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 ===================
14 * ================== PLATFORM CHANGES =================
17 * ==================== TOOL CHANGES ===================
20 * ==================== FIXED BUGS ====================
22 The following bugs have been fixed or resolved. Note that "n-i-bz"
23 stands for "not in bugzilla" -- that is, a bug that was reported to us
24 but never got a bugzilla entry. We encourage you to file bugs in
25 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
26 than mailing the developers (or mailing lists) directly -- bugs that
27 are not entered into bugzilla tend to get forgotten about or ignored.
29 469049 link failure on ppc64 (big endian) valgrind 3.20
30 469146 massif --ignore-fn does not ignore inlined functions
32 To see details of a given bug, visit
33 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
34 where XXXXXX is the bug number as listed above.
36 Release 3.21.0 (28 Apr 2023)
37 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
39 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
40 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
41 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
42 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
43 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
44 AMD64/macOS 10.13 and nanoMIPS/Linux.
46 * ==================== CORE CHANGES ===================
48 * When GDB is used to debug a program running under valgrind using
49 the valgrind gdbserver, GDB will automatically load some
50 python code provided in valgrind defining GDB front end commands
51 corresponding to the valgrind monitor commands.
52 These GDB front end commands accept the same format as
53 the monitor commands directly sent to the Valgrind gdbserver.
54 These GDB front end commands provide a better integration
55 in the GDB command line interface, so as to use for example
56 GDB auto-completion, command specific help, searching for
57 a command or command help matching a regexp, ...
58 For relevant monitor commands, GDB will evaluate arguments
59 to make the use of monitor commands easier.
60 For example, instead of having to print the address of a variable
61 to pass it to a subsequent monitor command, the GDB front end
62 command will evaluate the address argument. It is for example
64 (gdb) memcheck who_points_at &some_struct sizeof(some_struct)
67 $2 = (some_struct_type *) 0x1130a0 <some_struct>
68 (gdb) p sizeof(some_struct)
70 (gdb) monitor who_point_at 0x1130a0 40
72 * The vgdb utility now supports extended-remote protocol when
73 invoked with --multi. In this mode the GDB run command is
74 supported. Which means you don't need to run gdb and valgrind
75 from different terminals. So for example to start your program
76 in gdb and run it under valgrind you can do:
78 (gdb) set remote exec-file prog
80 (gdb) target extended-remote | vgdb --multi
83 * The behaviour of realloc with a size of zero can now
84 be changed for tools that intercept malloc. Those
85 tools are memcheck, helgrind, drd, massif and dhat.
86 Realloc implementations generally do one of two things
87 - free the memory like free() and return NULL
88 (GNU libc and ptmalloc).
89 - either free the memory and then allocate a
90 minimum sized block or just return the
91 original pointer. Return NULL if the
92 allocation of the minimum sized block fails
93 (jemalloc, musl, snmalloc, Solaris, macOS).
94 When Valgrind is configured and built it will
95 try to match the OS and libc behaviour. However
96 if you are using a non-default library to replace
97 malloc and family (e.g., musl on a glibc Linux or
98 tcmalloc on FreeBSD) then you can use a command line
99 option to change the behaviour of Valgrind:
100 --realloc-zero-bytes-frees=yes|no [yes on Linux glibc, no otherwise]
102 * ================== PLATFORM CHANGES =================
104 * Make the address space limit on FreeBSD amd64 128Gbytes
105 (the same as Linux and Solaris, it was 32Gbytes)
107 * ==================== TOOL CHANGES ===================
110 - When doing a delta leak_search, it is now possible to only
111 output the new loss records compared to the previous leak search.
112 This is available in the memcheck monitor command 'leak_search'
113 by specifying the "new" keyword or in your program by using
114 the client request VALGRIND_DO_NEW_LEAK_CHECK.
115 Whenever a "delta" leak search is done (i.e. when specifying
116 "new" or "increased" or "changed" in the monitor command),
117 the new loss records have a "new" marker.
118 - Valgrind now contains python code that defines GDB memcheck
119 front end monitor commands. See CORE CHANGES.
120 - Performs checks for the use of realloc with a size of zero.
121 This is non-portable and a source of errors. If memcheck
122 detects such a usage it will generate an error
123 realloc() with size 0
124 followed by the usual callstacks.
