021 oneAPI compilers
527 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
528 440670 unhandled ppc64le-linux syscall: 252 statfs64 and 253 fstatfs64
529 440906 Fix impossible constraint issue in P10 testcase.
530 441512 Remove a unneeded / unnecessary prefix check.
531 441534 Update the expected output for test_isa_3_1_VRT.
532 442061 very slow execution under Fedora 34 (readdwarf3)
533 443031 Gcc -many change requires explicit .machine directives
534 443033 Add support for the ISA 3.0 mcrxrx instruction
535 443034 Sraw, srawi, srad, sradi, mfs
536 443178 Powerpc, test jm-mfspr expected output needs to be updated.
537 443179 Need new test for the lxvx and stxvx instructions on ISA 2.07 and
539 443180 The subnormal test and the ISA 3.0 test generate compiler warnings
540 443314 In the latest GIT version, Valgrind with "--trace-flags" crashes
542 443605 Don't call final_tidyup (__libc_freeres) on FatalSignal
544 To see details of a given bug, visit
545 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
546 where XXXXXX is the bug number as listed below.
548 (3.18.0.RC1: 12 Oct 2021)
549 (3.18.0: 15 Oct 2021)
553 Release 3.17.0 (19 Mar 2021)
554 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
556 3.17.0 fixes a number of bugs and adds some functional changes: support for
557 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
558 some new instructions for Arm64, S390 and POWER. There are also some tool
561 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
562 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
563 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
564 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
565 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
567 * ==================== CORE CHANGES ===================
569 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
572 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
573 debugging information. When a debuginfo file cannot be found locally,
574 Valgrind is able to query debuginfod servers for the file using its
575 build-id. See the user manual for more information about debuginfod support.
577 * ================== PLATFORM CHANGES =================
581 - Inaccuracies resulting from double-rounding in the simulation of
582 floating-point multiply-add/subtract instructions have been fixed. These
583 should now behave exactly as the hardware does.
585 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
586 ongoing. Support for the half-word variants of at least the following
587 instructions has been added:
589 FABS <Vd>.<T>, <Vn>.<T>
591 FNEG <Vd>.<T>, <Vn>.<T>
593 FSQRT <Vd>.<T>, <Vn>.<T>
598 - Implement the new instructions/features that were added to z/Architecture
599 with the vector-enhancements facility 1. Also cover the instructions from
600 the vector-packed-decimal facility that are defined outside the chapter
601 "Vector Decimal Instructions", but not the ones from that chapter itself.
603 For a detailed list of newly supported instructions see the updates to
604 `docs/internals/s390-opcodes.csv'.
606 Since the miscellaneous instruction extensions facility 2 was already
607 added in Valgrind 3.16.0, this completes the support necessary to run
608 general programs built with `--march=z14' under Valgrind. The
609 vector-packed-decimal facility is currently not exploited by the standard
610 toolchain and libraries.
614 - Various bug fixes. Fix for the sync field to limit setting just two of
615 the two bits in the L-field. Fix the write size for the stxsibx and
616 stxsihx instructions. Fix the modsw and modsd instructions.
618 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
619 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
620 conversion instructions are still missing.
622 * ==================== TOOL CHANGES ====================
624 * General tool changes
626 - All the tools and their vgpreload libraries are now installed under
627 libexec because they cannot be executed directly and should be run through
628 the valgrind executable. This should be an internal, not user visible,
629 change, but might impact valgrind packagers.
631 - The --track-fds option now respects -q, --quiet and won't output anything
632 if no file descriptors are leaked. It also won't report the standard stdin
633 (0), stdout (1) or stderr (2) descriptors as being leaked with
634 --trace-fds=yes anymore. To track whether the standard file descriptors
635 are still open at the end of the program run use --trace-fds=all.
639 - DHAT has been extended, with two new modes of operation. The new
640 --mode=copy flag triggers copy profiling, which records calls to memcpy,
641 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
642 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
643 the new dhat/dhat.h file. This is useful for learning more about hot code
644 paths. See the user manual for more information about the new modes.
646 - Because of these changes, DHAT's file format has changed. DHAT output
647 files produced with earlier versions of DHAT will not work with this
648 version of DHAT's viewer, and DHAT output files produced with this version
649 of DHAT will not work with earlier versions of DHAT's viewer.
651 * ==================== FIXED BUGS ====================
653 The following bugs have been fixed or resolved. Note that "n-i-bz"
654 stands for "not in bugzilla" -- that is, a bug that was reported to us
655 but never got a bugzilla entry. We encourage you to file bugs in
656 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
657 than mailing the developers (or mailing lists) directly -- bugs that
658 are not entered into bugzilla tend to get forgotten about or ignored.
660 To see details of a given bug, visit
661 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
662 where XXXXXX is the bug number as listed below.
664 140178 open("/proc/self/exe", ...); doesn't quite work
665 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
666 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
667 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
668 345077 linux syscall execveat support (linux 3.19)
669 361770 Missing F_ADD_SEALS
670 369029 handle linux syscalls sched_getattr and sched_setattr
671 384729 __libc_freeres inhibits cross-platform valgrind
672 388787 Support for C++17 new/delete
673 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
674 396656 Warnings while reading debug info
675 397605 ioctl FICLONE mishandled
676 401416 Compile failure with openmpi 4.0
677 408663 Suppression file for musl libc
678 404076 s390x: z14 vector instructions not implemented
679 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
680 (actually affected all x86 and nanomips regardless of host bitness)
681 413547 regression test does not check for Arm 64 features.
682 414268 Enable AArch64 feature detection and decoding for v8.x instructions
683 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
684 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
685 422261 platform selection fails for unqualified client name
686 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
687 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
688 423195 PPC ISA 3.1 support is missing, part 1
689 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
690 424012 crash with readv/writev having invalid but not NULL arg2 iovec
691 424298 amd64: Implement RDSEED
692 425232 PPC ISA 3.1 support is missing, part 2
693 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
694 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
695 426123 PPC ISA 3.1 support is missing, part 3
696 426144 Fix "condition variable has not been initialized" on Fedora 33.
697 427400 PPC ISA 3.1 support is missing, part 4
698 427401 PPC ISA 3.1 support is missing, part 5
699 427404 PPC ISA 3.1 support is missing, part 6
700 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
701 427787 Support new faccessat2 linux syscall (439)
702 427969 debuginfo section duplicates a section in the main ELF file
703 428035 drd: Unbreak the musl build
704 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
705 428716 cppcheck detects potential leak in VEX/useful/smchash.c
706 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
707 429352 PPC ISA 3.1 support is missing, part 7
708 429354 PPC ISA 3.1 support is missing, part 8
709 429692 unhandled ppc64le-linux syscall: 147 (getsid)
710 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
712 429952 Errors when building regtest with clang
713 430354 ppc stxsibx and stxsihx instructions write too much data
714 430429 valgrind.h doesn't compile on s390x with clang
715 430485 expr_is_guardable doesn't handle Iex_Qop
716 431556 Complete arm64 FADDP v8.2 instruction support
717 432102 Add support for DWARF5 as produced by GCC11
718 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
719 432381 drd: Process STACK_REGISTER client requests
720 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
721 432672 vg_regtest: test-specific environment variables not reset between tests
722 432809 VEX should support REX.W + POPF
723 432861 PPC modsw and modsd give incorrect results for 1 mod 12
724 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
725 432215 Add debuginfod functionality
726 433323 Use pkglibexecdir as vglibdir
727 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
728 433629 valgrind/README has type "abd" instead of "and"
729 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
730 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
731 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
732 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
733 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
735 (3.17.0.RC1: 13 Mar 2021)
736 (3.17.0.RC2: 17 Mar 2021)
737 (3.17.0: 19 Mar 2021)
741 Release 3.16.1 (22 June 2020)
742 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
744 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
745 fixes character encoding problems in the documentation HTML.
747 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
748 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
750 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
754 Release 3.16.0 (27 May 2020)
755 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
757 3.16.0 is a feature release with many improvements and the usual collection of
760 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
761 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
762 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
763 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
764 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
766 * ==================== CORE CHANGES ===================
768 * It is now possible to dynamically change the value of many command line
769 options while your program (or its children) are running under Valgrind.
771 To see the list of dynamically changeable options, run
772 "valgrind --help-dyn-options".
774 You can change the options from the shell by using vgdb to launch
775 the monitor command "v.clo <clo option>...".
776 The same monitor command can be used from a gdb connected
777 to the valgrind gdbserver.
778 Your program can also change the dynamically changeable options using
779 the client request VALGRIND_CLO_CHANGE(option).
781 * ================== PLATFORM CHANGES =================
783 * MIPS: preliminary support for nanoMIPS instruction set has been added.
785 * ==================== TOOL CHANGES ====================
789 - The implicit memcpy done by each call to realloc now counts towards the
790 read and write counts of resized heap blocks, making those counts higher
795 - cg_annotate's --auto and --show-percs options now default to 'yes', because
796 they are usually wanted.
800 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
801 because they are usually wanted.
803 - The command option --collect-systime has been enhanced to specify
804 the unit used to record the elapsed time spent during system calls.
805 The command option now accepts the values no|yes|msec|usec|nsec,
806 where yes is a synonym of msec. When giving the value nsec, the
807 system cpu time of system calls is also recorded.
811 - Several memcheck options are now dynamically changeable.
812 Use valgrind --help-dyn-options to list them.
814 - The release 3.15 introduced a backward incompatible change for
815 some suppression entries related to preadv and pwritev syscalls.
816 When reading a suppression entry using the unsupported 3.14 format,
817 valgrind will now produce a warning to say the suppression entry will not
818 work, and suggest the needed change.
820 - Significantly fewer false positive errors on optimised code generated by
821 Clang and GCC. In particular, Memcheck now deals better with the
822 situation where the compiler will transform C-level "A && B" into "B && A"
823 under certain circumstances (in which the transformation is valid).
824 Handling of integer equality/non-equality checks on partially defined
825 values is also improved on some architectures.
829 - The exprimental Stack and Global Array Checking tool has been removed.
830 It only ever worked on x86 and amd64, and even on those it had a
831 high false positive rate and was slow. An alternative for detecting
832 stack and global array overruns is using the AddressSanitizer (ASAN)
833 facility of the GCC and Clang compilers, which require you to rebuild
834 your code with -fsanitize=address.
836 * ==================== OTHER CHANGES ====================
838 * New and modified GDB server monitor features:
840 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
842 - The gdbserver monitor commands that require an address and an optional
843 length argument now accepts the alternate 'C like' syntax "address[length]".
844 For example, the memcheck command "monitor who_points_at 0x12345678 120"
845 can now also be given as "monitor who_points_at 0x12345678[120]".
847 * ==================== FIXED BUGS ====================
849 The following bugs have been fixed or resolved. Note that "n-i-bz"
850 stands for "not in bugzilla" -- that is, a bug that was reported to us
851 but never got a bugzilla entry. We encourage you to file bugs in
852 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
853 than mailing the developers (or mailing lists) directly -- bugs that
854 are not entered into bugzilla tend to get forgotten about or ignored.
856 To see details of a given bug, visit
857 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
858 where XXXXXX is the bug number as listed below.
860 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
861 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
862 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
863 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
864 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
865 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
866 400872 Add nanoMIPS support to Valgrind
867 403212 drd/tests/trylock hangs on FreeBSD
868 404406 s390x: z14 miscellaneous instructions not implemented
869 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
870 406561 mcinfcallWSRU gdbserver_test fails on ppc64
871 406824 Unsupported baseline
872 407218 Add support for the copy_file_range syscall
873 407307 Intercept stpcpy also in ld.so for arm64
874 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
876 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
877 408009 Expose rdrand and f16c even on avx if host cpu supports them
878 408091 Missing pkey syscalls
879 408414 Add support for missing for preadv2 and pwritev2 syscalls
880 409141 Valgrind hangs when SIGKILLed
881 409206 Support for Linux PPS and PTP ioctls
882 409367 exit_group() after signal to thread waiting in futex() causes hangs
883 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
884 409780 References to non-existent configure.in
885 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
886 410599 Non-deterministic behaviour of pth_self_kill_15_other test
887 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
888 411134 Allow the user to change a set of command line options during execution
889 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
890 412344 Problem setting mips flags with specific paths
891 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
892 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
893 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
894 413603 callgrind_annotate/cg_annotate truncate function names at '#'
895 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
896 415136 ARMv8.1 Compare-and-Swap instructions are not supported
897 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
898 416239 valgrind crashes when handling clock_adjtime
899 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
900 416286 DRD reports "conflicting load" error on std::mutex::lock()
901 416301 s390x: "compare and signal" not supported
902 416387 finit_module and bpf syscalls are unhandled on arm64
903 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
904 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
905 416753 new 32bit time syscalls for 2038+
906 417075 pwritev(vector[...]) suppression ignored
907 417075 is not fixed, but incompatible supp entries are detected
908 and a warning is produced for these.
909 417187 [MIPS] Conditional branch problem since 'grail' changes
910 417238 Test memcheck/tests/vbit-test fails on mips64 BE
911 417266 Make memcheck/tests/linux/sigqueue usable with musl
912 417281 s390x: /bin/true segfaults with "grail" enabled
913 417427 commit to fix vki_siginfo_t definition created numerous regression
915 417452 s390_insn_store_emit: dst->tag for HRcVec128
916 417578 Add suppressions for glibc DTV leaks
917 417906 clone with CLONE_VFORK and no CLONE_VM fails
918 418004 Grail code additions break ppc64.
919 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
920 418997 s390x: Support Iex_ITE for float and vector types
921 419503 s390x: Avoid modifying registers returned from isel functions
922 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
923 421570 std_mutex fails on Arm v8.1 h/w
924 434035 vgdb might crash if valgrind is killed
925 n-i-bz Fix minor one time leaks in dhat.
926 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
927 n-i-bz Add support for the Linux io_uring system calls
928 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
929 n-i-bz Fix non-glibc build of test suite with s390x_features
930 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
931 423195 PPC ISA 3.1 support is missing, part 1
933 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
934 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
935 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
939 Release 3.15.0 (12 April 2019)
940 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
942 3.15.0 is a feature release with many improvements and the usual collection of
945 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
946 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
947 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
948 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
949 support for X86/macOS 10.13 and AMD64/macOS 10.13.
951 * ==================== CORE CHANGES ===================
953 * The XTree Massif output format now makes use of the information obtained
954 when specifying --read-inline-info=yes.
956 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
958 * ==================== TOOL CHANGES ====================
962 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
963 it has been promoted from an experimental tool to a regular tool. Run it
964 with --tool=dhat instead of --tool=exp-dhat.
966 - DHAT now prints only minimal data when the program ends, instead writing
967 the bulk of the profiling data to a file. As a result, the --show-top-n
968 and --sort-by options have been removed.
970 - Profile results can be viewed with the new viewer, dh_view.html. When
971 a run ends, a short message is printed, explaining how to view the result.
973 - See the documentation for more details.
977 - cg_annotate has a new option, --show-percs, which prints percentages next
982 - callgrind_annotate has a new option, --show-percs, which prints percentages
983 next to all event counts.
985 - callgrind_annotate now inserts commas in call counts, and
986 sort the caller/callee lists in the call tree.
990 - The default value for --read-inline-info is now "yes" on
991 Linux/Android/Solaris. It is still "no" on other OS.
995 - The option --xtree-leak=yes (to output leak result in xtree format)
996 automatically activates the option --show-leak-kinds=all, as xtree
997 visualisation tools such as kcachegrind can in any case select what kind
998 of leak to visualise.
1000 - There has been further work to avoid false positives. In particular,
1001 integer equality on partially defined inputs (C == and !=) is now handled
1004 * ==================== OTHER CHANGES ====================
1006 * The new option --show-error-list=no|yes displays, at the end of the run, the
1007 list of detected errors and the used suppressions. Prior to this change,
1008 showing this information could only be done by specifying "-v -v", but that
1009 also produced a lot of other possibly-non-useful messages. The option -s is
1010 equivalent to --show-error-list=yes.
1012 * ==================== FIXED BUGS ====================
1014 The following bugs have been fixed or resolved. Note that "n-i-bz"
1015 stands for "not in bugzilla" -- that is, a bug that was reported to us
1016 but never got a bugzilla entry. We encourage you to file bugs in
1017 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1018 than mailing the developers (or mailing lists) directly -- bugs that
1019 are not entered into bugzilla tend to get forgotten about or ignored.
1021 To see details of a given bug, visit
1022 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1023 where XXXXXX is the bug number as listed below.
1025 385411 s390x: z13 vector floating-point instructions not implemented
1026 397187 z13 vector register support for vgdb gdbserver
1027 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
1028 398870 Please add support for instruction vcvtps2ph
1029 399287 amd64 front end: Illegal Instruction vcmptrueps
1030 399301 Use inlined frames in Massif XTree output.
1031 399322 Improve callgrind_annotate output
1032 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
1033 400164 helgrind test encounters mips x-compiler warnings and assembler error
1034 400490 s390x: VRs allocated as if separate from FPRs
1035 400491 s390x: Operand of LOCH treated as unsigned integer
1036 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
1037 options, which specify a mips64 processor
1038 401112 LLVM 5.0 generates comparison against partially initialized data
1039 401277 More bugs in z13 support
1040 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
1041 401578 drd: crashes sometimes on fork()
1042 401627 memcheck errors with glibc avx2 optimized wcsncmp
1043 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
1044 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
1045 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
1047 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
1048 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
1049 402123 invalid assembler opcodes for mips32r2
1050 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
1051 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
1052 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
1053 402351 mips64 libvexmultiarch_test fails on s390x
1054 402369 Overhaul DHAT
1055 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
1056 402480 Do not use %rsp in clobber list
1057 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
1058 402515 Implement new option --show-error-list=no|yes / -s
1059 402519 POWER 3.0 addex instruction incorrectly implemented
1060 402781 Redo the cache used to process indirect branch targets
1061 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
1062 403552 s390x: wrong facility bit checked for vector facility
1063 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
1064 404638 Add VG_(replaceIndexXA)
1065 404843 s390x: backtrace sometimes ends prematurely
1066 404888 autotools cleanup series
1067 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
1068 405182 Valgrind fails to build with Clang
1069 405205 filter_libc: remove the line holding the futex syscall error entirely
1070 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
1071 the upper and lower 32-bits of the 64-bit result
1072 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
1073 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
1074 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
1076 405403 s390x disassembler cannot be used on x86
1077 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
1078 405458 MIPS mkFormVEC arguments swapped?
1079 405716 drd: Fix an integer overflow in the stack margin calculation
1080 405722 Support arm64 core dump
1081 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
1082 of the 64-bit destination field.
1083 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
1084 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
1085 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
1087 406256 PPC64, vector floating point instructions don't handle subnormal
1088 according to VSCR[NJ] bit setting.