125 A switch has been added to allow this to be turned off:
126 --show-realloc-size-zero=yes|no [yes]
129 - The option ---history-backtrace-size=<number> allows to configure
130 the number of entries to record in the stack traces of "old"
131 accesses. Previously, this number was hardcoded to 8.
132 - Valgrind now contains python code that defines GDB helgrind
133 front end monitor commands. See CORE CHANGES.
136 - `--cache-sim=no` is now the default. The cache simulation is old and
137 unlikely to match any real modern machine. This means only the `Ir`
138 event are gathered by default, but that is by far the most useful
140 - `cg_annotate`, `cg_diff`, and `cg_merge` have been rewritten in
141 Python. As a result, they all have more flexible command line
142 argument handling, e.g. supporting `--show-percs` and
143 `--no-show-percs` forms as well as the existing `--show-percs=yes`
144 and `--show-percs=no`.
145 - `cg_annotate` has some functional changes.
146 - It's much faster, e.g. 3-4x on common cases.
147 - It now supports diffing (with `--diff`, `--mod-filename`, and
148 `--mod-funcname`) and merging (by passing multiple data files).
149 - It now provides more information at the file and function level.
150 There are now "File:function" and "Function:file" sections. These
151 are very useful for programs that use inlining a lot.
152 - Support for user-annotated files and the `-I`/`--include` option
153 has been removed, because it was of little use and blocked other
155 - The `--auto` option is renamed `--annotate`, though the old
156 `--auto=yes`/`--auto=no` forms are still supported.
157 - `cg_diff` and `cg_merge` are now deprecated, because `cg_annotate`
158 now does a better job of diffing and merging.
159 - The Cachegrind output file format has changed very slightly, but in
160 ways nobody is likely to notice.
163 - Valgrind now contains python code that defines GDB callgrind
164 front end monitor commands. See CORE CHANGES.
167 - Valgrind now contains python code that defines GDB massif
168 front end monitor commands. See CORE CHANGES.
171 - A new kind of user request has been added which allows you to
172 override the 1024 byte limit on access count histograms for blocks
173 of memory. The client request is DHAT_HISTOGRAM_MEMORY.
175 * ==================== FIXED BUGS ====================
177 The following bugs have been fixed or resolved. Note that "n-i-bz"
178 stands for "not in bugzilla" -- that is, a bug that was reported to us
179 but never got a bugzilla entry. We encourage you to file bugs in
180 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
181 than mailing the developers (or mailing lists) directly -- bugs that
182 are not entered into bugzilla tend to get forgotten about or ignored.
184 170510 Don't warn about ioctl of size 0 without direction hint
185 241072 List tools in --help output
186 327548 false positive while destroying mutex
187 382034 Testcases build fixes for musl
188 351857 confusing error message about valid command line option
189 374596 inconsistent RDTSCP support on x86_64
190 392331 Spurious lock not held error from inside pthread_cond_timedwait
191 397083 Likely false positive "uninitialised value(s)" for __wmemchr_avx2 and __wmemcmp_avx2_movbe
192 400793 pthread_rwlock_timedwrlock false positive
193 419054 Unhandled syscall getcpu on arm32
194 433873 openat2 syscall unimplemented on Linux
195 434057 Add stdio mode to valgrind's gdbserver
196 435441 valgrind fails to interpose malloc on musl 1.2.2 due to weak symbol name and no libc soname
197 436413 Warn about realloc of size zero
198 439685 compiler warning in callgrind/main.c
199 444110 priv/guest_ppc_toIR.c:36198:31: warning: duplicated 'if' condition.
200 444487 hginfo test detects an extra lock inside data symbol "_rtld_local"
201 444488 Use glibc.pthread.stack_cache_size tunable
202 444568 drd/tests/pth_barrier_thr_cr fails on Fedora 38
203 445743 "The impossible happened: mutex is locked simultaneously by two threads"
204 while using mutexes with priority inheritance and signals
205 449309 Missing loopback device ioctl(s)
206 459476 vgdb: allow address reuse to avoid "address already in use" errorsuse" errors
207 460356 s390: Sqrt32Fx4 -- cannot reduce tree
208 462830 WARNING: unhandled amd64-freebsd syscall: 474
209 463027 broken check for MPX instruction support in assembler
210 464103 Enhancement: add a client request to DHAT to mark memory to be histogrammed
211 464476 Firefox fails to start under Valgrind
212 464609 Valgrind memcheck should support Linux pidfd_open
213 464680 Show issues caused by memory policies like selinux deny_execmem
214 464859 Build failures with GCC-13 (drd tsan_unittest)
215 464969 D language demangling
216 465435 m_libcfile.c:66 (vgPlain_safe_fd): Assertion 'newfd >= VG_(fd_hard_limit)' failed.