1089 406352 cachegrind/callgrind fails ann tests because of missing a.c
1090 406354 dhat is broken on x86 (32bit)
1091 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
1092 406357 gdbserver_tests fails because of gdb output change
1093 406360 memcheck/tests/libstdc++.supp needs more supression variants
1094 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
1095 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
1096 407340 PPC64, does not support the vlogefp, vexptefp instructions.
1097 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
1098 n-i-bz Fix callgrind_annotate non deterministic order for equal total
1099 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
1100 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
1101 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
1103 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
1104 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
1105 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
1109 Release 3.14.0 (9 October 2018)
1110 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1112 3.14.0 is a feature release with many improvements and the usual collection of
1115 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1116 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1117 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1118 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
1119 support for X86/macOS 10.13, AMD64/macOS 10.13.
1121 * ==================== CORE CHANGES ===================
1123 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
1124 debug info for unloaded code. This allows saved stack traces (e.g. for
1125 memory leaks) to include file/line info for code that has been dlclose'd (or
1126 similar). See the user manual for more information and known limitations.
1128 * Ability to specify suppressions based on source file name and line number.
1130 * Majorly overhauled register allocator. No end-user changes, but the JIT
1131 generates code a bit more quickly now.
1133 * ================== PLATFORM CHANGES =================
1135 * Preliminary support for macOS 10.13 has been added.
1137 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
1139 * mips: support for MIPS SIMD architecture (MSA) has been added.
1141 * mips: support for MIPS N32 ABI has been added.
1143 * s390: partial support for vector instructions (integer and string) has been
1146 * ==================== TOOL CHANGES ====================
1148 * Helgrind: Addition of a flag
1149 --delta-stacktrace=no|yes [yes on linux amd64/x86]
1150 which specifies how full history stack traces should be computed.
1151 Setting this to =yes can speed up Helgrind by 25% when using
1152 --history-level=full.
1154 * Memcheck: reduced false positive rate for optimised code created by Clang 6
1155 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
1156 blocks more carefully to determine where it can avoid expensive definedness
1157 checks without loss of precision. This is controlled by the flag
1158 --expensive-definedness-checks=no|auto|yes [auto].
1160 * ==================== OTHER CHANGES ====================
1162 * Valgrind is now buildable with link-time optimisation (LTO). A new
1163 configure option --enable-lto=yes allows building Valgrind with LTO. If the
1164 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
1165 Note that if you are doing Valgrind development, --enable-lto=yes massively
1166 slows down the build process.
1168 * ==================== FIXED BUGS ====================
1170 The following bugs have been fixed or resolved. Note that "n-i-bz"
1171 stands for "not in bugzilla" -- that is, a bug that was reported to us
1172 but never got a bugzilla entry. We encourage you to file bugs in
1173 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1174 than mailing the developers (or mailing lists) directly -- bugs that
1175 are not entered into bugzilla tend to get forgotten about or ignored.
1177 To see details of a given bug, visit
1178 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1179 where XXXXXX is the bug number as listed below.
1181 79362 Debug info is lost for .so files when they are dlclose'd
1182 208052 strlcpy error when n = 0
1183 255603 exp-sgcheck Assertion '!already_present' failed
1184 338252 building valgrind with -flto (link time optimisation) fails
1185 345763 MIPS N32 ABI support
1186 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
1187 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
1188 372347 Replacement problem of the additional c++14/c++17 new/delete operators
1189 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1190 376257 helgrind history full speed up using a cached stack
1191 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
1193 379748 Fix missing pselect syscall (OS X 10.11)
1194 379754 Fix missing syscall ulock_wait (OS X 10.12)
1195 380397 s390x: __GI_strcspn() replacemenet needed
1196 381162 possible array overrun in VEX register allocator
1197 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
1198 381274 powerpc too chatty even with --sigill-diagnostics=no
1199 381289 epoll_pwait can have a NULL sigmask
1200 381553 VEX register allocator v3
1201 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
1202 381769 Use ucontext_t instead of struct ucontext
1203 381805 arm32 needs ld.so index hardwire for new glibc security fixes
1204 382256 gz compiler flag test doesn't work for gold
1205 382407 vg_perf needs "--terse" command line option
1206 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
1207 382563 MIPS MSA ASE support
1208 382998 xml-socket doesn't work
1209 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
1210 383723 Fix missing kevent_qos syscall (macOS 10.11)
1211 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
1212 384096 Mention AddrCheck at Memcheck's command line option [..]
1213 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
1214 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
1215 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
1216 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
1217 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
1218 384337 performance improvements to VEX register allocator v2 and v3
1219 384526 reduce number of spill insns generated by VEX register allocator v3
1220 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
1221 384631 Sanitise client args as printed with -v
1222 384633 Add a simple progress-reporting facility
1223 384987 VEX regalloc: allocate caller-save registers for short lived vregs
1224 385055 PPC VEX temporary storage exhausted
1225 385182 PPC64 is missing support for the DSCR
1226 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
1227 385207 PPC64, generate_store_FPRF() generates too many Iops
1228 385208 PPC64, xxperm instruction exhausts temporary memory
1229 385210 PPC64, vpermr instruction could exhaust temporary memory
1230 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
1231 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
1232 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
1233 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
1234 385408 s390x: z13 vector "support" instructions not implemented
1235 385409 s390x: z13 vector integer instructions not implemented
1236 385410 s390x: z13 vector string instructions not implemented
1237 385412 s390x: new non-vector z13 instructions not implemented
1238 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
1239 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
1240 385939 Optionally exit on the first error
1241 386318 valgrind.org/info/tools.html is missing SGCheck
1242 386425 running valgrind + wine on armv7l gives illegal opcode
1243 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
1244 387410 MIPSr6 support
1245 387664 Memcheck: make expensive-definedness-checks be the default
1246 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
1247 387766 asm shifts cause false positive "Conditional jump or move depends
1248 on uninitialised value"
1249 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
1250 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
1251 388786 Support bpf syscall in amd64 Linux
1252 388862 Add replacements for wmemchr and wcsnlen on Linux
1253 389065 valgrind meets gcc flag -Wlogical-op
1254 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
1255 390471 suppression by specification of source-file line number
1256 390723 make xtree dump files world wide readable, similar to log files
1257 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
1258 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
1259 392118 unhandled amd64-linux syscall: 332 (statx)
1260 392449 callgrind not clearing the number of calls properly
1261 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
1262 lxssp, stxssp and stxvl instructions.
1263 393023 callgrind_control risks using the wrong vgdb
1264 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
1265 393099 posix_memalign() invalid write if alignment == 0
1266 393146 failing assert "is_DebugInfo_active(di)"
1267 395709 PPC64 is missing support for the xvnegsp instruction
1268 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
1270 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
1271 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
1272 396839 s390x: Trap instructions not implemented
1273 396887 arch_prctl should return EINVAL on unknown option
1274 == 397286 crash before launching binary (Unsupported arch_prctl option)
1275 == 397393 valgrind: the 'impossible' happened: (Archlinux)
1276 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
1277 396906 compile tests failure on mips32-linux: broken inline asm in tests on
1279 397012 glibc ld.so uses arch_prctl on i386
1280 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
1281 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
1282 397424 glibc 2.27 and gdb_server tests
1283 398028 Assertion `cfsi_fits` failing in simple C program
1284 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
1286 n-i-bz Fix missing workq_ops operations (macOS)
1287 n-i-bz fix bug in strspn replacement
1288 n-i-bz Add support for the Linux BLKFLSBUF ioctl
1289 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
1290 n-i-bz Fix possible stack trashing by semctl syscall wrapping
1291 n-i-bz Add support for the Linux membarrier() system call
1292 n-i-bz x86 front end: recognise and handle UD2 correctly
1293 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
1294 correctly aligned before entering the handler.
1296 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
1297 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
1298 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
1302 Release 3.13.0 (15 June 2017)
1303 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1305 3.13.0 is a feature release with many improvements and the usual collection of
1308 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1309 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1310 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1311 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
1313 * ==================== CORE CHANGES ===================
1315 * The translation cache size has been increased to keep up with the demands of
1316 large applications. The maximum number of sectors has increased from 24 to
1317 48. The default number of sectors has increased from 16 to 32 on all
1318 targets except Android, where the increase is from 6 to 12.
1320 * The amount of memory that Valgrind can use has been increased from 64GB to
1321 128GB. In particular this means your application can allocate up to about
1322 60GB when running on Memcheck.
1324 * Valgrind's default load address has been changed from 0x3800'0000 to
1325 0x5800'0000, so as to make it possible to load larger executables. This
1326 should make it possible to load executables of size at least 1200MB.
1328 * A massive spaceleak caused by reading compressed debuginfo files has been
1329 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
1332 * The C++ demangler has been updated.
1334 * Support for demangling Rust symbols has been added.
1336 * A new representation of stack traces, the "XTree", has been added. An XTree
1337 is a tree of stacktraces with data associated with the stacktraces. This is
1338 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
1339 consumption of your program. Reporting is controlled by the new options
1340 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
1342 A report can also be produced on demand using the gdbserver monitor command
1343 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
1344 format' and 'massif format. The existing visualisers for these formats (e.g.
1345 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
1346 analyse these reports.
1348 Memcheck can also produce XTree leak reports using the Callgrind file
1349 format. For more details, see the user manual.
1351 * ================== PLATFORM CHANGES =================
1353 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
1355 * amd64: fixes for JIT failure problems on long AVX2 code blocks
1357 * amd64 and x86: support for CET prefixes has been added
1359 * arm32: a few missing ARMv8 instructions have been implemented
1361 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
1362 Store-Conditional instructions has been added. This is to deal with
1363 processor implementations that implement the LL/SC specifications strictly
1364 and as a result cause Valgrind to hang in certain situations. The
1365 alternative implementation is automatically enabled at startup, as required.
1366 You can use the option --sim-hints=fallback-llsc to force-enable it if you
1369 * Support for OSX 10.12 has been improved.
1371 * On Linux, clone handling has been improved to honour CLONE_VFORK that
1372 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
1373 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
1374 CLONE_VM exact semantics will (still) not work.
1376 * The TileGX/Linux port has been removed because it appears to be both unused
1379 * ==================== TOOL CHANGES ====================
1383 - Memcheck should give fewer false positives when running optimised
1384 Clang/LLVM generated code.
1386 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1388 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
1389 to produce the end of execution leak report in a xtree callgrind format
1392 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
1393 the leak report in an xtree file.
1397 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1399 - For some workloads (typically, for big applications), Massif memory
1400 consumption and CPU consumption has decreased significantly.
1404 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1406 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
1407 for Ada gnat compiled applications.
1409 * ==================== OTHER CHANGES ====================
1411 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
1412 append the inner guest stacktrace to the inner host stacktrace. This helps
1413 to investigate the errors reported by the outer, when they are caused by the
1414 inner guest program (such as an inner regtest). See README_DEVELOPERS for
1417 * To allow fast detection of callgrind files by desktop environments and file
1418 managers, the format was extended to have an optional first line that
1419 uniquely identifies the format ("# callgrind format"). Callgrind creates
1420 this line now, as does the new xtree functionality.
1422 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
1423 have a new %n format letter that is replaced by a sequence number.
1425 * "--version -v" now shows the SVN revision numbers from which Valgrind was
1428 * ==================== FIXED BUGS ====================
1430 The following bugs have been fixed or resolved. Note that "n-i-bz"
1431 stands for "not in bugzilla" -- that is, a bug that was reported to us
1432 but never got a bugzilla entry. We encourage you to file bugs in
1433 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1434 than mailing the developers (or mailing lists) directly -- bugs that
1435 are not entered into bugzilla tend to get forgotten about or ignored.
1437 To see details of a given bug, visit
1438 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1439 where XXXXXX is the bug number as listed below.
1441 162848 --log-file output isn't split when a program forks
1442 340777 Illegal instruction on mips (ar71xx)
1443 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
1444 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
1445 to a different stack.
1446 344139 x86 stack-seg overrides, needed by the Wine people
1447 344524 store conditional of guest applications always fail - observed on
1449 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
1450 352395 Please provide SVN revision info in --version -v
1451 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
1452 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
1453 INVALID_POSIX_THREADID' failed
1454 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
1455 barrier implementation
1456 358697 valgrind.h: Some code remains even when defining NVALGRIND
1457 359202 Add musl libc configure/compile
1458 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
1459 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
1460 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
1461 362223 assertion failed when .valgrindrc is a directory instead of a file
1462 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
1463 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
1464 368507 can't malloc chunks larger than about 34GB
1465 368529 Android arm target link error, missing atexit and pthread_atfork
1466 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
1467 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
1468 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
1469 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
1470 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
1471 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
1472 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
1473 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
1474 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
1475 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
1476 370028 Reduce the number of compiler warnings on MIPS platforms
1477 370635 arm64 missing syscall getcpu
1478 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
1479 371227 Clean AArch64 syscall table
1480 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
1481 371471 Valgrind complains about non legit memory leaks on placement new (C++)
1482 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
1483 371503 disInstr(arm64): unhandled instruction 0xF89F0000
1484 371869 support '%' in symbol Z-encoding
1485 371916 execution tree xtree concept
1486 372120 c++ demangler demangles symbols which are not c++
1487 372185 Support of valgrind on ARMv8 with 32 bit executable
1488 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1489 372195 Power PC, xxsel instruction is not always recognized.
1490 372504 Hanging on exit_group
1491 372600 process loops forever when fatal signals are arriving quickly
1492 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1493 373046 Stacks registered by core are never deregistered
1494 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1495 373086 Implement additional Xen hypercalls
1496 373192 Calling posix_spawn in glibc 2.24 completely broken
1497 373488 Support for fanotify API on ARM64 architecture
1498 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1499 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1500 373938 const IRExpr arguments for matchIRExpr()
1501 374719 some spelling fixes
1502 374963 increase valgrind's load address to prevent mmap failure
1503 375514 valgrind_get_tls_addr() does not work in case of static TLS
1504 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1505 for ML_(find_rx_mapping)()
1506 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1507 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1508 == 377159 "vex: the `impossible' happened" still present
1509 == 375150 Assertion 'tres.status == VexTransOK' failed
1510 == 378068 valgrind crashes on AVX2 function in FFmpeg
1511 376142 Segfaults on MIPS Cavium Octeon boards
1512 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1513 376455 Solaris: unhandled syscall lgrpsys(180)
1514 376518 Solaris: unhandled fast trap getlgrp(6)
1515 376611 ppc64 and arm64 don't know about prlimit64 syscall
1516 376729 PPC64, remove R2 from the clobber list
1518 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1519 to be wrongly marked as addressable
1520 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1521 PIE enabled by default
1522 377376 memcheck/tests/linux/getregset fails with glibc2.24
1523 377427 PPC64, lxv instruction failing on odd destination register
1524 377478 PPC64: ISA 3.0 setup fixes
1525 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1526 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1527 and 2 args for FUTEX_TRYLOCK_PI
1528 377717 Fix massive space leak when reading compressed debuginfo sections
1529 377891 Update Xen 4.6 domctl wrappers
1530 377930 fcntl syscall wrapper is missing flock structure check
1531 378524 libvexmultiarch_test regression on s390x and ppc64
1532 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1533 378673 Update libiberty demangler
1534 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1535 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1536 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1537 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1538 (task_register_dyld_image_infos)
1539 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1540 (task_register_dyld_shared_cache_image_info)
1541 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1542 379473 MIPS: add support for rdhwr cycle counter register
1543 379504 remove TileGX/Linux port
1544 379525 Support more x86 nop opcodes
1545 379838 disAMode(x86): not an addr!
1546 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1547 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1548 379895 clock_gettime does not execute POST syscall wrapper
1549 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1550 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1551 380200 xtree generated callgrind files refer to files without directory name
1552 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1553 380397 s390x: __GI_strcspn() replacement needed
1554 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1556 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1557 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1558 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1562 Release 3.12.0 (20 October 2016)
1563 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1565 3.12.0 is a feature release with many improvements and the usual
1566 collection of bug fixes.
1568 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1569 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1570 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1571 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1572 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1573 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1575 * ================== PLATFORM CHANGES =================
1577 * POWER: Support for ISA 3.0 has been added
1579 * mips: support for O32 FPXX ABI has been added.
1580 * mips: improved recognition of different processors
1581 * mips: determination of page size now done at run time
1583 * amd64: Partial support for AMD FMA4 instructions.
1585 * arm, arm64: Support for v8 crypto and CRC instructions.
1587 * Improvements and robustification of the Solaris port.
1589 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1591 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1592 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1593 where possible. Valgrind's support for 32-bit x86 has stagnated in
1594 recent years and has fallen far behind that for 64-bit x86
1595 instructions. By contrast 64-bit x86 is well supported, up to and
1598 * ==================== TOOL CHANGES ====================
1602 - Added meta mempool support for describing a custom allocator which:
1603 - Auto-frees all chunks assuming that destroying a pool destroys all
1605 - Uses itself to allocate other memory blocks
1607 - New flag --ignore-range-below-sp to ignore memory accesses below
1608 the stack pointer, if you really have to. The related flag
1609 --workaround-gcc296-bugs=yes is now deprecated. Use
1610 --ignore-range-below-sp=1024-1 as a replacement.
1614 - Improved thread startup time significantly on non-Linux platforms.
1618 - Added collection of the metric "tot-blocks-allocd"
1620 * ==================== OTHER CHANGES ====================
1622 * Replacement/wrapping of malloc/new related functions is now done not just
1623 for system libraries by default, but for any globally defined malloc/new
1624 related function (both in shared libraries and statically linked alternative
1625 malloc implementations). The dynamic (runtime) linker is excluded, though.
1626 To only intercept malloc/new related functions in
1627 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1628 "nouserintercepts" can be any non-existing library name).
1629 This new functionality is not implemented for MacOS X.
1631 * The maximum number of callers in a suppression entry is now equal to
1632 the maximum size for --num-callers (500).
1633 Note that --gen-suppressions=yes|all similarly generates suppressions
1634 containing up to --num-callers frames.
1636 * New and modified GDB server monitor features:
1638 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1639 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1642 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1643 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1646 * Valgrind is able to read compressed debuginfo sections in two formats:
1647 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1648 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1650 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1651 for the most common use case (x86_64-linux, Memcheck) has been
1654 * Improved performance for programs that do a lot of discarding of
1655 instruction address ranges of 8KB or less.
1657 * The C++ symbol demangler has been updated.
1659 * More robustness against invalid syscall parameters on Linux.
1661 * ==================== FIXED BUGS ====================
1663 The following bugs have been fixed or resolved. Note that "n-i-bz"
1664 stands for "not in bugzilla" -- that is, a bug that was reported to us
1665 but never got a bugzilla entry. We encourage you to file bugs in
1666 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1667 than mailing the developers (or mailing lists) directly -- bugs that
1668 are not entered into bugzilla tend to get forgotten about or ignored.
1670 To see details of a given bug, visit
1671 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1672 where XXXXXX is the bug number as listed below.
1674 191069 Exiting due to signal not reported in XML output
1675 199468 Suppressions: stack size limited to 25
1676 while --num-callers allows more frames
1677 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1678 278744 cvtps2pd with redundant RexW
1679 303877 valgrind doesn't support compressed debuginfo sections.
1680 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1681 348345 Assertion fails for negative lineno
1682 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1683 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1684 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1685 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1686 352197 mips: mmap2() not wrapped correctly for page size > 4096
1687 353083 arm64 doesn't implement various xattr system calls
1688 353084 arm64 doesn't support sigpending system call
1689 353137 www: update info for Supported Platforms
1690 353138 www: update "The Valgrind Developers" page
1691 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1694 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1695 353398 WARNING: unhandled amd64-solaris syscall: 207
1696 353660 XML in auxwhat tag not escaping reserved symbols properly
1697 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1698 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1699 353802 ELF debug info reader confused with multiple .rodata sections
1700 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1701 353917 unhandled amd64-solaris syscall fchdir(120)
1702 353920 unhandled amd64-solaris syscall: 170
1703 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1704 354392 unhandled amd64-solaris syscall: 171
1705 354797 Vbit test does not include Iops for Power 8 instruction support
1706 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1710 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1711 355188 valgrind should intercept all malloc related global functions
1712 355454 do not intercept malloc related symbols from the runtime linker
1713 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1714 356044 Dwarf line info reader misinterprets is_stmt register
1715 356112 mips: replace addi with addiu
1716 356393 valgrind (vex) crashes because isZeroU happened
1719 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1720 356678 arm64-linux: unhandled syscall 232 (mincore)
1721 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1722 356823 Unsupported ARM instruction: stlex
1723 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1724 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1725 357673 crash if I try to run valgrind with a binary link with libcurl
1726 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1727 357871 pthread_spin_destroy not properly wrapped
1728 357887 Calls to VG_(fclose) do not close the file descriptor
1729 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1730 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1731 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1732 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1733 359181 Buffer Overflow during Demangling
1734 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1735 359289 s390x: popcnt (B9E1) not implemented
1736 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1737 359503 Add missing syscalls for aarch64 (arm64)
1738 359645 "You need libc6-dbg" help message could be more helpful
1739 359703 s390: wire up separate socketcalls system calls
1740 359724 getsockname might crash - deref_UInt should call safe_to_deref
1741 359733 amd64 implement ld.so strchr/index override like x86
1742 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1743 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1745 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1746 359871 Incorrect mask handling in ppoll
1747 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1748 360008 Contents of Power vr registers contents is not printed correctly when
1749 the --vgdb-shadow-registers=yes option is used
1750 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1751 non-zero shadow bits
1752 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1753 360425 arm64 unsupported instruction ldpsw
1755 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1756 360571 Error about the Android Runtime reading below the stack pointer on ARM
1757 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1758 360749 kludge for multiple .rodata sections on Solaris no longer needed
1759 360752 raise the number of reserved fds in m_main.c from 10 to 12
1760 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1761 361226 s390x: risbgn (EC59) not implemented
1762 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1763 361354 ppc64[le]: wire up separate socketcalls system calls
1764 361615 Inconsistent termination for multithreaded process terminated by signal
1765 361926 Unhandled Solaris syscall: sysfs(84)
1766 362009 V dumps core on unimplemented functionality before threads are created
1767 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1768 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1769 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1770 362953 Request for an update to the Valgrind Developers page
1771 363680 add renameat2() support
1772 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1773 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1774 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1775 364058 clarify in manual limitations of array overruns detections
1776 364413 pselect sycallwrapper mishandles NULL sigmask
1777 364728 Power PC, missing support for several HW registers in
1778 get_otrack_shadow_offset_wrk()
1779 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1780 365273 Invalid write to stack location reported after signal handler runs
1781 365912 ppc64BE segfault during jm-insns test (RELRO)
1782 366079 FPXX Support for MIPS32 Valgrind
1783 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1784 366344 Multiple unhandled instruction for Aarch64
1785 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1786 367995 Integration of memcheck with custom memory allocator
1787 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1788 368412 False positive result for altivec capability check
1789 368416 Add tc06_two_races_xml.exp output for ppc64
1790 368419 Perf Events ioctls not implemented
1791 368461 mmapunmap test fails on ppc64
1792 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1793 369000 AMD64 fma4 instructions unsupported.
1794 369169 ppc64 fails jm_int_isa_2_07 test
1795 369175 jm_vec_isa_2_07 test crashes on ppc64
1796 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1797 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1798 369359 msghdr_foreachfield can crash when handling bad iovec
1799 369360 Bad sigprocmask old or new sets can crash valgrind
1800 369361 vmsplice syscall wrapper crashes on bad iovec
1801 369362 Bad sigaction arguments crash valgrind
1802 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1803 369402 Bad set/get_thread_area pointer crashes valgrind
1804 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1805 369446 valgrind crashes on unknown fcntl command
1806 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1807 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1808 370265 ISA 3.0 HW cap stuff needs updating
1809 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1810 372195 Power PC, xxsel instruction is not always recognized
1812 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1813 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1814 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1815 n-i-bz Fix ppoll_alarm exclusion on OS X
1816 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1817 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1818 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1819 n-i-bz amd64: memcheck false positive with shr %edx
1820 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1821 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1822 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1823 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1825 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1826 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1827 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1831 Release 3.11.0 (22 September 2015)
1832 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1834 3.11.0 is a feature release with many improvements and the usual
1835 collection of bug fixes.
1837 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1838 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1839 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1840 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1841 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1842 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1844 * ================== PLATFORM CHANGES =================
1846 * Support for Solaris/x86 and Solaris/amd64 has been added.
1848 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1850 * Preliminary support for the Tilera TileGX architecture has been added.
1852 * s390x: It is now required for the host to have the "long displacement"
1853 facility. The oldest supported machine model is z990.
1855 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1856 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1858 * The JIT's register allocator is significantly faster, making the JIT
1859 as a whole somewhat faster, so JIT-intensive activities, for example
1860 program startup, are modestly faster, around 5%.
1862 * There have been changes to the default settings of several command
1863 line flags, as detailed below.
1865 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1866 capable hosts, the simulated CPUID will now indicate AVX2 support.
1868 * ==================== TOOL CHANGES ====================
1872 - The default value for --leak-check-heuristics has been changed from
1873 "none" to "all". This helps to reduce the number of possibly
1874 lost blocks, in particular for C++ applications.
1876 - The default value for --keep-stacktraces has been changed from
1877 "malloc-then-free" to "malloc-and-free". This has a small cost in
1878 memory (one word per malloc-ed block) but allows Memcheck to show the
1879 3 stacktraces of a dangling reference: where the block was allocated,
1880 where it was freed, and where it is acccessed after being freed.
1882 - The default value for --partial-loads-ok has been changed from "no" to
1883 "yes", so as to avoid false positive errors resulting from some kinds
1884 of vectorised loops.
1886 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1887 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1888 than get_vbits when you need to associate byte data value with
1889 their corresponding validity bits.
1891 - The 'block_list' monitor command has been enhanced:
1892 o it can print a range of loss records
1893 o it now accepts an optional argument 'limited <max_blocks>'
1894 to control the number of blocks printed.
1895 o if a block has been found using a heuristic, then
1896 'block_list' now shows the heuristic after the block size.
1897 o the loss records/blocks to print can be limited to the blocks
1898 found via specified heuristics.
1900 - The C helper functions used to instrument loads on
1901 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1902 replaced by handwritten assembly sequences. This gives speedups
1903 in the region of 0% to 7% for those targets only.
1905 - A new command line option, --expensive-definedness-checks=yes|no,
1906 has been added. This is useful for avoiding occasional invalid
1907 uninitialised-value errors in optimised code. Watch out for
1908 runtime degradation, as this can be up to 25%. As always, though,
1909 the slowdown is highly application specific. The default setting
1914 - A new monitor command 'all_snapshots <filename>' dumps all
1915 snapshots taken so far.
1919 - Significant memory reduction and moderate speedups for
1920 --history-level=full for applications accessing a lot of memory
1921 with many different stacktraces.
1923 - The default value for --conflict-cache-size=N has been doubled to
1924 2000000. Users that were not using the default value should
1925 preferably also double the value they give.
1927 The default was changed due to the changes in the "full history"
1928 implementation. Doubling the value gives on average a slightly more
1929 complete history and uses similar memory (or significantly less memory
1930 in the worst case) than the previous implementation.
1932 - The Helgrind monitor command 'info locks' now accepts an optional
1933 argument 'lock_addr', which shows information about the lock at the
1936 - When using --history-level=full, the new Helgrind monitor command
1937 'accesshistory <addr> [<len>]' will show the recorded accesses for
1938 <len> (or 1) bytes at <addr>.
1940 * ==================== OTHER CHANGES ====================
1942 * The default value for the --smc-check option has been changed from
1943 "stack" to "all-non-file" on targets that provide automatic D-I
1944 cache coherence (x86, amd64 and s390x). The result is to provide,
1945 by default, transparent support for JIT generated and self-modifying
1946 code on all targets.
1948 * Mac OS X only: the default value for the --dsymutil option has been
1949 changed from "no" to "yes", since any serious usage on Mac OS X
1950 always required it to be "yes".
1952 * The command line options --db-attach and --db-command have been removed.
1953 They were deprecated in 3.10.0.
1955 * When a process dies due to a signal, Valgrind now shows the signal
1956 and the stacktrace at default verbosity (i.e. verbosity 1).
1958 * The address description logic used by Memcheck and Helgrind now
1959 describes addresses in anonymous segments, file mmap-ed segments,
1960 shared memory segments and the brk data segment.
1962 * The new option --error-markers=<begin>,<end> can be used to mark the
1963 begin/end of errors in textual output mode, to facilitate
1964 searching/extracting errors in output files that mix valgrind errors
1965 with program output.
1967 * The new option --max-threads=<number> can be used to change the number
1968 of threads valgrind can handle. The default is 500 threads which
1969 should be more than enough for most applications.
1971 * The new option --valgrind-stacksize=<number> can be used to change the
1972 size of the private thread stacks used by Valgrind. This is useful
1973 for reducing memory use or increasing the stack size if Valgrind
1974 segfaults due to stack overflow.
1976 * The new option --avg-transtab-entry-size=<number> can be used to specify
1977 the expected instrumented block size, either to reduce memory use or
1978 to avoid excessive retranslation.
1980 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1982 * New and modified GDB server monitor features:
1984 - When a signal is reported in GDB, you can now use the GDB convenience
1985 variable $_siginfo to examine detailed signal information.
1987 - Valgrind's gdbserver now allows the user to change the signal
1988 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1989 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1990 continue without passing the signal to the process.
1992 - With GDB >= 7.10, the command 'target remote'
1993 will automatically load the executable file of the process running
1994 under Valgrind. This means you do not need to specify the executable
1995 file yourself, GDB will discover it itself. See GDB documentation about
1996 'qXfer:exec-file:read' packet for more info.
1998 * ==================== FIXED BUGS ====================
2000 The following bugs have been fixed or resolved. Note that "n-i-bz"
2001 stands for "not in bugzilla" -- that is, a bug that was reported to us
2002 but never got a bugzilla entry. We encourage you to file bugs in
2003 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2004 than mailing the developers (or mailing lists) directly -- bugs that
2005 are not entered into bugzilla tend to get forgotten about or ignored.
2007 To see details of a given bug, visit
2008 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2009 where XXXXXX is the bug number as listed below.
2011 116002 VG_(printf): Problems with justification of strings and integers
2012 155125 avoid cutting away file:lineno after long function name
2013 197259 Unsupported arch_prtctl PR_SET_GS option
2014 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
2015 201216 Fix Valgrind does not support pthread_sigmask() on OS X
2016 201435 Fix Darwin: -v does not show kernel version
2017 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
2018 211256 Fixed an outdated comment regarding the default platform.
2019 211529 Incomplete call stacks for code compiled by newer versions of MSVC
2020 211926 Avoid compilation warnings in valgrind.h with -pedantic
2021 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
2023 226609 Crediting upstream authors in man page
2024 231257 Valgrind omits path when executing script from shebang line
2025 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
2026 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
2027 269360 s390x: Fix addressing mode selection for compare-and-swap
2028 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
2030 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
2031 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
2032 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
2033 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
2034 330147 libmpiwrap PMPI_Get_count returns undefined value
2035 333051 mmap of huge pages fails due to incorrect alignment
2037 334802 valgrind does not always explain why a given option is bad
2038 335618 mov.w rN, pc/sp (ARM32)
2039 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
2044 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
2045 338602 AVX2 bit in CPUID missing
2046 338606 Strange message for scripts with invalid interpreter
2047 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
2048 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
2049 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
2051 339156 gdbsrv not called for fatal signal
2052 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
2053 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
2054 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
2055 339442 Fix testsuite build failure on OS X 10.9
2056 339542 Enable compilation with Intel's ICC compiler
2057 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
2058 339688 Mac-specific ASM does not support .version directive (cpuid,
2059 tronical and pushfpopf tests)
2060 339745 Valgrind crash when check Marmalade app (partial fix)
2061 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
2062 339778 Linux/TileGx platform support for Valgrind
2063 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
2064 339789 Fix none/tests/execve test on Mac OS X 10.9
2065 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
2066 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
2067 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
2068 340392 Allow user to select more accurate definedness checking in memcheck
2069 to avoid invalid complaints on optimised code
2070 340430 Fix some grammatical weirdness in the manual.
2071 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
2072 341419 Signal handler ucontext_t not filled out correctly on OS X
2073 341539 VG_(describe_addr) should not describe address as belonging to client
2074 segment if it is past the heap end
2075 341613 Enable building of manythreads and thread-exits tests on Mac OS X
2076 341615 Fix none/tests/darwin/access_extended test on Mac OS X
2077 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
2078 341789 aarch64: shmat fails with valgrind on ARMv8
2079 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
2080 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
2081 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
2082 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
2083 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
2084 342221 socket connect false positive uninit memory for unknown af family
2085 342353 Allow dumping full massif output while valgrind is still running
2086 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
2090 342603 Add I2C_SMBUS ioctl support
2091 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
2092 342683 Mark memory past the initial brk limit as unaddressable
2093 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
2094 342795 Internal glibc __GI_mempcpy call should be intercepted
2095 342841 s390x: Support instructions fiebr(a) and fidbr(a)
2096 343012 Unhandled syscall 319 (memfd_create)
2097 343069 Patch updating v4l2 API support
2098 343173 helgrind crash during stack unwind
2099 343219 fix GET_STARTREGS for arm
2100 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
2101 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
2102 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
2103 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
2104 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
2105 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
2106 343597 ppc64le: incorrect use of offseof macro
2107 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
2108 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
2109 343732 Unhandled syscall 144 (setgid) on aarch64
2110 343733 Unhandled syscall 187 (msgctl and related) on aarch64
2111 343802 s390x: False positive "conditional jump or move depends on [..]
2112 343902 --vgdb=yes doesn't break when --xml=yes is used
2113 343967 Don't warn about setuid/setgid/setcap executable for directories
2114 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
2115 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
2116 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
2117 344054 www - update info for Solaris/illumos
2118 344416 'make regtest' does not work cleanly on OS X
2119 344235 Remove duplicate include of pub_core_aspacemgr.h
2120 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
2121 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
2122 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
2123 344314 callgrind_annotate ... warnings about commands containing newlines
2124 344318 socketcall should wrap recvmmsg and sendmmsg
2125 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
2126 344416 Fix 'make regtest' does not work cleanly on OS X
2127 344499 Fix compilation for Linux kernel >= 4.0.0
2128 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
2129 unix:349 (__pthread_fchdir)
2130 344559 Garbage collection of unused segment names in address space manager
2131 344560 Fix stack traces missing penultimate frame on OS X
2132 344621 Fix memcheck/tests/err_disable4 test on OS X
2133 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
2134 344702 Fix missing libobjc suppressions on OS X 10.10
2136 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
2137 344939 Fix memcheck/tests/xml1 on OS X 10.10
2138 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
2139 345079 Fix build problems in VEX/useful/test_main.c
2140 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
2141 345177 arm64: prfm (reg) not implemented
2142 345215 Performance improvements for the register allocator
2143 345248 add support for Solaris OS in valgrind
2144 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
2145 345394 Fix memcheck/tests/strchr on OS X
2146 345637 Fix memcheck/tests/sendmsg on OS X
2147 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
2148 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
2149 345887 Fix an assertion in the address space manager
2150 345928 amd64: callstack only contains current function for small stacks
2151 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
2152 345987 MIPS64: Implement cavium LHX instruction
2153 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
2154 346185 Fix typo saving altivec register v24
2155 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
2156 and LibVEX_GuestPPC64_get_CR()
2157 346270 Regression tests none/tests/jm_vec/isa_2_07 and
2158 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
2159 346307 fuse filesystem syscall deadlocks
2160 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
2161 346411 MIPS: SysRes::_valEx handling is incorrect
2162 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
2163 346474 PPC64 Power 8, spr TEXASRU register not supported
2164 346487 Compiler generates "note" about a future ABI change for PPC64
2165 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
2166 and causing spurious "invalid read of size 8" warnings
2167 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
2168 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
2169 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
2170 347322 Power PC regression test cleanup
2171 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
2173 347389 unhandled syscall: 373 (Linux ARM syncfs)
2174 347686 Patch set to cleanup PPC64 regtests
2175 347978 Remove bash dependencies where not needed
2176 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
2177 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
2179 348102 Patch updating v4l2 API support
2180 348247 amd64 front end: jno jumps wrongly when overflow is not set
2181 348269 Improve mmap MAP_HUGETLB support.
2182 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
2183 348345 Assertion fails for negative lineno
2184 348377 Unsupported ARM instruction: yield
2185 348565 Fix detection of command line option availability for clang
2186 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
2187 348728 Fix broken check for VIDIOC_G_ENC_INDEX
2188 348748 Fix redundant condition
2189 348890 Fix clang warning about unsupported --param inline-unit-growth=900
2190 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
2191 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
2192 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
2193 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
2194 349626 Implemented additional Xen hypercalls
2195 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
2196 349790 Clean up of the hardware capability checking utilities.
2197 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
2198 349874 Fix typos in source code
2199 349879 memcheck: add handwritten assembly for helperc_LOADV*
2200 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
2201 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
2202 350202 Add limited param to 'monitor block_list'
2203 350290 s390x: Support instructions fixbr(a)
2204 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
2205 350809 Fix none/tests/async-sigs for Solaris
2206 350811 Remove reference to --db-attach which has been removed.