217 466104 aligned_alloc problems, part 1
218 467036 Add time cost statistics for Regtest
219 467482 Build failure on aarch64 Alpine
220 467714 fdleak_* and rlimit tests fail when parent process has more than
221 64 descriptors opened
222 467839 Gdbserver: Improve compatibility of library directory name
223 468401 [PATCH] Add a style file for clang-format
224 468556 Build failure for vgdb
225 468606 build: remove "Valgrind relies on GCC" check/output
226 469097 ppc64(be) doesn't support SCV syscall instruction
227 n-i-bz FreeBSD rfork syscall fail with EINVAL or ENOSYS rather than VG_(unimplemented)
229 To see details of a given bug, visit
230 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
231 where XXXXXX is the bug number as listed above.
233 * ==================== KNOWN ISSUES ===================
235 * configure --enable-lto=yes is know to not work in all setups.
236 See bug 469049. Workaround: Build without LTO.
238 (3.21.0.RC1: 14 Apr 2023)
239 (3.21.0.RC2: 21 Apr 2023)
241 Release 3.20.0 (24 Oct 2022)
242 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
244 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
245 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
246 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
247 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
248 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
249 AMD64/macOS 10.13 and nanoMIPS/Linux.
251 * ==================== CORE CHANGES ===================
253 * The option "--vgdb-stop-at=event1,event2,..." accepts the new value abexit.
254 This indicates to invoke gdbserver when your program exits abnormally
255 (i.e. with a non zero exit code).
256 * Fix Rust v0 name demangling.
257 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
258 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
259 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with
261 * The option --enable-debuginfod=<no|yes> [default: yes] has been added on
263 * More DWARF5 support as generated by clang14.
265 * ==================== FIXED BUGS ====================
267 The following bugs have been fixed or resolved. Note that "n-i-bz"
268 stands for "not in bugzilla" -- that is, a bug that was reported to us
269 but never got a bugzilla entry. We encourage you to file bugs in
270 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
271 than mailing the developers (or mailing lists) directly -- bugs that
272 are not entered into bugzilla tend to get forgotten about or ignored.
274 131186 writev reports error in (vector[...])
275 434764 iconv_open causes ld.so v2.28+ to use optimised strncmp
276 446754 Improve error codes from alloc functions under memcheck
277 452274 memcheck crashes with Assertion 'sci->status.what == SsIdle' failed
278 452779 Valgrind fails to build on FreeBSD 13.0 with llvm-devel (15.0.0)
279 453055 shared_timed_mutex drd test fails with "Lock shared failed" message
280 453602 Missing command line option to enable/disable debuginfod
281 452802 Handle lld 9+ split RW PT_LOAD segments correctly
282 454040 s390x: False-positive memcheck:cond in memmem on arch13 systems
283 456171 [PATCH] FreeBSD: Don't record address errors when accessing the 'kern.ps_strings' sysctl struct
284 n-i-bz Implement vgdb invoker on FreeBSD
285 458845 PowerPC: The L field for the dcbf and sync instruction should be
287 458915 Remove register cache to fix 458915 gdbserver causes wrong syscall return
288 459031 Documentation on --error-exitcode incomplete
289 459477 XERROR messages lacks ending '\n' in vgdb
290 462007 Implicit int in none/tests/faultstatus.c
292 To see details of a given bug, visit
293 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
294 where XXXXXX is the bug number as listed above.
296 (3.20.0.RC1: 20 Oct 2022)
299 Release 3.19.0 (11 Apr 2022)
300 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
302 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
303 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
304 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
305 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
306 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
307 AMD64/macOS 10.13 and nanoMIPS/Linux.
309 * ==================== CORE CHANGES ===================
311 * Fix Rust v0 name demangling.
312 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
313 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
314 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with HardenedBSD
316 * ================== PLATFORM CHANGES =================
319 - ignore the "v8.x" architecture levels, only look at actual CPU features
320 present. Fixes mismatch detected between RDMA and atomics features
321 preventing startup on some QEMU configurations.
322 - Implement LD{,A}XP and ST{,L}XP
323 - Fix incorrect code emitted for doubleword CAS.
326 - Fix sys_ipc semtimedop syscall
327 - Fix VFLRX and WFLRX instructions
328 - Fix EXRL instruction with negative offset
331 - Reimplement the vbpermq instruction support to generate less Iops and
332 avoid overflowing internal buffers.
333 - Fix checking for scv support to avoid "Facility 'SCV' unavailable (12),
334 exception" messages in dmsg.
335 - Fix setting condition code for Vector Compare quad word instructions.
336 - Fix fix lxsibzx, lxsihzx and lxsihzx instructions so they only load
337 their respective sized data.
338 - Fix the prefixed stq instruction in PC relative mode.
340 * ==================== TOOL CHANGES ===================
343 - Speed up --track-origins=yes for large (in the range of hundreds to
344 thousands of megabytes) mmap/munmaps.
346 - Several fixes for new versions of libstd++ using new posix try_lock
349 * ==================== FIXED BUGS ====================
351 The following bugs have been fixed or resolved. Note that "n-i-bz"
352 stands for "not in bugzilla" -- that is, a bug that was reported to us
353 but never got a bugzilla entry. We encourage you to file bugs in
354 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
355 than mailing the developers (or mailing lists) directly -- bugs that
356 are not entered into bugzilla tend to get forgotten about or ignored.
358 403802 leak_cpp_interior fails with some reachable blocks different than expected
359 435732 memcheck/tests/leak_cpp_interior fails with gcc11
360 444242 s390x: Valgrind crashes on EXRL with negative offset
361 444399 arm64: unhandled instruction 0xC87F2D89 (LD{,A}XP and ST{,L}XP).
363 444481 gdb_server test failures on s390x
364 444495 dhat/tests/copy fails on s390x
365 444552 memcheck/tests/sem fails on s390x with glibc 2.34
366 444571 PPC, fix the lxsibzx and lxsihzx so they only load their respective
368 444836 PPC, pstq instruction for R=1 is not storing to the correct address.
369 444925 fexecve syscall wrapper not properly implemented
370 445032 valgrind/memcheck crash with SIGSEGV when SIGVTALRM timer used and
372 445211 Fix out of tree builds
373 445300 [PATCH] Fix building tests with Musl
374 445011 SIGCHLD is sent when valgrind uses debuginfod-find
375 445354 arm64 backend: incorrect code emitted for doubleword CAS
376 445415 arm64 front end: alignment checks missing for atomic instructions
377 445504 Using C++ condition_variable results in bogus "mutex is locked simultaneously by two threads" warning
378 445607 Unhandled amd64-freebsd syscall: 247
379 445668 Inline stack frame generation is broken for Rust binaries
380 445916 Demangle Rust v0 symbols with .llvm suffix
381 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false positives
382 446138 DRD/Helgrind with std::timed_mutex::try_lock_until false positives
383 446281 Add a DRD suppression for fwrite
384 446103 Memcheck: `--track-origins=yes` causes extreme slowdowns for large mmap/munmap
385 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false
386 446251 TARGET_SIGNAL_THR added to enum target_signal
387 446823 FreeBSD - missing syscalls when using libzm4
388 447991 s390x: Valgrind indicates illegal instruction on wflrx
389 447995 Valgrind segfault on power10 due to hwcap checking code
390 449483 Powerpc: vcmpgtsq., vcmpgtuq,, vcmpequq. instructions not setting the
391 condition code correctly.
392 449672 ppc64 --track-origins=yes failures because of bad cmov addHRegUse
393 449838 sigsegv liburing the 'impossible' happened for io_uring_setup
394 450025 Powerc: ACC file not implemented as a logical overlay of the VSR
396 450437 Warn for execve syscall with argv or argv[0] being NULL
397 450536 Powerpc: valgrind throws 'facility scv unavailable exception'
398 451626 Syscall param bpf(attr->raw_tracepoint.name) points to unaddressable byte(s)
399 451827 [ppc64le] VEX temporary storage exhausted with several vbpermq instructions
400 451843 valgrind fails to start on a FreeBSD system which enforces W^X
402 To see details of a given bug, visit
403 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
404 where XXXXXX is the bug number as listed above.
406 (3.19.0.RC1: 02 Apr 2022)
407 (3.19.0.RC2: 08 Apr 2022)
410 Release 3.18.0 (15 Oct 2021)
411 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
413 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
414 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
415 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
416 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
417 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
418 AMD64/macOS 10.13 and nanoMIPS/Linux.
420 * ==================== CORE CHANGES ===================
422 * The libiberty demangler has been updated, which brings support for
423 Rust v0 name demangling. [Update: alas, due to a bug, this support
424 isn't working in 3.18.0.]