2207 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
2208 350854 hard-to-understand code in VG_(load_ELF)()
2209 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2210 351386 Solaris: Cannot run ld.so.1 under Valgrind
2211 351474 Fix VG_(iseqsigset) as obvious
2212 351531 Typo in /include/vki/vki-xen-physdev.h header guard
2213 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
2214 351858 ldsoexec support on Solaris
2215 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
2216 352130 helgrind reports false races for printfs using mempcpy on FILE* state
2217 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
2218 352320 arm64 crash on none/tests/nestedfs
2219 352765 Vbit test fails on Power 6
2220 352768 The mbar instruction is missing from the Power PC support
2221 352769 Power PC program priority register (PPR) is not supported
2222 n-i-bz Provide implementations of certain compiler builtins to support
2223 compilers that may not provide those
2224 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
2225 n-i-bz Fix compilation on distros with glibc < 2.5
2226 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
2227 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
2228 n-i-bz mremap did not work properly on shared memory
2229 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
2230 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
2232 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
2233 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
2234 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
2238 Release 3.10.1 (25 November 2014)
2239 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2240 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
2241 and backports fixes for all reported missing AArch64 ARMv8 instructions
2242 and syscalls from the trunk. If you package or deliver 3.10.0 for others
2243 to use, you might want to consider upgrading to 3.10.1 instead.
2245 The following bugs have been fixed or resolved. Note that "n-i-bz"
2246 stands for "not in bugzilla" -- that is, a bug that was reported to us
2247 but never got a bugzilla entry. We encourage you to file bugs in
2248 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2249 than mailing the developers (or mailing lists) directly -- bugs that
2250 are not entered into bugzilla tend to get forgotten about or ignored.
2252 To see details of a given bug, visit
2253 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2254 where XXXXXX is the bug number as listed below.
2256 335440 arm64: ld1 (single structure) is not implemented
2257 335713 arm64: unhanded instruction: prfm (immediate)
2258 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
2259 339182 ppc64: AvSplat ought to load destination vector register with [..]
2260 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
2261 339433 ppc64 lxvw4x instruction uses four 32-byte loads
2262 339645 Use correct tag names in sys_getdents/64 wrappers
2263 339706 Fix false positive for ioctl(TIOCSIG) on linux
2264 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
2265 339853 arm64 times syscall unknown
2266 339855 arm64 unhandled getsid/setsid syscalls
2267 339858 arm64 dmb sy not implemented
2268 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
2269 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
2270 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
2272 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
2273 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
2274 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
2275 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
2276 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
2277 340509 arm64: unhandled instruction fcvtas
2278 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
2279 340632 arm64: unhandled instruction fcvtas
2280 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
2281 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
2282 340788 warning: unhandled syscall: 318 (getrandom)
2283 340807 disInstr(arm): unhandled instruction: 0xEE989B20
2284 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
2285 340922 arm64: unhandled getgroups/setgroups syscalls
2286 350251 Fix typo in VEX utility program (test_main.c).
2287 350407 arm64: unhandled instruction ucvtf (vector, integer)
2288 350809 none/tests/async-sigs breaks when run under cron on Solaris
2289 350811 update README.solaris after r15445
2290 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
2291 350854 strange code in VG_(load_ELF)()
2292 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2293 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
2294 n-i-bz Add missing ]] to terminate CDATA.
2295 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
2296 n-i-bz Enable sys_fadvise64_64 on arm32.
2297 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
2298 n-i-bz Add test cases for all known arm64 load/store instructions.
2299 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
2300 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
2301 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
2302 n-i-bz arm64: Implement "fcvtpu w, s".
2303 n-i-bz arm64: implement ADDP and various others
2304 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
2305 n-i-bz arm64: enable FCVT{A,N}S X,S.
2307 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
2311 Release 3.10.0 (10 September 2014)
2312 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2314 3.10.0 is a feature release with many improvements and the usual
2315 collection of bug fixes.
2317 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
2318 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
2319 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
2320 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
2321 significantly improved relative to the 3.9.0 release.
2323 * ================== PLATFORM CHANGES =================
2325 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
2326 is mostly complete, and is usable, but some SIMD instructions are as
2329 * Support for little-endian variant of the 64-bit POWER architecture.
2331 * Support for Android on MIPS32.
2333 * Support for 64bit FPU on MIPS32 platforms.
2335 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
2337 * Configuration for and running on Android targets has changed.
2338 See README.android in the source tree for details.
2340 * ================== DEPRECATED FEATURES =================
2342 * --db-attach is now deprecated and will be removed in the next
2343 valgrind feature release. The built-in GDB server capabilities are
2344 superior and should be used instead. Learn more here:
2345 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
2347 * ==================== TOOL CHANGES ====================
2351 - Client code can now selectively disable and re-enable reporting of
2352 invalid address errors in specific ranges using the new client
2353 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
2354 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
2356 - Leak checker: there is a new leak check heuristic called
2357 "length64". This is used to detect interior pointers pointing 8
2358 bytes inside a block, on the assumption that the first 8 bytes
2359 holds the value "block size - 8". This is used by
2360 sqlite3MemMalloc, for example.
2362 - Checking of system call parameters: if a syscall parameter
2363 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
2364 several fields not initialised, an error is now reported for each
2365 field. Previously, an error was reported only for the first
2366 uninitialised field.
2368 - Mismatched alloc/free checking: a new flag
2369 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
2370 such checks if necessary.
2374 - Improvements to error messages:
2376 o Race condition error message involving heap allocated blocks also
2377 show the thread number that allocated the raced-on block.
2379 o All locks referenced by an error message are now announced.
2380 Previously, some error messages only showed the lock addresses.
2382 o The message indicating where a lock was first observed now also
2383 describes the address/location of the lock.
2385 - Helgrind now understands the Ada task termination rules and
2386 creates a happens-before relationship between a terminated task
2387 and its master. This avoids some false positives and avoids a big
2388 memory leak when a lot of Ada tasks are created and terminated.
2389 The interceptions are only activated with forthcoming releases of
2390 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
2392 - A new GDB server monitor command "info locks" giving the list of
2393 locks, their location, and their status.
2397 - callgrind_control now supports the --vgdb-prefix argument,
2398 which is needed if valgrind was started with this same argument.
2400 * ==================== OTHER CHANGES ====================
2402 * Unwinding through inlined function calls. Stack unwinding can now
2403 make use of Dwarf3 inlined-unwind information if it is available.
2404 The practical effect is that inlined calls become visible in stack
2405 traces. The suppression matching machinery has been adjusted
2406 accordingly. This is controlled by the new option
2407 --read-inline-info=yes|no. Currently this is enabled by default
2408 only on Linux and Android targets and only for the tools Memcheck,
2411 * Valgrind can now read EXIDX unwind information on 32-bit ARM
2412 targets. If an object contains both CFI and EXIDX unwind
2413 information, Valgrind will prefer the CFI over the EXIDX. This
2414 facilitates unwinding through system libraries on arm-android
2417 * Address description logic has been improved and is now common
2418 between Memcheck and Helgrind, resulting in better address
2419 descriptions for some kinds of error messages.
2421 * Error messages about dubious arguments (eg, to malloc or calloc) are
2422 output like other errors. This means that they can be suppressed
2423 and they have a stack trace.
2425 * The C++ demangler has been updated for better C++11 support.
2427 * New and modified GDB server monitor features:
2429 - Thread local variables/storage (__thread) can now be displayed.
2431 - The GDB server monitor command "v.info location <address>"
2432 displays information about an address. The information produced
2433 depends on the tool and on the options given to valgrind.
2434 Possibly, the following are described: global variables, local
2435 (stack) variables, allocated or freed blocks, ...
2437 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
2438 ask the GDB server to stop at the start of program execution, at
2439 the end of the program execution and on Valgrind internal errors.
2441 - A new monitor command "v.info stats" shows various Valgrind core
2442 and tool statistics.
2444 - A new monitor command "v.set hostvisibility" allows the GDB server
2445 to provide access to Valgrind internal host status/memory.
2447 * A new option "--aspace-minaddr=<address>" can in some situations
2448 allow the use of more memory by decreasing the address above which
2449 Valgrind maps memory. It can also be used to solve address
2450 conflicts with system libraries by increasing the default value.
2451 See user manual for details.
2453 * The amount of memory used by Valgrind to store debug info (unwind
2454 info, line number information and symbol data) has been
2455 significantly reduced, even though Valgrind now reads more
2456 information in order to support unwinding of inlined function calls.
2458 * Dwarf3 handling with --read-var-info=yes has been improved:
2460 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
2462 - Code compiled with
2463 -ffunction-sections -fdata-sections -Wl,--gc-sections
2464 no longer causes assertion failures.
2466 * Improved checking for the --sim-hints= and --kernel-variant=
2467 options. Unknown strings are now detected and reported to the user
2470 * The semantics of stack start/end boundaries in the valgrind.h
2471 VALGRIND_STACK_REGISTER client request has been clarified and
2472 documented. The convention is that start and end are respectively
2473 the lowest and highest addressable bytes of the stack.
2475 * ==================== FIXED BUGS ====================
2477 The following bugs have been fixed or resolved. Note that "n-i-bz"
2478 stands for "not in bugzilla" -- that is, a bug that was reported to us
2479 but never got a bugzilla entry. We encourage you to file bugs in
2480 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2481 than mailing the developers (or mailing lists) directly -- bugs that
2482 are not entered into bugzilla tend to get forgotten about or ignored.
2484 To see details of a given bug, visit
2485 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2486 where XXXXXX is the bug number as listed below.
2488 175819 Support for ipv6 socket reporting with --track-fds
2489 232510 make distcheck fails
2490 249435 Analyzing wine programs with callgrind triggers a crash
2491 278972 support for inlined function calls in stacktraces and suppression
2493 291310 FXSAVE instruction marks memory as undefined on amd64
2494 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2495 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2496 315199 vgcore file for threaded app does not show which thread crashed
2497 315952 tun/tap ioctls are not supported
2498 323178 Unhandled instruction: PLDW register (ARM)
2499 323179 Unhandled instruction: PLDW immediate (ARM)
2500 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2501 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2502 325124 [MIPSEL] Compilation error
2503 325477 Phase 4 support for IBM Power ISA 2.07
2504 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2505 325628 Phase 5 support for IBM Power ISA 2.07
2506 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2507 325751 Missing the two privileged Power PC Transactional Memory Instructions
2508 325816 Phase 6 support for IBM Power ISA 2.07
2509 325856 Make SGCheck fail gracefully on unsupported platforms
2510 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2511 326436 DRD: False positive in libstdc++ std::list::push_back
2512 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2513 326462 Refactor vgdb to isolate invoker stuff into separate module
2514 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2515 326623 DRD: false positive conflict report in a field assignment
2516 326724 Valgrind does not compile on OSX 1.9 Mavericks
2517 326816 Intercept for __strncpy_sse2_unaligned missing?
2518 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2519 326983 Clear direction flag after tests on amd64.
2520 327212 Do not prepend the current directory to absolute path names.
2521 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2522 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2523 327284 s390x: Fix translation of the risbg instruction
2524 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2525 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2526 327916 DW_TAG_typedef may have no name
2527 327943 s390x: add a redirection for the 'index' function
2528 328100 XABORT not implemented
2529 328205 Implement additional Xen hypercalls
2530 328454 add support Backtraces with ARM unwind tables (EXIDX)
2531 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2532 328711 valgrind.1 manpage "memcheck options" section is badly generated
2533 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2534 329612 Incorrect handling of AT_BASE for image execution
2535 329694 clang warns about using uninitialized variable
2536 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2537 330228 mmap must align to VKI_SHMLBA on mips32
2538 330257 LLVM does not support `-mno-dynamic-no-pic` option
2539 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2540 330459 --track-fds=yes doesn't track eventfds
2541 330469 Add clock_adjtime syscall support
2542 330594 Missing sysalls on PowerPC / uClibc
2543 330622 Add test to regression suite for POWER instruction: dcbzl
2544 330939 Support for AMD's syscall instruction on x86
2546 330941 Typo in PRE(poll) syscall wrapper
2547 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2548 331254 Fix expected output for memcheck/tests/dw4
2549 331255 Fix race condition in test none/tests/coolo_sigaction
2550 331257 Fix type of jump buffer in test none/tests/faultstatus
2551 331305 configure uses bash specific syntax
2552 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2553 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2554 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2555 331829 Unexpected ioctl opcode sign extension
2556 331830 ppc64: WARNING: unhandled syscall: 96/97
2557 331839 drd/tests/sem_open specifies invalid semaphore name
2558 331847 outcome of drd/tests/thread_name is nondeterministic
2559 332037 Valgrind cannot handle Thumb "add pc, reg"
2560 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2561 consistency checks enabled
2562 332263 intercepts for pthread_rwlock_timedrdlock and
2563 pthread_rwlock_timedwrlock are incorrect
2564 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2566 332276 Implement additional Xen hypercalls
2567 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2568 332765 Fix ms_print to create temporary files in a proper directory
2569 333072 drd: Add semaphore annotations
2570 333145 Tests for missaligned PC+#imm access for arm
2571 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2572 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2573 333248 WARNING: unhandled syscall: unix:443
2574 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2575 333501 cachegrind: assertion: Cache set count is not a power of two.
2578 333666 Recognize MPX instructions and bnd prefix.
2579 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2580 333817 Valgrind reports the memory areas written to by the SG_IO
2582 334049 lzcnt fails silently (x86_32)
2583 334384 Valgrind does not have support Little Endian support for
2585 334585 recvmmsg unhandled (+patch) (arm)
2586 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2587 334727 Build fails with -Werror=format-security
2588 334788 clarify doc about --log-file initial program directory
2589 334834 PPC64 Little Endian support, patch 2
2590 334836 PPC64 Little Endian support, patch 3 testcase fixes
2591 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2592 335034 Unhandled ioctl: HCIGETDEVLIST
2593 335155 vgdb, fix error print statement.
2594 335262 arm64: movi 8bit version is not supported
2595 335263 arm64: dmb instruction is not implemented
2596 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2597 335496 arm64: sbc/abc instructions are not implemented
2598 335554 arm64: unhandled instruction: abs
2599 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2600 335735 arm64: unhandled instruction: cnt
2601 335736 arm64: unhandled instruction: uaddlv
2602 335848 arm64: unhandled instruction: {s,u}cvtf
2603 335902 arm64: unhandled instruction: sli
2604 335903 arm64: unhandled instruction: umull (vector)
2605 336055 arm64: unhandled instruction: mov (element)
2606 336062 arm64: unhandled instruction: shrn{,2}
2607 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2608 336189 arm64: unhandled Instruction: mvn
2609 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2610 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2611 336772 Make moans about unknown ioctls more informative
2612 336957 Add a section about the Solaris/illumos port on the webpage
2613 337094 ifunc wrapper is broken on ppc64
2614 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2615 337528 leak check heuristic for block prefixed by length as 64bit number
2616 337740 Implement additional Xen hypercalls
2617 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2618 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2619 337871 deprecate --db-attach
2620 338023 Add support for all V4L2/media ioctls
2621 338024 inlined functions are not shown if DW_AT_ranges is used
2622 338106 Add support for 'kcmp' syscall
2623 338115 DRD: computed conflict set differs from actual after fork
2624 338160 implement display of thread local storage in gdbsrv
2625 338205 configure.ac and check for -Wno-tautological-compare
2626 338300 coredumps are missing one byte of every segment
2627 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2628 338499 --sim-hints parsing broken due to wrong order in tokens
2629 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2630 338681 Unable to unwind through clone thread created on i386-linux
2631 338698 race condition between gdbsrv and vgdb on startup
2632 338703 helgrind on arm-linux gets false positives in dynamic loader
2633 338791 alt dwz files can be relative of debug/main file
2634 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2635 338932 build V-trunk with gcc-trunk
2636 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2637 345079 Fix build problems in VEX/useful/test_main.c
2638 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2639 n-i-bz s390x: Fix memory corruption for multithreaded applications
2640 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2641 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2642 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2643 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2644 n-i-bz Add command-line option --default-suppressions
2645 n-i-bz Add support for BLKDISCARDZEROES ioctl
2646 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2647 n-i-bz Add support for sys_pivot_root and sys_unshare
2649 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2650 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2651 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2655 Release 3.9.0 (31 October 2013)
2656 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2657 3.9.0 is a feature release with many improvements and the usual
2658 collection of bug fixes.
2660 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2661 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2662 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2663 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2665 * ================== PLATFORM CHANGES =================
2667 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2668 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2670 * Support for MIPS DSP ASE on MIPS32 platforms.
2672 * Support for s390x Decimal Floating Point instructions on hosts that
2673 have the DFP facility installed.
2675 * Support for POWER8 (Power ISA 2.07) instructions
2677 * Support for Intel AVX2 instructions. This is available only on 64
2680 * Initial support for Intel Transactional Synchronization Extensions,
2683 * Initial support for Hardware Transactional Memory on POWER.
2685 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2686 run large GUI apps tolerably well.
2688 * ==================== TOOL CHANGES ====================
2692 - Improvements in handling of vectorised code, leading to
2693 significantly fewer false error reports. You need to use the flag
2694 --partial-loads-ok=yes to get the benefits of these changes.
2696 - Better control over the leak checker. It is now possible to
2697 specify which leak kinds (definite/indirect/possible/reachable)
2698 should be displayed, which should be regarded as errors, and which
2699 should be suppressed by a given leak suppression. This is done
2700 using the options --show-leak-kinds=kind1,kind2,..,
2701 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2702 "match-leak-kinds:" line in suppression entries, respectively.
2704 Note that generated leak suppressions contain this new line and
2705 are therefore more specific than in previous releases. To get the
2706 same behaviour as previous releases, remove the "match-leak-kinds:"
2707 line from generated suppressions before using them.
2709 - Reduced "possible leak" reports from the leak checker by the use
2710 of better heuristics. The available heuristics provide detection
2711 of valid interior pointers to std::stdstring, to new[] allocated
2712 arrays with elements having destructors and to interior pointers
2713 pointing to an inner part of a C++ object using multiple
2714 inheritance. They can be selected individually using the
2715 option --leak-check-heuristics=heur1,heur2,...
2717 - Better control of stacktrace acquisition for heap-allocated
2718 blocks. Using the --keep-stacktraces option, it is possible to
2719 control independently whether a stack trace is acquired for each
2720 allocation and deallocation. This can be used to create better
2721 "use after free" errors or to decrease Valgrind's resource
2722 consumption by recording less information.
2724 - Better reporting of leak suppression usage. The list of used
2725 suppressions (shown when the -v option is given) now shows, for
2726 each leak suppressions, how many blocks and bytes it suppressed
2727 during the last leak search.
2731 - False errors resulting from the use of statically initialised
2732 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2735 - False errors resulting from the use of pthread_cond_waits that
2736 timeout, have been removed.
2738 * ==================== OTHER CHANGES ====================
2740 * Some attempt to tune Valgrind's space requirements to the expected
2741 capabilities of the target:
2743 - The default size of the translation cache has been reduced from 8
2744 sectors to 6 on Android platforms, since each sector occupies
2745 about 40MB when using Memcheck.
2747 - The default size of the translation cache has been increased to 16
2748 sectors on all other platforms, reflecting the fact that large
2749 applications require instrumentation and storage of huge amounts
2750 of code. For similar reasons, the number of memory mapped
2751 segments that can be tracked has been increased by a factor of 6.
2753 - In all cases, the maximum number of sectors in the translation
2754 cache can be controlled by the new flag --num-transtab-sectors.
2756 * Changes in how debug info (line numbers, etc) is read:
2758 - Valgrind no longer temporarily mmaps the entire object to read
2759 from it. Instead, reading is done through a small fixed sized
2760 buffer. This avoids virtual memory usage spikes when Valgrind
2761 reads debuginfo from large shared objects.
2763 - A new experimental remote debug info server. Valgrind can read
2764 debug info from a different machine (typically, a build host)
2765 where debuginfo objects are stored. This can save a lot of time
2766 and hassle when running Valgrind on resource-constrained targets
2767 (phones, tablets) when the full debuginfo objects are stored
2768 somewhere else. This is enabled by the --debuginfo-server=
2771 - Consistency checking between main and debug objects can be
2772 disabled using the --allow-mismatched-debuginfo option.
2774 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2775 scanning can recover stack traces in some cases when the normal
2776 unwind mechanisms fail. Stack scanning is best described as "a
2777 nasty, dangerous and misleading hack" and so is disabled by default.
2778 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2781 * Detection and merging of recursive stack frame cycles. When your
2782 program has recursive algorithms, this limits the memory used by
2783 Valgrind for recorded stack traces and avoids recording
2784 uninteresting repeated calls. This is controlled by the command
2785 line option --merge-recursive-frame and by the monitor command
2786 "v.set merge-recursive-frames".
2788 * File name and line numbers for used suppressions. The list of used
2789 suppressions (shown when the -v option is given) now shows, for each
2790 used suppression, the file name and line number where the suppression
2793 * New and modified GDB server monitor features:
2795 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2796 that can be used to execute gdbserver monitor commands from the
2799 - A new monitor command, "v.info open_fds", that gives the list of
2800 open file descriptors and additional details.
2802 - An optional message in the "v.info n_errs_found" monitor command,
2803 for example "v.info n_errs_found test 1234 finished", allowing a
2804 comment string to be added to the process output, perhaps for the
2805 purpose of separating errors of different tests or test phases.
2807 - A new monitor command "v.info execontext" that shows information
2808 about the stack traces recorded by Valgrind.
2810 - A new monitor command "v.do expensive_sanity_check_general" to run
2811 some internal consistency checks.
2813 * New flag --sigill-diagnostics to control whether a diagnostic
2814 message is printed when the JIT encounters an instruction it can't
2815 translate. The actual behavior -- delivery of SIGILL to the
2816 application -- is unchanged.
2818 * The maximum amount of memory that Valgrind can use on 64 bit targets
2819 has been increased from 32GB to 64GB. This should make it possible
2820 to run applications on Memcheck that natively require up to about 35GB.
2822 * ==================== FIXED BUGS ====================
2824 The following bugs have been fixed or resolved. Note that "n-i-bz"
2825 stands for "not in bugzilla" -- that is, a bug that was reported to us
2826 but never got a bugzilla entry. We encourage you to file bugs in
2827 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2828 than mailing the developers (or mailing lists) directly -- bugs that
2829 are not entered into bugzilla tend to get forgotten about or ignored.
2831 To see details of a given bug, visit
2832 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2833 where XXXXXX is the bug number as listed below.
2835 123837 system call: 4th argument is optional, depending on cmd
2836 135425 memcheck should tell you where Freed blocks were Mallocd
2837 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2838 207815 Adds some of the drm ioctls to syswrap-linux.c
2839 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2840 252955 Impossible to compile with ccache
2841 253519 Memcheck reports auxv pointer accesses as invalid reads.
2842 263034 Crash when loading some PPC64 binaries
2843 269599 Increase deepest backtrace
2844 274695 s390x: Support "compare to/from logical" instructions (z196)
2845 275800 s390x: Autodetect cache info (part 2)
2846 280271 Valgrind reports possible memory leaks on still-reachable std::string
2847 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2848 289578 Backtraces with ARM unwind tables (stack scan flags)
2849 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2850 304832 ppc32: build failure
2851 305431 Use find_buildid shdr fallback for separate .debug files
2852 305728 Add support for AVX2 instructions
2853 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2854 306035 s390x: Fix IR generation for LAAG and friends
2855 306054 s390x: Condition code computation for convert-to-int/logical
2856 306098 s390x: alternate opcode form for convert to/from fixed
2857 306587 Fix cache line detection from auxiliary vector for PPC.
2858 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2859 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2860 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2861 307101 sys_capget second argument can be NULL
2862 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2863 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2864 307113 s390x: DFP support
2865 307141 valgrind does't work in mips-linux system
2866 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2867 307285 x86_amd64 feature test for avx in test suite is wrong
2868 307290 memcheck overlap testcase needs memcpy version filter
2869 307463 Please add "&limit=0" to the "all open bugs" link
2870 307465 --show-possibly-lost=no should reduce the error count / exit code
2871 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2872 307729 pkgconfig support broken valgrind.pc
2873 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2874 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2875 308089 Unhandled syscall on ppc64: prctl
2876 308135 PPC32 MPC8xx has 16 bytes cache size
2877 308321 testsuite memcheck filter interferes with gdb_filter
2879 308341 vgdb should report process exit (or fatal signal)
2880 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2881 308495 Remove build dependency on installed Xen headers
2882 308573 Internal error on 64-bit instruction executed in 32-bit mode
2884 308627 pmovmskb validity bit propagation is imprecise
2885 308644 vgdb command for having the info for the track-fds option
2886 308711 give more info about aspacemgr and arenas in out_of_memory
2887 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2888 308718 ARM implement SMLALBB family of instructions
2889 308886 Missing support for PTRACE_SET/GETREGSET
2890 308930 syscall name_to_handle_at (303 on amd64) not handled
2891 309229 V-bit tester does not report number of tests generated
2892 309323 print unrecognized instuction on MIPS
2893 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2894 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2895 309430 Self hosting ppc64 encounters a vassert error on operand type
2896 309600 valgrind is a bit confused about 0-sized sections
2897 309823 Generate errors for still reachable blocks
2898 309921 PCMPISTRI validity bit propagation is imprecise
2899 309922 none/tests/ppc64/test_dfp5 sometimes fails
2900 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2901 310424 --read-var-info does not properly describe static variables
2902 310792 search additional path for debug symbols
2903 310931 s390x: Message-security assist (MSA) instruction extension [..]
2904 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2905 311318 ARM: "128-bit constant is not implemented" error message
2906 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2907 311690 V crashes because it redirects branches inside of a redirected function
2908 311880 x86_64: make regtest hangs at shell_valid1
2909 311922 WARNING: unhandled syscall: 170
2911 312171 ppc: insn selection for DFP
2912 312571 Rounding mode call wrong for the DFP Iops [..]
2913 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2914 312913 Dangling pointers error should also report the alloc stack trace
2915 312980 Building on Mountain Lion generates some compiler warnings
2916 313267 Adding MIPS64/Linux port to Valgrind
2919 313811 Buffer overflow in assert_fail
2920 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2921 314269 ppc: dead code in insn selection
2922 314718 ARM: implement integer divide instruction (sdiv and udiv)
2923 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2924 315441 sendmsg syscall should ignore unset msghdr msg_flags
2925 315534 msgrcv inside a thread causes valgrind to hang (block)
2926 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2927 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2928 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2929 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2930 316144 valgrind.1 manpage contains unknown ??? strings [..]
2931 316145 callgrind command line options in manpage reference (unknown) [..]
2932 316145 callgrind command line options in manpage reference [..]
2933 316181 drd: Fixed a 4x slowdown for certain applications
2934 316503 Valgrind does not support SSE4 "movntdqa" instruction
2935 316535 Use of |signed int| instead of |size_t| in valgrind messages
2936 316696 fluidanimate program of parsec 2.1 stuck
2937 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2938 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2939 317186 "Impossible happens" when occurs VCVT instruction on ARM
2940 317318 Support for Threading Building Blocks "scalable_malloc"
2941 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2942 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2943 317463 bmi testcase IR SANITY CHECK FAILURE
2944 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2945 318050 libmpiwrap fails to compile with out-of-source build
2946 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2947 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2948 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2949 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2950 318932 Add missing PPC64 and PPC32 system call support
2951 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2952 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2953 319494 VEX Makefile-gcc standalone build update after r2702
2954 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2955 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2956 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2957 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2958 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2959 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2960 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2961 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2962 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2963 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2964 320895 add fanotify support (patch included)
2965 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2966 321065 Valgrind updates for Xen 4.3
2967 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2968 321363 Unhandled instruction: SSAX (ARM + Thumb)
2969 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2970 321466 Unhandled instruction: SHASX (ARM + Thumb)
2971 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2972 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2973 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2974 321620 Unhandled instruction: UASX (ARM + Thumb)
2975 321621 Unhandled instruction: USAX (ARM + Thumb)
2976 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2977 321693 Unhandled instruction: LDRSBT (Thumb)
2978 321694 Unhandled instruction: UQASX (ARM + Thumb)
2979 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2980 321697 Unhandled instruction: UHASX (ARM + Thumb)
2981 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2982 321704 Unhandled instruction: REVSH (ARM + Thumb)
2983 321730 Add cg_diff and cg_merge man pages
2984 321738 Add vgdb and valgrind-listener man pages
2986 321891 Unhandled instruction: LDRHT (Thumb)
2987 321960 pthread_create() then alloca() causing invalid stack write errors
2988 321969 ppc32 and ppc64 don't support [lf]setxattr
2989 322254 Show threadname together with tid if set by application
2990 322294 Add initial support for IBM Power ISA 2.07
2991 322368 Assertion failure in wqthread_hijack under OS X 10.8
2992 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2993 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2994 322851 0bXXX binary literal syntax is not standard
2995 323035 Unhandled instruction: LDRSHT(Thumb)
2996 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2997 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2998 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2999 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
3000 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
3001 323437 Phase 2 support for IBM Power ISA 2.07
3002 323713 Support mmxext (integer sse) subset on i386 (athlon)
3003 323803 Transactional memory instructions are not supported for Power
3004 323893 SSE3 not available on amd cpus in valgrind
3005 323905 Probable false positive from Valgrind/drd on close()
3006 323912 valgrind.h header isn't compatible for mingw64
3007 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
3008 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
3009 324181 mmap does not handle MAP_32BIT
3010 324227 memcheck false positive leak when a thread calls exit+block [..]
3011 324421 Support for fanotify API on ARM architecture
3012 324514 gdbserver monitor cmd output behaviour consistency [..]
3013 324518 ppc64: Emulation of dcbt instructions does not handle [..]
3014 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
3015 324582 When access is made to freed memory, report both allocation [..]
3016 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
3017 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
3018 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
3019 324834 Unhandled instructions in Microsoft C run-time for x86_64
3020 324894 Phase 3 support for IBM Power ISA 2.07
3021 326091 drd: Avoid false race reports from optimized strlen() impls
3022 326113 valgrind libvex hwcaps error on AMD64
3023 n-i-bz Some wrong command line options could be ignored
3024 n-i-bz patch to allow fair-sched on android
3025 n-i-bz report error for vgdb snapshot requested before execution
3026 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
3028 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
3032 Release 3.8.1 (19 September 2012)
3033 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3034 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
3035 that occur moderately frequently in real use cases, adds support for
3036 some missing instructions on ARM, and fixes a deadlock condition on
3037 MacOSX. If you package or deliver 3.8.0 for others to use, you might
3038 want to consider upgrading to 3.8.1 instead.
3040 The following bugs have been fixed or resolved. Note that "n-i-bz"
3041 stands for "not in bugzilla" -- that is, a bug that was reported to us
3042 but never got a bugzilla entry. We encourage you to file bugs in
3043 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3044 than mailing the developers (or mailing lists) directly -- bugs that
3045 are not entered into bugzilla tend to get forgotten about or ignored.
3047 To see details of a given bug, visit
3048 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3049 where XXXXXX is the bug number as listed below.
3052 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
3053 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
3054 298281 wcslen causes false(?) uninitialised value warnings
3055 301281 valgrind hangs on OS X when the process calls system()
3056 304035 disInstr(arm): unhandled instruction 0xE1023053
3057 304867 implement MOVBE instruction in x86 mode
3058 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
3059 305042 amd64: implement 0F 7F encoding of movq between two registers
3060 305199 ARM: implement QDADD and QDSUB
3061 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
3062 305513 killed by fatal signal: SIGSEGV
3063 305690 DRD reporting invalid semaphore when sem_trywait fails
3064 305926 Invalid alignment checks for some AVX instructions
3065 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
3066 306310 3.8.0 release tarball missing some files
3067 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
3068 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
3069 n-i-bz shmat of a segment > 4Gb does not work
3070 n-i-bz simulate_control_c script wrong USR1 signal number on mips
3071 n-i-bz vgdb ptrace calls wrong on mips [...]
3072 n-i-bz Fixes for more MPI false positives
3073 n-i-bz exp-sgcheck's memcpy causes programs to segfault
3074 n-i-bz OSX build w/ clang: asserts at startup
3075 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
3076 n-i-bz fix a couple of union tag-vs-field mixups
3077 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
3079 The following bugs were fixed in 3.8.0 but not listed in this NEWS
3082 254088 Valgrind should know about UD2 instruction
3085 304754 NEWS blows TeX's little mind
3087 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
3091 Release 3.8.0 (10 August 2012)
3092 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3093 3.8.0 is a feature release with many improvements and the usual
3094 collection of bug fixes.
3096 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3097 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
3098 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
3099 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
3100 There is initial support for MacOSX 10.8, but it is not usable for
3101 serious work at present.
3103 * ================== PLATFORM CHANGES =================
3105 * Support for MIPS32 platforms running Linux. Valgrind has been
3106 tested on MIPS32 and MIPS32r2 platforms running different Debian
3107 Squeeze and MeeGo distributions. Both little-endian and big-endian
3108 cores are supported. The tools Memcheck, Massif and Lackey have
3109 been tested and are known to work. See README.mips for more details.
3111 * Preliminary support for Android running on x86.
3113 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
3115 * Support for Intel AVX instructions and for AES instructions. This
3116 support is available only for 64 bit code.
3118 * Support for POWER Decimal Floating Point instructions.
3120 * ==================== TOOL CHANGES ====================
3122 * Non-libc malloc implementations are now supported. This is useful
3123 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
3124 Using the new option --soname-synonyms, such tools can be informed
3125 that the malloc implementation is either linked statically into the
3126 executable, or is present in some other shared library different
3127 from libc.so. This makes it possible to process statically linked
3128 programs, and programs using other malloc libraries, for example
3129 TCMalloc or JEMalloc.
3131 * For tools that provide their own replacement for malloc et al, the
3132 option --redzone-size=<number> allows users to specify the size of
3133 the padding blocks (redzones) added before and after each client
3134 allocated block. Smaller redzones decrease the memory needed by
3135 Valgrind. Bigger redzones increase the chance to detect blocks
3136 overrun or underrun. Prior to this change, the redzone size was
3137 hardwired to 16 bytes in Memcheck.
3141 - The leak_check GDB server monitor command now can
3142 control the maximum nr of loss records to output.
3144 - Reduction of memory use for applications allocating
3145 many blocks and/or having many partially defined bytes.
3147 - Addition of GDB server monitor command 'block_list' that lists
3148 the addresses/sizes of the blocks of a leak search loss record.
3150 - Addition of GDB server monitor command 'who_points_at' that lists
3151 the locations pointing at a block.
3153 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
3154 detect an invalid access of these redzones, by marking them
3155 noaccess. Similarly, if a redzone size is given for a memory
3156 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
3157 This still allows to find some bugs if the user has forgotten to
3158 mark the pool superblock noaccess.
3160 - Performance of memory leak check has been improved, especially in
3161 cases where there are many leaked blocks and/or many suppression
3162 rules used to suppress leak reports.
3164 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
3165 more precise analysis, which is important for LLVM/Clang
3166 generated code. This is at the cost of somewhat reduced
3167 performance. Note there is no change to analysis precision or
3168 costs on Linux targets.
3172 - Added even more facilities that can help finding the cause of a data
3173 race, namely the command-line option --ptrace-addr and the macro
3174 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
3176 - Fixed a subtle bug that could cause false positive data race reports.
3178 * ==================== OTHER CHANGES ====================
3180 * The C++ demangler has been updated so as to work well with C++
3181 compiled by up to at least g++ 4.6.
3183 * Tool developers can make replacement/wrapping more flexible thanks
3184 to the new option --soname-synonyms. This was reported above, but
3185 in fact is very general and applies to all function
3186 replacement/wrapping, not just to malloc-family functions.
3188 * Round-robin scheduling of threads can be selected, using the new
3189 option --fair-sched= yes. Prior to this change, the pipe-based
3190 thread serialisation mechanism (which is still the default) could
3191 give very unfair scheduling. --fair-sched=yes improves
3192 responsiveness of interactive multithreaded applications, and
3193 improves repeatability of results from the thread checkers Helgrind
3196 * For tool developers: support to run Valgrind on Valgrind has been
3197 improved. We can now routinely Valgrind on Helgrind or Memcheck.
3199 * gdbserver now shows the float shadow registers as integer
3200 rather than float values, as the shadow values are mostly
3201 used as bit patterns.
3203 * Increased limit for the --num-callers command line flag to 500.
3205 * Performance improvements for error matching when there are many
3206 suppression records in use.
3208 * Improved support for DWARF4 debugging information (bug 284184).
3210 * Initial support for DWZ compressed Dwarf debug info.
3212 * Improved control over the IR optimiser's handling of the tradeoff
3213 between performance and precision of exceptions. Specifically,
3214 --vex-iropt-precise-memory-exns has been removed and replaced by
3215 --vex-iropt-register-updates, with extended functionality. This
3216 allows the Valgrind gdbserver to always show up to date register
3219 * Modest performance gains through the use of translation chaining for
3222 * ==================== FIXED BUGS ====================
3224 The following bugs have been fixed or resolved. Note that "n-i-bz"
3225 stands for "not in bugzilla" -- that is, a bug that was reported to us
3226 but never got a bugzilla entry. We encourage you to file bugs in
3227 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3228 than mailing the developers (or mailing lists) directly -- bugs that
3229 are not entered into bugzilla tend to get forgotten about or ignored.
3231 To see details of a given bug, visit
3232 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3233 where XXXXXX is the bug number as listed below.
3235 197914 Building valgrind from svn now requires automake-1.10
3236 203877 increase to 16Mb maximum allowed alignment for memalign et al
3237 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
3238 247386 make perf does not run all performance tests
3239 270006 Valgrind scheduler unfair
3240 270777 Adding MIPS/Linux port to Valgrind
3241 270796 s390x: Removed broken support for the TS insn
3242 271438 Fix configure for proper SSE4.2 detection
3243 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
3244 273475 Add support for AVX instructions
3245 274078 improved configure logic for mpicc
3246 276993 fix mremap 'no thrash checks'
3247 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
3248 281482 memcheck incorrect byte allocation count in realloc() for silly argument
3249 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
3250 283413 Fix wrong sanity check
3251 283671 Robustize alignment computation in LibVEX_Alloc
3252 283961 Adding support for some HCI IOCTLs
3253 284124 parse_type_DIE: confused by: DWARF 4
3254 284864 == 273475 (Add support for AVX instructions)
3255 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
3256 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
3257 285725 == 273475 (Add support for AVX instructions)
3258 286261 add wrapper for linux I2C_RDWR ioctl
3259 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
3260 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
3261 286384 configure fails "checking for a supported version of gcc"
3262 286497 == 273475 (Add support for AVX instructions)
3263 286596 == 273475 (Add support for AVX instructions)
3264 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
3265 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
3266 287260 Incorrect conditional jump or move depends on uninitialised value(s)
3267 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
3268 287307 == 273475 (Add support for AVX instructions)
3269 287858 VG_(strerror): unknown error
3270 288298 (MacOSX) unhandled syscall shm_unlink
3271 288995 == 273475 (Add support for AVX instructions)
3272 289470 Loading of large Mach-O thin binaries fails.
3273 289656 == 273475 (Add support for AVX instructions)
3274 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
3275 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
3276 289839 s390x: Provide support for unicode conversion instructions
3277 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
3278 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
3279 290655 Add support for AESKEYGENASSIST instruction
3280 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
3281 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
3282 291253 ES register not initialised in valgrind simulation
3283 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
3284 291865 s390x: Support the "Compare Double and Swap" family of instructions
3285 292300 == 273475 (Add support for AVX instructions)
3286 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
3287 292493 == 273475 (Add support for AVX instructions)
3288 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
3289 292627 Missing support for some SCSI ioctls
3290 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
3291 292841 == 273475 (Add support for AVX instructions)
3292 292993 implement the getcpu syscall on amd64-linux
3293 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
3294 293088 Add some VEX sanity checks for ppc64 unhandled instructions
3295 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
3296 293754 PCMPxSTRx not implemented for 16-bit characters
3297 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
3298 293808 CLFLUSH not supported by latest VEX for amd64
3299 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
3300 294048 MPSADBW instruction not implemented
3301 294055 regtest none/tests/shell fails when locale is not set to C
3302 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
3303 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
3304 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
3305 294260 disInstr_AMD64: disInstr miscalculated next %rip
3306 294523 --partial-loads-ok=yes causes false negatives
3307 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
3308 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
3309 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
3310 295089 can not annotate source for both helgrind and drd
3311 295221 POWER Processor decimal floating point instruction support missing
3312 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
3313 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
3314 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
3315 295617 ARM - Add some missing syscalls
3316 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
3317 296229 Linux user input device ioctls missing wrappers
3318 296318 ELF Debug info improvements (more than one rx/rw mapping)
3319 296422 Add translation chaining support
3320 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
3321 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
3322 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
3323 297078 gdbserver signal handling problems [..]
3324 297147 drd false positives on newly allocated memory
3325 297329 disallow decoding of IBM Power DFP insns on some machines
3326 297497 POWER Processor decimal floating point instruction support missing
3327 297701 Another alias for strncasecmp_l in libc-2.13.so
3328 297911 'invalid write' not reported when using APIs for custom mem allocators.
3329 297976 s390x: revisit EX implementation
3330 297991 Valgrind interferes with mmap()+ftell()
3331 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
3332 297993 Fix compilation of valgrind with gcc -g3.
3333 298080 POWER Processor DFP support missing, part 3
3334 298227 == 273475 (Add support for AVX instructions)
3335 298335 == 273475 (Add support for AVX instructions)
3336 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
3337 298394 s390x: Don't bail out on an unknown machine model. [..]
3338 298421 accept4() syscall (366) support is missing for ARM
3339 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
3340 298732 valgrind installation problem in ubuntu with kernel version 3.x
3341 298862 POWER Processor DFP instruction support missing, part 4
3342 298864 DWARF reader mis-parses DW_FORM_ref_addr
3343 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
3344 299053 Support DWARF4 DW_AT_high_pc constant form
3345 299104 == 273475 (Add support for AVX instructions)
3346 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
3347 299629 dup3() syscall (358) support is missing for ARM
3348 299694 POWER Processor DFP instruction support missing, part 5
3349 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
3350 299803 == 273475 (Add support for AVX instructions)
3351 299804 == 273475 (Add support for AVX instructions)
3352 299805 == 273475 (Add support for AVX instructions)
3353 300140 ARM - Missing (T1) SMMUL
3354 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
3355 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
3356 300414 FCOM and FCOMP unimplemented for amd64 guest
3357 301204 infinite loop in canonicaliseSymtab with ifunc symbol
3358 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
3359 301265 add x86 support to Android build
3360 301984 configure script doesn't detect certain versions of clang
3361 302205 Fix compiler warnings for POWER VEX code and POWER test cases
3362 302287 Unhandled movbe instruction on Atom processors
3363 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
3364 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
3365 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
3366 302656 == 273475 (Add support for AVX instructions)
3367 302709 valgrind for ARM needs extra tls support for android emulator [..]
3368 302827 add wrapper for CDROM_GET_CAPABILITY
3369 302901 Valgrind crashes with dwz optimized debuginfo
3370 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
3371 303116 Add support for the POWER instruction popcntb
3372 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
3373 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
3374 303466 == 273475 (Add support for AVX instructions)
3375 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
3376 303963 strstr() function produces wrong results under valgrind callgrind
3377 304054 CALL_FN_xx macros need to enforce stack alignment
3378 304561 tee system call not supported
3379 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
3380 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
3381 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
3382 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
3383 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
3384 n-i-bz s390x: Shadow registers can now be examined using vgdb
3386 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
3387 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
3391 Release 3.7.0 (5 November 2011)
3392 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3393 3.7.0 is a feature release with many significant improvements and the
3394 usual collection of bug fixes.
3396 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3397 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
3398 Support for recent distros and toolchain components (glibc 2.14, gcc
3399 4.6, MacOSX 10.7) has been added.
3401 * ================== PLATFORM CHANGES =================
3403 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
3404 analyse 64-bit programs running on z/Architecture. Most user space
3405 instructions up to and including z10 are supported. Valgrind has
3406 been tested extensively on z9, z10, and z196 machines running SLES
3407 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
3408 known to work well. Callgrind, Helgrind, and DRD work reasonably
3409 well on z9 and later models. See README.s390 for more details.
3411 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
3412 64-bit processes are supported. Some complex threaded applications
3413 (Firefox) are observed to hang when run as 32 bit applications,
3414 whereas 64-bit versions run OK. The cause is unknown. Memcheck
3415 will likely report some false errors. In general, expect some rough
3416 spots. This release also supports MacOSX 10.6, but drops support
3419 * Preliminary support for Android (on ARM). Valgrind can now run
3420 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
3421 README.android for more details, plus instructions on how to get
3424 * Support for the IBM Power ISA 2.06 (Power7 instructions)
3426 * General correctness and performance improvements for ARM/Linux, and,
3427 by extension, ARM/Android.
3429 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
3430 instruction set support is under development but is not available in
3433 * Support for AIX5 has been removed.
3435 * ==================== TOOL CHANGES ====================
3437 * Memcheck: some incremental changes:
3439 - reduction of memory use in some circumstances
3441 - improved handling of freed memory, which in some circumstances
3442 can cause detection of use-after-free that would previously have
3445 - fix of a longstanding bug that could cause false negatives (missed
3446 errors) in programs doing vector saturated narrowing instructions.
3448 * Helgrind: performance improvements and major memory use reductions,
3449 particularly for large, long running applications which perform many
3450 synchronisation (lock, unlock, etc) events. Plus many smaller
3453 - display of locksets for both threads involved in a race
3455 - general improvements in formatting/clarity of error messages
3457 - addition of facilities and documentation regarding annotation
3458 of thread safe reference counted C++ classes
3460 - new flag --check-stack-refs=no|yes [yes], to disable race checking
3461 on thread stacks (a performance hack)
3463 - new flag --free-is-write=no|yes [no], to enable detection of races
3464 where one thread accesses heap memory but another one frees it,
3465 without any coordinating synchronisation event
3467 * DRD: enabled XML output; added support for delayed thread deletion
3468 in order to detect races that occur close to the end of a thread
3469 (--join-list-vol); fixed a memory leak triggered by repeated client
3470 memory allocatation and deallocation; improved Darwin support.
3472 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
3474 * exp-sgcheck: this tool has been reduced in scope so as to improve
3475 performance and remove checking that Memcheck does better.
3476 Specifically, the ability to check for overruns for stack and global
3477 arrays is unchanged, but the ability to check for overruns of heap
3478 blocks has been removed. The tool has accordingly been renamed to
3479 exp-sgcheck ("Stack and Global Array Checking").
3481 * ==================== OTHER CHANGES ====================
3483 * GDB server: Valgrind now has an embedded GDB server. That means it
3484 is possible to control a Valgrind run from GDB, doing all the usual
3485 things that GDB can do (single stepping, breakpoints, examining
3486 data, etc). Tool-specific functionality is also available. For
3487 example, it is possible to query the definedness state of variables
3488 or memory from within GDB when running Memcheck; arbitrarily large
3489 memory watchpoints are supported, etc. To use the GDB server, start
3490 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3493 * Improved support for unfriendly self-modifying code: a new option
3494 --smc-check=all-non-file is available. This adds the relevant
3495 consistency checks only to code that originates in non-file-backed
3496 mappings. In effect this confines the consistency checking only to
3497 code that is or might be JIT generated, and avoids checks on code
3498 that must have been compiled ahead of time. This significantly
3499 improves performance on applications that generate code at run time.
3501 * It is now possible to build a working Valgrind using Clang-2.9 on
3504 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3505 These enable and disable error reporting on a per-thread, and
3506 nestable, basis. This is useful for hiding errors in particularly
3507 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3508 now uses this facility.
3510 * Added the --mod-funcname option to cg_diff.
3512 * ==================== FIXED BUGS ====================
3514 The following bugs have been fixed or resolved. Note that "n-i-bz"
3515 stands for "not in bugzilla" -- that is, a bug that was reported to us
3516 but never got a bugzilla entry. We encourage you to file bugs in
3517 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3518 mailing the developers (or mailing lists) directly -- bugs that are
3519 not entered into bugzilla tend to get forgotten about or ignored.
3521 To see details of a given bug, visit
3522 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3523 where XXXXXX is the bug number as listed below.
3525 79311 malloc silly arg warning does not give stack trace
3526 210935 port valgrind.h (not valgrind) to win32 to support client requests
3527 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3528 243404 Port to zSeries
3529 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3530 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3531 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3532 253206 Some fixes for the faultstatus testcase
3533 255223 capget testcase fails when running as root
3534 256703 xlc_dbl_u32.c testcase broken
3535 256726 Helgrind tests have broken inline asm
3536 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3537 264800 testcase compile failure on zseries
3538 265762 make public VEX headers compilable by G++ 3.x
3539 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3540 266753 configure script does not give the user the option to not use QtCore
3541 266931 gen_insn_test.pl is broken
3542 266961 ld-linux.so.2 i?86-linux strlen issues
3543 266990 setns instruction causes false positive
3544 267020 Make directory for temporary files configurable at run-time.
3545 267342 == 267997 (segmentation fault on Mac OS 10.6)
3546 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3547 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3548 267488 regtest: darwin support for 64-bit build
3549 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3550 267630 Add support for IBM Power ISA 2.06 -- stage 1
3551 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3552 267819 Add client request for informing the core about reallocation
3553 267925 laog data structure quadratic for a single sequence of lock
3554 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3555 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3556 268513 missed optimizations in fold_Expr
3557 268619 s390x: fpr - gpr transfer facility
3558 268620 s390x: reconsider "long displacement" requirement
3559 268621 s390x: improve IR generation for XC
3560 268715 s390x: FLOGR is not universally available
3561 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3562 268930 s390x: MHY is not universally available
3563 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3564 269079 Support ptrace system call on ARM
3565 269144 missing "Bad option" error message
3566 269209 conditional load and store facility (z196)
3567 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3568 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3569 269736 s390x: minor code generation tweaks
3570 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3571 269863 s390x: remove unused function parameters
3572 269864 s390x: tweak s390_emit_load_cc
3573 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3574 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3575 270115 s390x: rewrite some testcases
3576 270309 == 267997 (valgrind crash on startup)
3577 270320 add support for Linux FIOQSIZE ioctl() call
3578 270326 segfault while trying to sanitize the environment passed to execle
3579 270794 IBM POWER7 support patch causes regression in none/tests
3580 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3581 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3582 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3583 270959 s390x: invalid use of R0 as base register
3584 271042 VSX configure check fails when it should not
3585 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3586 271259 s390x: fix code confusion
3587 271337 == 267997 (Valgrind segfaults on MacOS X)
3588 271385 s390x: Implement Ist_MBE
3589 271501 s390x: misc cleanups
3590 271504 s390x: promote likely and unlikely
3591 271579 ppc: using wrong enum type
3592 271615 unhandled instruction "popcnt" (arch=amd10h)
3593 271730 Fix bug when checking ioctls: duplicate check
3594 271776 s390x: provide STFLE instruction support
3595 271779 s390x: provide clock instructions like STCK
3596 271799 Darwin: ioctls without an arg report a memory error
3597 271820 arm: fix type confusion
3598 271917 pthread_cond_timedwait failure leads to not-locked false positive
3599 272067 s390x: fix DISP20 macro
3600 272615 A typo in debug output in mc_leakcheck.c
3601 272661 callgrind_annotate chokes when run from paths containing regex chars
3602 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3603 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3604 272967 make documentation build-system more robust
3605 272986 Fix gcc-4.6 warnings with valgrind.h
3606 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3607 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3608 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3609 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3610 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3611 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3612 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3613 273778 exp-ptrcheck: unhandled sysno == 259
3614 274089 exp-ptrcheck: unhandled sysno == 208
3615 274378 s390x: Various dispatcher tweaks
3616 274447 WARNING: unhandled syscall: 340
3617 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3618 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3619 274926 valgrind does not build against linux-3
3620 275148 configure FAIL with glibc-2.14
3621 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3622 275168 Make Valgrind work for MacOSX 10.7 Lion
3623 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3624 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3625 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3626 275308 Fix implementation for ppc64 fres instruc
3627 275339 s390x: fix testcase compile warnings
3628 275517 s390x: Provide support for CKSM instruction
3629 275710 s390x: get rid of redundant address mode calculation
3630 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3631 275852 == 250101 (valgrind uses all swap space and is killed)
3632 276784 Add support for IBM Power ISA 2.06 -- stage 3
3633 276987 gdbsrv: fix tests following recent commits
3634 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3635 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3636 277471 Unhandled syscall: 340
3637 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3638 277653 ARM: support Thumb2 PLD instruction
3639 277663 ARM: NEON float VMUL by scalar incorrect
3640 277689 ARM: tests for VSTn with register post-index are broken
3641 277694 ARM: BLX LR instruction broken in ARM mode
3642 277780 ARM: VMOV.F32 (immediate) instruction is broken
3643 278057 fuse filesystem syscall deadlocks
3644 278078 Unimplemented syscall 280 on ppc32
3645 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3646 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3647 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3648 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3649 279027 Support for MVCL and CLCL instruction
3650 279027 s390x: Provide support for CLCL and MVCL instructions
3651 279062 Remove a redundant check in the insn selector for ppc.
3652 279071 JDK creates PTEST with redundant REX.W prefix
3653 279212 gdbsrv: add monitor cmd v.info scheduler.
3654 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3655 279698 memcheck discards valid-bits for packuswb
3656 279795 memcheck reports uninitialised values for mincore on amd64
3657 279994 Add support for IBM Power ISA 2.06 -- stage 3
3658 280083 mempolicy syscall check errors
3659 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3660 280710 s390x: config files for nightly builds
3661 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3662 280965 Valgrind breaks fcntl locks when program does mmap
3663 281138 WARNING: unhandled syscall: 340
3664 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3665 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3666 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3667 281468 s390x: handle do_clone and gcc clones in call traces
3668 281488 ARM: VFP register corruption
3669 281828 == 275284 (false memmove warning: "Source and destination overlap")
3670 281883 s390x: Fix system call wrapper for "clone".
3671 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3672 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3673 282238 SLES10: make check fails
3674 282979 strcasestr needs replacement with recent(>=2.12) glibc
3675 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3676 283243 Regression in ppc64 memcheck tests
3677 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3678 283427 re-connect epoll_pwait syscall on ARM linux
3679 283600 gdbsrv: android: port vgdb.c
3680 283709 none/tests/faultstatus needs to account for page size
3681 284305 filter_gdb needs enhancement to work on ppc64
3682 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3683 284472 Thumb2 ROR.W encoding T2 not implemented
3684 284621 XML-escape process command line in XML output
3685 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3686 that have non-power-of-2 sizes (also AMDs)
3687 n-i-bz don't be spooked by libraries mashed by elfhack
3688 n-i-bz don't be spooked by libxul.so linked with gold
3689 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3691 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3692 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3693 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3697 Release 3.6.1 (16 February 2011)
3698 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3699 3.6.1 is a bug fix release. It adds support for some SSE4
3700 instructions that were omitted in 3.6.0 due to lack of time. Initial
3701 support for glibc-2.13 has been added. A number of bugs causing
3702 crashing or assertion failures have been fixed.
3704 The following bugs have been fixed or resolved. Note that "n-i-bz"
3705 stands for "not in bugzilla" -- that is, a bug that was reported to us
3706 but never got a bugzilla entry. We encourage you to file bugs in
3707 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3708 mailing the developers (or mailing lists) directly -- bugs that are
3709 not entered into bugzilla tend to get forgotten about or ignored.
3711 To see details of a given bug, visit
3712 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3713 where XXXXXX is the bug number as listed below.
3715 188572 Valgrind on Mac should suppress setenv() mem leak
3716 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3717 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3718 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3719 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3720 254420 memory pool tracking broken
3721 254957 Test code failing to compile due to changes in memcheck.h
3722 255009 helgrind/drd: crash on chmod with invalid parameter
3723 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3724 255355 helgrind/drd: crash on threaded programs doing fork
3726 255418 (SSE4.x) rint call compiled with ICC
3727 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3728 255888 closing valgrindoutput tag outputted to log-stream on error
3729 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3730 255966 Slowness when using mempool annotations
3731 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3732 256600 super-optimized strcasecmp() false positive
3733 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3734 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3735 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3736 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3737 257276 Missing case in memcheck --track-origins=yes
3738 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3739 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3740 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3741 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3742 263099 callgrind_annotate counts Ir improperly [...]
3743 263877 undefined coprocessor instruction on ARMv7
3744 265964 configure FAIL with glibc-2.13
3745 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3746 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3747 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3748 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3749 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3751 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3755 Release 3.6.0 (21 October 2010)
3756 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3757 3.6.0 is a feature release with many significant improvements and the
3758 usual collection of bug fixes.
3760 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3761 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3762 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3764 -------------------------
3766 Here are some highlights. Details are shown further down:
3768 * Support for ARM/Linux.
3770 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3772 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3774 * Support for the SSE4.2 instruction set.
3776 * Enhancements to the Callgrind profiler, including the ability to
3777 handle CPUs with three levels of cache.
3779 * A new experimental heap profiler, DHAT.
3781 * A huge number of bug fixes and small enhancements.
3783 -------------------------
3785 Here are details of the above changes, together with descriptions of
3786 many other changes, and a list of fixed bugs.
3788 * ================== PLATFORM CHANGES =================
3790 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3791 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3792 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3794 This requires a CPU capable of running the ARMv7-A instruction set
3795 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3796 of the user space instruction set, including ARM and Thumb integer
3797 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3798 Cachegrind and Massif tools work properly; other tools work to
3801 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3802 with support for recent releases of the underlying toolchain
3803 components, notably gcc-4.5 and glibc-2.12.
3805 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3806 support also works much better on OS X 10.5, and is as solid as
3809 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3810 64-bit mode. In 32-bit mode, support is only available up to and
3811 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3812 supported in 64-bit mode, and 32-bit mode does in fact support the
3813 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3814 10.6 on 32-bit targets.
3816 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3817 and including version 2.05 is supported.
3819 * ==================== TOOL CHANGES ====================
3821 * Cachegrind has a new processing script, cg_diff, which finds the
3822 difference between two profiles. It's very useful for evaluating
3823 the performance effects of a change in a program.
3825 Related to this change, the meaning of cg_annotate's (rarely-used)
3826 --threshold option has changed; this is unlikely to affect many
3827 people, if you do use it please see the user manual for details.
3829 * Callgrind now can do branch prediction simulation, similar to
3830 Cachegrind. In addition, it optionally can count the number of
3831 executed global bus events. Both can be used for a better
3832 approximation of a "Cycle Estimation" as derived event (you need to
3833 update the event formula in KCachegrind yourself).
3835 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3836 rather than the L2 cache. This is to accommodate machines with
3837 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3838 cache configuration of such a machine it will run the simulation as
3839 if the L2 cache isn't present. This means the results are less
3840 likely to match the true result for the machine, but
3841 Cachegrind/Callgrind's results are already only approximate, and
3842 should not be considered authoritative. The results are still
3843 useful for giving a general idea about a program's locality.
3845 * Massif has a new option, --pages-as-heap, which is disabled by
3846 default. When enabled, instead of tracking allocations at the level
3847 of heap blocks (as allocated with malloc/new/new[]), it instead
3848 tracks memory allocations at the level of memory pages (as mapped by
3849 mmap, brk, etc). Each mapped page is treated as its own block.
3850 Interpreting the page-level output is harder than the heap-level
3851 output, but this option is useful if you want to account for every
3852 byte of memory used by a program.
3854 * DRD has two new command-line options: --free-is-write and
3855 --trace-alloc. The former allows to detect reading from already freed
3856 memory, and the latter allows tracing of all memory allocations and
3859 * DRD has several new annotations. Custom barrier implementations can
3860 now be annotated, as well as benign races on static variables.
3862 * DRD's happens before / happens after annotations have been made more
3863 powerful, so that they can now also be used to annotate e.g. a smart
3864 pointer implementation.
3866 * Helgrind's annotation set has also been drastically improved, so as
3867 to provide to users a general set of annotations to describe locks,
3868 semaphores, barriers and condition variables. Annotations to
3869 describe thread-safe reference counted heap objects have also been
3872 * Memcheck has a new command-line option, --show-possibly-lost, which
3873 is enabled by default. When disabled, the leak detector will not
3874 show possibly-lost blocks.
3876 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3877 has been added. DHAT keeps track of allocated heap blocks, and also
3878 inspects every memory reference to see which block (if any) is being
3879 accessed. This gives a lot of insight into block lifetimes,
3880 utilisation, turnover, liveness, and the location of hot and cold
3881 fields. You can use DHAT to do hot-field profiling.
3883 * ==================== OTHER CHANGES ====================
3885 * Improved support for unfriendly self-modifying code: the extra
3886 overhead incurred by --smc-check=all has been reduced by
3887 approximately a factor of 5 as compared with 3.5.0.
3889 * Ability to show directory names for source files in error messages.
3890 This is combined with a flexible mechanism for specifying which
3891 parts of the paths should be shown. This is enabled by the new flag
3894 * A new flag, --require-text-symbol, which will stop the run if a
3895 specified symbol is not found it a given shared object when it is
3896 loaded into the process. This makes advanced working with function
3897 intercepting and wrapping safer and more reliable.
3899 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3900 and control of Valgrind is now available for the tools Memcheck and
3901 Helgrind. XML output from Valgrind is available for Memcheck,
3902 Helgrind and exp-Ptrcheck.
3904 * More reliable stack unwinding on amd64-linux, particularly in the
3905 presence of function wrappers, and with gcc-4.5 compiled code.
3907 * Modest scalability (performance improvements) for massive
3908 long-running applications, particularly for those with huge amounts
3911 * Support for analyzing programs running under Wine with has been
3912 improved. The header files <valgrind/valgrind.h>,
3913 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3914 Windows-programs compiled with MinGW or one of the Microsoft Visual
3917 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3918 The 32-bit simulator was not affected. This did not occur often,
3919 but when it did would usually crash the program under test.
3922 * A large number of bugs were fixed. These are shown below.
3924 * A number of bugs were investigated, and were candidates for fixing,
3925 but are not fixed in 3.6.0, due to lack of developer time. They may
3926 get fixed in later releases. They are:
3928 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3929 212419 false positive "lock order violated" (A+B vs A)
3930 213685 Undefined value propagates past dependency breaking instruction
3931 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3932 237920 valgrind segfault on fork failure
3933 242137 support for code compiled by LLVM-2.8
3934 242423 Another unknown Intel cache config value
3935 243232 Inconsistent Lock Orderings report with trylock
3936 243483 ppc: callgrind triggers VEX assertion failure
3937 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3938 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3940 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3941 249435 Analyzing wine programs with callgrind triggers a crash
3942 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3943 250065 Handling large allocations
3944 250101 huge "free" memory usage due to m_mallocfree.c
3945 "superblocks fragmentation"
3946 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3947 252091 Callgrind on ARM does not detect function returns correctly
3948 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3949 254420 memory pool tracking broken
3950 n-i-bz support for adding symbols for JIT generated code
3953 The following bugs have been fixed or resolved. Note that "n-i-bz"
3954 stands for "not in bugzilla" -- that is, a bug that was reported to us
3955 but never got a bugzilla entry. We encourage you to file bugs in
3956 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3957 mailing the developers (or mailing lists) directly -- bugs that are
3958 not entered into bugzilla tend to get forgotten about or ignored.
3960 To see details of a given bug, visit
3961 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3962 where XXXXXX is the bug number as listed below.
3964 135264 dcbzl instruction missing
3966 153699 Valgrind should report unaligned reads with movdqa
3968 190429 Valgrind reports lost of errors in ld.so
3969 with x86_64 2.9.90 glibc
3970 197266 valgrind appears to choke on the xmms instruction
3972 197988 Crash when demangling very large symbol names
3973 202315 unhandled syscall: 332 (inotify_init1)
3974 203256 Add page-level profiling to Massif
3975 205093 dsymutil=yes needs quotes, locking (partial fix)
3976 205241 Snow Leopard 10.6 support (partial fix)
3977 206600 Leak checker fails to upgrade indirect blocks when their
3978 parent becomes reachable
3979 210935 port valgrind.h (not valgrind) to win32 so apps run under
3980 wine can make client requests
3981 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3982 within Linux ip-stack checksum functions
3983 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3985 213685 Undefined value propagates past dependency breaking instruction
3987 215914 Valgrind inserts bogus empty environment variable
3989 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3990 222545 shmat fails under valgind on some arm targets
3991 222560 ARM NEON support
3994 232509 Docs build fails with formatting inside <title></title> elements
3996 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3997 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3998 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3999 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
4000 237485 symlink (syscall 57) is not supported on Mac OS
4001 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
4003 238208 is_just_below_ESP doesn't take into account red-zone
4004 238345 valgrind passes wrong $0 when executing a shell script
4005 238679 mq_timedreceive syscall doesn't flag the reception buffer
4007 238696 fcntl command F_DUPFD_CLOEXEC not supported
4008 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4009 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4010 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
4012 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
4018 242606 unhandled syscall: setegid (in Ptrcheck)
4019 242814 Helgrind "Impossible has happened" during
4020 QApplication::initInstance();
4021 243064 Valgrind attempting to read debug information from iso
4022 243270 Make stack unwinding in Valgrind wrappers more reliable
4023 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
4024 sysno = 277 (mq_open)
4025 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
4026 244493 ARM VFP d16-d31 registers support
4027 244670 add support for audit_session_self syscall on Mac OS 10.6
4028 244921 The xml report of helgrind tool is not well format
4029 244923 In the xml report file, the <preamble> not escape the
4030 xml char, eg '<','&','>'
4031 245535 print full path names in plain text reports
4032 245925 x86-64 red zone handling problem
4033 246258 Valgrind not catching integer underruns + new [] s
4034 246311 reg/reg cmpxchg doesn't work on amd64
4035 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
4036 246888 Improve Makefile.vex.am
4037 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
4038 to [f]chmod_extended
4039 247526 IBM POWER6 (ISA 2.05) support is incomplete
4040 247561 Some leak testcases fails due to reachable addresses in
4042 247875 sizeofIRType to handle Ity_I128
4043 247894 [PATCH] unhandled syscall sys_readahead
4044 247980 Doesn't honor CFLAGS passed to configure
4045 248373 darwin10.supp is empty in the trunk
4046 248822 Linux FIBMAP ioctl has int parameter instead of long
4047 248893 [PATCH] make readdwarf.c big endianess safe to enable
4048 unwinding on big endian systems
4049 249224 Syscall 336 not supported (SYS_proc_info)
4051 249775 Incorrect scheme for detecting NEON capabilities of host CPU
4052 249943 jni JVM init fails when using valgrind
4053 249991 Valgrind incorrectly declares AESKEYGENASSIST support
4055 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
4056 250799 frexp$fenv_access_off function generates SIGILL
4057 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
4058 251251 support pclmulqdq insn
4059 251362 valgrind: ARM: attach to debugger either fails or provokes
4061 251674 Unhandled syscall 294
4064 254257 Add support for debugfiles found by build-id
4065 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
4066 254646 Wrapped functions cause stack misalignment on OS X
4067 (and possibly Linux)
4068 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
4070 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
4074 Release 3.5.0 (19 August 2009)
4075 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4076 3.5.0 is a feature release with many significant improvements and the
4077 usual collection of bug fixes. The main improvement is that Valgrind
4078 now works on Mac OS X.
4080 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
4081 and X86/Darwin. Support for recent distros and toolchain components
4082 (glibc 2.10, gcc 4.5) has been added.
4084 -------------------------
4086 Here is a short summary of the changes. Details are shown further
4089 * Support for Mac OS X (10.5.x).
4091 * Improvements and simplifications to Memcheck's leak checker.
4093 * Clarification and simplifications in various aspects of Valgrind's
4096 * XML output for Helgrind and Ptrcheck.
4098 * Performance and stability improvements for Helgrind and DRD.
4100 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
4102 * A new experimental tool, BBV, useful for computer architecture
4105 * Improved Wine support, including ability to read Windows PDB
4108 -------------------------
4110 Here are details of the above changes, followed by descriptions of
4111 many other minor changes, and a list of fixed bugs.
4114 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
4115 called "Darwin" because that is the name of the OS core, which is the
4116 level that Valgrind works at.)
4120 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
4121 because it would require work and 10.4 is only becoming less common.
4123 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
4124 fairly well. For 10.5.x, 32-bit programs are the default even on
4125 64-bit machines, so it handles most current programs.
4127 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
4128 officially supported, but simple programs at least will probably work.
4129 However, start-up is slow.
4131 - PowerPC machines are not supported.
4133 Things that don't work:
4135 - The Ptrcheck tool.
4137 - Objective-C garbage collection.
4141 - If you have Rogue Amoeba's "Instant Hijack" program installed,
4142 Valgrind will fail with a SIGTRAP at start-up. See
4143 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
4148 - You will likely find --dsymutil=yes a useful option, as error
4149 messages may be imprecise without it.
4151 - Mac OS X support is new and therefore will be less robust than the
4152 Linux support. Please report any bugs you find.
4154 - Threaded programs may run more slowly than on Linux.
4156 Many thanks to Greg Parker for developing this port over several years.
4159 * Memcheck's leak checker has been improved.
4161 - The results for --leak-check=summary now match the summary results
4162 for --leak-check=full. Previously they could differ because
4163 --leak-check=summary counted "indirectly lost" blocks and
4164 "suppressed" blocks as "definitely lost".
4166 - Blocks that are only reachable via at least one interior-pointer,
4167 but are directly pointed to by a start-pointer, were previously
4168 marked as "still reachable". They are now correctly marked as
4171 - The default value for the --leak-resolution option has been
4172 changed from "low" to "high". In general, this means that more
4173 leak reports will be produced, but each leak report will describe
4174 fewer leaked blocks.
4176 - With --leak-check=full, "definitely lost" and "possibly lost"
4177 leaks are now considered as proper errors, ie. they are counted
4178 for the "ERROR SUMMARY" and affect the behaviour of
4179 --error-exitcode. These leaks are not counted as errors if
4180 --leak-check=summary is specified, however.
4182 - Documentation for the leak checker has been improved.
4185 * Various aspects of Valgrind's text output have changed.
4187 - Valgrind's start-up message has changed. It is shorter but also
4188 includes the command being run, which makes it easier to use
4189 --trace-children=yes. An example:
4191 - Valgrind's shut-down messages have also changed. This is most
4192 noticeable with Memcheck, where the leak summary now occurs before
4193 the error summary. This change was necessary to allow leaks to be
4194 counted as proper errors (see the description of the leak checker
4195 changes above for more details). This was also necessary to fix a
4196 longstanding bug in which uses of suppressions against leaks were
4197 not "counted", leading to difficulties in maintaining suppression
4198 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
4200 - Behavior of -v has changed. In previous versions, -v printed out
4201 a mixture of marginally-user-useful information, and tool/core
4202 statistics. The statistics printing has now been moved to its own
4203 flag, --stats=yes. This means -v is less verbose and more likely
4204 to convey useful end-user information.
4206 - The format of some (non-XML) stack trace entries has changed a
4207 little. Previously there were six possible forms:
4209 0x80483BF: really (a.c:20)
4210 0x80483BF: really (in /foo/a.out)
4212 0x80483BF: (within /foo/a.out)
4213 0x80483BF: ??? (a.c:20)
4216 The third and fourth of these forms have been made more consistent
4217 with the others. The six possible forms are now:
4219 0x80483BF: really (a.c:20)
4220 0x80483BF: really (in /foo/a.out)
4221 0x80483BF: really (in ???)
4222 0x80483BF: ??? (in /foo/a.out)
4223 0x80483BF: ??? (a.c:20)
4226 Stack traces produced when --xml=yes is specified are different
4230 * Helgrind and Ptrcheck now support XML output, so they can be used
4231 from GUI tools. Also, the XML output mechanism has been
4234 - The XML format has been overhauled and generalised, so it is more
4235 suitable for error reporting tools in general. The Memcheck
4236 specific aspects of it have been removed. The new format, which
4237 is an evolution of the old format, is described in
4238 docs/internals/xml-output-protocol4.txt.
4240 - Memcheck has been updated to use the new format.
4242 - Helgrind and Ptrcheck are now able to emit output in this format.
4244 - The XML output mechanism has been overhauled. XML is now output
4245 to its own file descriptor, which means that:
4247 * Valgrind can output text and XML independently.
4249 * The longstanding problem of XML output being corrupted by
4250 unexpected un-tagged text messages is solved.
4252 As before, the destination for text output is specified using
4253 --log-file=, --log-fd= or --log-socket=.
4255 As before, XML output for a tool is enabled using --xml=yes.
4257 Because there's a new XML output channel, the XML output
4258 destination is now specified by --xml-file=, --xml-fd= or
4261 Initial feedback has shown this causes some confusion. To
4262 clarify, the two envisaged usage scenarios are:
4264 (1) Normal text output. In this case, do not specify --xml=yes
4265 nor any of --xml-file=, --xml-fd= or --xml-socket=.
4267 (2) XML output. In this case, specify --xml=yes, and one of
4268 --xml-file=, --xml-fd= or --xml-socket= to select the XML
4269 destination, one of --log-file=, --log-fd= or --log-socket=
4270 to select the destination for any remaining text messages,
4271 and, importantly, -q.
4273 -q makes Valgrind completely silent on the text channel,
4274 except in the case of critical failures, such as Valgrind
4275 itself segfaulting, or failing to read debugging information.
4276 Hence, in this scenario, it suffices to check whether or not
4277 any output appeared on the text channel. If yes, then it is
4278 likely to be a critical error which should be brought to the
4279 attention of the user. If no (the text channel produced no
4280 output) then it can be assumed that the run was successful.
4282 This allows GUIs to make the critical distinction they need to
4283 make (did the run fail or not?) without having to search or
4284 filter the text output channel in any way.
4286 It is also recommended to use --child-silent-after-fork=yes in
4290 * Improvements and changes in Helgrind:
4292 - XML output, as described above
4294 - Checks for consistent association between pthread condition
4295 variables and their associated mutexes are now performed.
4297 - pthread_spinlock functions are supported.
4299 - Modest performance improvements.
4301 - Initial (skeletal) support for describing the behaviour of
4302 non-POSIX synchronisation objects through ThreadSanitizer
4303 compatible ANNOTATE_* macros.
4305 - More controllable tradeoffs between performance and the level of
4306 detail of "previous" accesses in a race. There are now three
4309 * --history-level=full. This is the default, and was also the
4310 default in 3.4.x. It shows both stacks involved in a race, but
4311 requires a lot of memory and can be very slow in programs that
4312 do many inter-thread synchronisation events.
4314 * --history-level=none. This only shows the later stack involved
4315 in a race. This can be much faster than --history-level=full,
4316 but makes it much more difficult to find the other access
4317 involved in the race.
4319 The new intermediate setting is
4321 * --history-level=approx
4323 For the earlier (other) access, two stacks are presented. The
4324 earlier access is guaranteed to be somewhere in between the two
4325 program points denoted by those stacks. This is not as useful
4326 as showing the exact stack for the previous access (as per
4327 --history-level=full), but it is better than nothing, and it's
4328 almost as fast as --history-level=none.
4331 * New features and improvements in DRD:
4333 - The error messages printed by DRD are now easier to interpret.
4334 Instead of using two different numbers to identify each thread
4335 (Valgrind thread ID and DRD thread ID), DRD does now identify
4336 threads via a single number (the DRD thread ID). Furthermore
4337 "first observed at" information is now printed for all error
4338 messages related to synchronization objects.
4340 - Added support for named semaphores (sem_open() and sem_close()).
4342 - Race conditions between pthread_barrier_wait() and
4343 pthread_barrier_destroy() calls are now reported.
4345 - Added support for custom allocators through the macros
4346 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
4347 in <valgrind/valgrind.h>). An alternative for these two macros is
4348 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
4351 - Added support for annotating non-POSIX synchronization objects
4352 through several new ANNOTATE_*() macros.
4354 - OpenMP: added support for the OpenMP runtime (libgomp) included
4355 with gcc versions 4.3.0 and 4.4.0.
4359 - Added two new command-line options (--first-race-only and
4360 --segment-merging-interval).
4363 * Genuinely atomic support for x86/amd64/ppc atomic instructions
4365 Valgrind will now preserve (memory-access) atomicity of LOCK-
4366 prefixed x86/amd64 instructions, and any others implying a global
4367 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
4369 This means that Valgrinded processes will "play nicely" in
4370 situations where communication with other processes, or the kernel,
4371 is done through shared memory and coordinated with such atomic
4372 instructions. Prior to this change, such arrangements usually
4373 resulted in hangs, races or other synchronisation failures, because
4374 Valgrind did not honour atomicity of such instructions.
4377 * A new experimental tool, BBV, has been added. BBV generates basic
4378 block vectors for use with the SimPoint analysis tool, which allows
4379 a program's overall behaviour to be approximated by running only a
4380 fraction of it. This is useful for computer architecture
4381 researchers. You can run BBV by specifying --tool=exp-bbv (the
4382 "exp-" prefix is short for "experimental"). BBV was written by
4386 * Modestly improved support for running Windows applications under
4387 Wine. In particular, initial support for reading Windows .PDB debug
4388 information has been added.
4391 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
4392 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
4396 * The Valgrind client requests VALGRIND_PRINTF and
4397 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
4398 the string was always printed immediately on its own line. Now, the
4399 string will be added to a buffer but not printed until a newline is
4400 encountered, or other Valgrind output is printed (note that for
4401 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
4402 "other Valgrind output"). This allows you to use multiple
4403 VALGRIND_PRINTF calls to build up a single output line, and also to
4404 print multiple output lines with a single request (by embedding
4405 multiple newlines in the string).
4408 * The graphs drawn by Massif's ms_print program have changed slightly:
4410 - The half-height chars '.' and ',' are no longer drawn, because
4411 they are confusing. The --y option can be used if the default
4412 y-resolution is not high enough.
4414 - Horizontal lines are now drawn after the top of a snapshot if
4415 there is a gap until the next snapshot. This makes it clear that
4416 the memory usage has not dropped to zero between snapshots.
4419 * Something that happened in 3.4.0, but wasn't clearly announced: the
4420 option --read-var-info=yes can be used by some tools (Memcheck,
4421 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
4422 variable type and location information. This makes those tools
4423 start up more slowly and increases memory consumption, but
4424 descriptions of data addresses in error messages become more
4428 * exp-Omega, an experimental instantaneous leak-detecting tool, was
4429 disabled in 3.4.0 due to a lack of interest and maintenance,
4430 although the source code was still in the distribution. The source
4431 code has now been removed from the distribution. For anyone
4432 interested, the removal occurred in SVN revision r10247.
4435 * Some changes have been made to the build system.
4437 - VEX/ is now integrated properly into the build system. This means
4438 that dependency tracking within VEX/ now works properly, "make
4439 install" will work without requiring "make" before it, and
4440 parallel builds (ie. 'make -j') now work (previously a
4441 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
4442 was effectively ignored).
4444 - The --with-vex configure option has been removed. It was of
4445 little use and removing it simplified the build system.
4447 - The location of some install files has changed. This should not
4448 affect most users. Those who might be affected:
4450 * For people who use Valgrind with MPI programs, the installed
4451 libmpiwrap.so library has moved from
4452 $(INSTALL)/<platform>/libmpiwrap.so to
4453 $(INSTALL)/libmpiwrap-<platform>.so.
4455 * For people who distribute standalone Valgrind tools, the
4456 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
4457 have moved to $(INSTALL)/libcoregrind-<platform>.a.
4459 These changes simplify the build system.
4461 - Previously, all the distributed suppression (*.supp) files were
4462 installed. Now, only default.supp is installed. This should not
4463 affect users as the other installed suppression files were not
4464 read; the fact that they were installed was a mistake.
4467 * KNOWN LIMITATIONS:
4469 - Memcheck is unusable with the Intel compiler suite version 11.1,
4470 when it generates code for SSE2-and-above capable targets. This
4471 is because of icc's use of highly optimised inlined strlen
4472 implementations. It causes Memcheck to report huge numbers of
4473 false errors even in simple programs. Helgrind and DRD may also
4476 Versions 11.0 and earlier may be OK, but this has not been
4480 The following bugs have been fixed or resolved. Note that "n-i-bz"
4481 stands for "not in bugzilla" -- that is, a bug that was reported to us
4482 but never got a bugzilla entry. We encourage you to file bugs in
4483 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4484 mailing the developers (or mailing lists) directly -- bugs that are
4485 not entered into bugzilla tend to get forgotten about or ignored.
4487 To see details of a given bug, visit
4488 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4489 where XXXXXX is the bug number as listed below.
4491 84303 How about a LockCheck tool?
4492 91633 dereference of null ptr in vgPlain_st_basetype
4493 97452 Valgrind doesn't report any pthreads problems
4494 100628 leak-check gets assertion failure when using
4495 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4496 108528 NPTL pthread cleanup handlers not called
4497 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4498 110128 mallinfo is not implemented...
4499 110770 VEX: Generated files not always updated when making valgrind
4500 111102 Memcheck: problems with large (memory footprint) applications
4501 115673 Vex's decoder should never assert
4502 117564 False positive: Syscall param clone(child_tidptr) contains
4503 uninitialised byte(s)
4504 119404 executing ssh from inside valgrind fails
4505 133679 Callgrind does not write path names to sources with dwarf debug
4507 135847 configure.in problem with non gnu compilers (and possible fix)
4508 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4509 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4510 136230 memcheck reports "possibly lost", should be "still reachable"
4511 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4512 137904 Valgrind reports a memory leak when using POSIX threads,
4514 139076 valgrind VT_GETSTATE error
4515 142228 complaint of elf_dynamic_do_rela in trivial usage
4516 145347 spurious warning with USBDEVFS_REAPURB
4517 148441 (wine) can't find memory leak in Wine, win32 binary
4519 148742 Leak-check fails assert on exit
4520 149878 add (proper) check for calloc integer overflow
4521 150606 Call graph is broken when using callgrind control
4522 152393 leak errors produce an exit code of 0. I need some way to
4523 cause leak errors to result in a nonzero exit code.
4524 157154 documentation (leak-resolution doc speaks about num-callers
4525 def=4) + what is a loss record
4526 159501 incorrect handling of ALSA ioctls
4527 162020 Valgrinding an empty/zero-byte file crashes valgrind
4528 162482 ppc: Valgrind crashes while reading stabs information
4529 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4530 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4531 163560 VEX/test_main.c is missing from valgrind-3.3.1
4532 164353 malloc_usable_size() doesn't return a usable size
4533 165468 Inconsistent formatting in memcheck manual -- please fix
4534 169505 main.c:286 (endOfInstr):
4535 Assertion 'ii->cost_offset == *cost_offset' failed
4536 177206 Generate default.supp during compile instead of configure
4537 177209 Configure valt_load_address based on arch+os
4538 177305 eventfd / syscall 323 patch lost
4539 179731 Tests fail to build because of inlining of non-local asm labels
4540 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4541 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4542 181594 Bogus warning for empty text segment
4543 181707 dwarf doesn't require enumerations to have name
4544 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4545 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4546 Assertion '!already_present' failed.
4547 185359 exp-ptrcheck: unhandled syscall getresuid()
4548 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4549 185816 Valgrind is unable to handle debug info for files with split
4550 debug info that are prelinked afterwards
4551 185980 [darwin] unhandled syscall: sem_open
4552 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4553 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4554 186790 Suppression pattern used for leaks are not reported
4555 186796 Symbols with length>200 in suppression files are ignored
4556 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4557 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4558 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4559 188046 bashisms in the configure script
4560 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4561 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4562 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4563 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4564 assertion fail "!lock->heldBy"
4565 188427 Add support for epoll_create1 (with patch)
4566 188530 Support for SIOCGSTAMPNS
4567 188560 Include valgrind.spec in the tarball
4568 188572 Valgrind on Mac should suppress setenv() mem leak
4569 189054 Valgrind fails to build because of duplicate non-local asm labels
4570 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4571 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4572 189763 drd assertion failure: s_threadinfo[tid].is_recording
4573 190219 unhandled syscall: 328 (x86-linux)
4574 190391 dup of 181394; see above
4575 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4576 190820 No debug information on powerpc-linux
4577 191095 PATCH: Improve usbdevfs ioctl handling
4578 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4580 191189 --xml=yes should obey --gen-suppressions=all
4581 191192 syslog() needs a suppression on macosx
4582 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4583 191761 getrlimit on MacOSX
4584 191992 multiple --fn-skip only works sometimes; dependent on order
4585 192634 V. reports "aspacem sync_check_mapping_callback:
4586 segment mismatch" on Darwin
4587 192954 __extension__ missing on 2 client requests
4588 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4589 194474 "INSTALL" file has different build instructions than "README"
4590 194671 Unhandled syscall (sem_wait?) from mac valgrind
4591 195069 memcheck: reports leak (memory still reachable) for
4593 195169 drd: (vgDrd_barrier_post_wait):
4594 Assertion 'r->sg[p->post_iteration]' failed.
4595 195268 valgrind --log-file doesn't accept ~/...
4596 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4597 195860 WARNING: unhandled syscall: unix:223
4598 196528 need a error suppression for pthread_rwlock_init under os x?
4599 197227 Support aio_* syscalls on Darwin
4600 197456 valgrind should reject --suppressions=(directory)
4601 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4602 197591 unhandled syscall 27 (mincore)
4603 197793 Merge DCAS branch to the trunk == 85756, 142103
4604 197794 Avoid duplicate filenames in Vex
4605 197898 make check fails on current SVN
4606 197901 make check fails also under exp-ptrcheck in current SVN
4607 197929 Make --leak-resolution=high the default
4608 197930 Reduce spacing between leak reports
4609 197933 Print command line of client at start-up, and shorten preamble
4610 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4611 198395 add BBV to the distribution as an experimental tool
4612 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4613 198649 callgrind_annotate doesn't cumulate counters
4614 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4615 199977 Valgrind complains about an unrecognized instruction in the
4616 atomic_incs test program
4617 200029 valgrind isn't able to read Fedora 12 debuginfo
4618 200760 darwin unhandled syscall: unix:284
4619 200827 DRD doesn't work on Mac OS X
4620 200990 VG_(read_millisecond_timer)() does not work correctly
4621 201016 Valgrind does not support pthread_kill() on Mac OS
4622 201169 Document --read-var-info
4623 201323 Pre-3.5.0 performance sanity checking
4624 201384 Review user manual for the 3.5.0 release
4625 201585 mfpvr not implemented on ppc
4626 201708 tests failing because x86 direction flag is left set
4627 201757 Valgrind doesn't handle any recent sys_futex additions
4628 204377 64-bit valgrind can not start a shell script
4629 (with #!/path/to/shell) if the shell is a 32-bit executable
4630 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4631 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4632 about memory allocation events with memory access tracing enabled
4633 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4635 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4639 Release 3.4.1 (28 February 2009)
4640 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4641 3.4.1 is a bug-fix release that fixes some regressions and assertion
4642 failures in debug info reading in 3.4.0, most notably incorrect stack
4643 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4644 other debug info problems are also fixed. A number of bugs in the
4645 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4647 In view of the fact that 3.4.0 contains user-visible regressions
4648 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4649 encouraged to ship 3.4.1 in preference to 3.4.0.
4651 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4652 bugzilla" -- that is, a bug that was reported to us but never got a
4653 bugzilla entry. We encourage you to file bugs in bugzilla
4654 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4655 developers (or mailing lists) directly -- bugs that are not entered
4656 into bugzilla tend to get forgotten about or ignored.
4658 n-i-bz Fix various bugs reading icc-11 generated debug info
4659 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4660 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4661 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4662 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4663 179618 exp-ptrcheck crashed / exit prematurely
4664 179624 helgrind: false positive races with pthread_create and
4665 recv/open/close/read
4666 134207 pkg-config output contains @VG_PLATFORM@
4667 176926 floating point exception at valgrind startup with PPC 440EPX
4668 181594 Bogus warning for empty text segment
4669 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4670 181707 Dwarf3 doesn't require enumerations to have name
4671 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4672 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4673 Assertion '!already_present' failed.
4674 185359 exp-ptrcheck unhandled syscall getresuid()
4676 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4677 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4681 Release 3.4.0 (2 January 2009)
4682 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4683 3.4.0 is a feature release with many significant improvements and the
4684 usual collection of bug fixes. This release supports X86/Linux,
4685 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4686 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4688 3.4.0 brings some significant tool improvements. Memcheck can now
4689 report the origin of uninitialised values, the thread checkers
4690 Helgrind and DRD are much improved, and we have a new experimental
4691 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4692 global arrays. In detail:
4694 * Memcheck is now able to track the origin of uninitialised values.
4695 When it reports an uninitialised value error, it will try to show
4696 the origin of the value, as either a heap or stack allocation.
4697 Origin tracking is expensive and so is not enabled by default. To
4698 use it, specify --track-origins=yes. Memcheck's speed will be
4699 essentially halved, and memory usage will be significantly
4700 increased. Nevertheless it can drastically reduce the effort
4701 required to identify the root cause of uninitialised value errors,
4702 and so is often a programmer productivity win, despite running more
4705 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4706 3.4.0, will be released shortly.
4708 * Helgrind's race detection algorithm has been completely redesigned
4709 and reimplemented, to address usability and scalability concerns:
4711 - The new algorithm has a lower false-error rate: it is much less
4712 likely to report races that do not really exist.
4714 - Helgrind will display full call stacks for both accesses involved
4715 in a race. This makes it easier to identify the root causes of
4718 - Limitations on the size of program that can run have been removed.
4720 - Performance has been modestly improved, although that is very
4723 - Direct support for Qt4 threading has been added.
4725 - pthread_barriers are now directly supported.
4727 - Helgrind works well on all supported Linux targets.
4729 * The DRD thread debugging tool has seen major improvements:
4731 - Greatly improved performance and significantly reduced memory
4734 - Support for several major threading libraries (Boost.Thread, Qt4,
4735 glib, OpenMP) has been added.
4737 - Support for atomic instructions, POSIX semaphores, barriers and
4738 reader-writer locks has been added.
4740 - Works now on PowerPC CPUs too.
4742 - Added support for printing thread stack usage at thread exit time.
4744 - Added support for debugging lock contention.
4746 - Added a manual for Drd.
4748 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4749 checks for misuses of pointers. In that sense it is a bit like
4750 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4751 detect overruns of stack and global arrays, it can detect
4752 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4753 detect accesses heap blocks that have been freed a very long time
4754 ago (millions of blocks in the past).
4756 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4757 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4758 of the main Valgrind documentation. As this is an experimental
4759 tool, we would be particularly interested in hearing about your
4760 experiences with it.
4762 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4763 longer built by default, although the code remains in the repository
4764 and the tarball. This is due to three factors: a perceived lack of
4765 users, a lack of maintenance, and concerns that it may not be
4766 possible to achieve reliable operation using the existing design.
4768 * As usual, support for the latest Linux distros and toolchain
4769 components has been added. It should work well on Fedora Core 10,
4770 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4771 state) is supported, as is glibc-2.9. The C++ demangler has been
4772 updated so as to work well with C++ compiled by even the most recent
4775 * You can now use frame-level wildcards in suppressions. This was a
4776 frequently-requested enhancement. A line "..." in a suppression now
4777 matches zero or more frames. This makes it easier to write
4778 suppressions which are precise yet insensitive to changes in
4781 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4783 * Very basic support for IBM Power6 has been added (64-bit processes only).
4785 * Valgrind is now cross-compilable. For example, it is possible to
4786 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4787 on a ppc32/64-linux target.
4789 * You can set the main thread's stack size at startup using the
4790 new --main-stacksize= flag (subject of course to ulimit settings).
4791 This is useful for running apps that need a lot of stack space.
4793 * The limitation that you can't use --trace-children=yes together
4794 with --db-attach=yes has been removed.
4796 * The following bugs have been fixed. Note that "n-i-bz" stands for
4797 "not in bugzilla" -- that is, a bug that was reported to us but
4798 never got a bugzilla entry. We encourage you to file bugs in
4799 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4800 mailing the developers (or mailing lists) directly.
4802 n-i-bz Make return types for some client requests 64-bit clean
4803 n-i-bz glibc 2.9 support
4804 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4805 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4806 n-i-bz Building in an env without gdb gives bogus gdb attach
4807 92456 Tracing the origin of uninitialised memory
4808 106497 Valgrind does not demangle some C++ template symbols
4810 151612 Suppression with "..." (frame-level wildcards in .supp files)
4811 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4812 159285 unhandled syscall:25 (stime, on x86-linux)
4813 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4814 160954 ppc build of valgrind crashes with illegal instruction (isel)
4815 160956 mallinfo implementation, w/ patch
4816 162092 Valgrind fails to start gnome-system-monitor
4817 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4818 163794 assertion failure with "--track-origins=yes"
4819 163933 sigcontext.err and .trapno must be set together
4820 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4821 164476 Missing kernel module loading system calls
4822 164669 SVN regression: mmap() drops posix file locks
4823 166581 Callgrind output corruption when program forks
4824 167288 Patch file for missing system calls on Cell BE
4825 168943 unsupported scas instruction pentium
4826 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4827 172417 x86->IR: 0x82 ...
4828 172563 amd64->IR: 0xD9 0xF5 - fprem1
4829 173099 .lds linker script generation error
4830 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4831 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4833 174908 --log-file value not expanded correctly for core file
4834 175044 Add lookup_dcookie for amd64
4835 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4837 Developer-visible changes:
4839 * Valgrind's debug-info reading machinery has been majorly overhauled.
4840 It can now correctly establish the addresses for ELF data symbols,
4841 which is something that has never worked properly before now.
4843 Also, Valgrind can now read DWARF3 type and location information for
4844 stack and global variables. This makes it possible to use the
4845 framework to build tools that rely on knowing the type and locations
4846 of stack and global variables, for example exp-Ptrcheck.
4848 Reading of such information is disabled by default, because most
4849 tools don't need it, and because it is expensive in space and time.
4850 However, you can force Valgrind to read it, using the
4851 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4852 make use of such information, if present, to provide source-level
4853 descriptions of data addresses in the error messages they create.
4855 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4856 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).