426 * __libc_freeres isn't called anymore after the program recieves a
427 fatal signal. Causing some internal glibc resources to hang around,
428 but preventing any crashes after the program has ended.
430 * The DWARF reader is now very much faster at startup when just
431 --read-inline-info=yes (the default in most cases) is given.
433 * glibc 2.34, which moved various functions from libpthread.so into
434 libc.so, is now supported.
436 * ================== PLATFORM CHANGES =================
440 - v8.2 scalar and vector FABD, FACGE, FACGT and FADD.
441 - v8.2 FP compare & conditional compare instructions.
442 - Zero variants of v8.2 FP compare instructions.
446 - Support the miscellaneous-instruction-extensions facility 3 and
447 the vector-enhancements facility 2. This enables programs
448 compiled with "-march=arch13" or "-march=z15" to be executed
453 - ISA 3.1 support is now complete
454 - ISA 3.0 support for the darn instruction added.
455 - ISA 3.0 support for the vector system call instruction scv added.
456 - ISA 3.0 support for the copy, paste and cpabort instructions added.
458 * Support for X86/FreeBSD and AMD64/FreeBSD has been added.
460 * ==================== OTHER CHANGES ====================
462 * Memcheck on amd64: minor fixes to remove some false positive
465 * ==================== FIXED BUGS ====================
467 The following bugs have been fixed or resolved. Note that "n-i-bz"
468 stands for "not in bugzilla" -- that is, a bug that was reported to us
469 but never got a bugzilla entry. We encourage you to file bugs in
470 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
471 than mailing the developers (or mailing lists) directly -- bugs that
472 are not entered into bugzilla tend to get forgotten about or ignored.
474 208531 [PATCH]: FreeBSD support for valgrind
475 368960 WARNING: unhandled amd64-linux syscall: 163 (acct)
476 407589 [Linux] Add support for C11 aligned_alloc() and GNU reallocarray()
477 423963 Error in child thread when CLONE_PIDFD is used
478 426148 crash with "impossible happened" when running BPF CO-RE programs
479 429375 PPC ISA 3.1 support is missing, part 9
480 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
481 431306 Update demangler to support Rust v0 name mangling
482 432387 s390x: z15 instructions support
483 433437 FreeBSD support, part 1
484 433438 FreeBSD support, part 2
485 433439 FreeBSD support, part 3
486 433469 FreeBSD support, part 4
487 433473 FreeBSD support, part 5
488 433477 FreeBSD support, part 6
489 433479 FreeBSD support, part 7
490 433504 FreeBSD support, part 8
491 433506 FreeBSD support, part 9
492 433507 FreeBSD support, part 10
493 433508 FreeBSD support, part 11
494 433510 FreeBSD support, part 12
495 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
496 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
497 434296 s390x: False-positive memcheck diagnostics from vector string
499 434840 PPC64 darn instruction not supported
500 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
501 435908 valgrind tries to fetch from deubginfod for files which already
502 have debug information
503 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2
504 439046 valgrind is unusably large when linked with lld
505 439090 Implement close_range(2)
506 439326 Valgrind 3.17.0 won't compile with Intel 2021 oneAPI compilers
507 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
508 440670 unhandled ppc64le-linux syscall: 252 statfs64 and 253 fstatfs64
509 440906 Fix impossible constraint issue in P10 testcase.
510 441512 Remove a unneeded / unnecessary prefix check.
511 441534 Update the expected output for test_isa_3_1_VRT.
512 442061 very slow execution under Fedora 34 (readdwarf3)
513 443031 Gcc -many change requires explicit .machine directives
514 443033 Add support for the ISA 3.0 mcrxrx instruction
515 443034 Sraw, srawi, srad, sradi, mfs
516 443178 Powerpc, test jm-mfspr expected output needs to be updated.
517 443179 Need new test for the lxvx and stxvx instructions on ISA 2.07 and
519 443180 The subnormal test and the ISA 3.0 test generate compiler warnings
520 443314 In the latest GIT version, Valgrind with "--trace-flags" crashes
522 443605 Don't call final_tidyup (__libc_freeres) on FatalSignal
524 To see details of a given bug, visit
525 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
526 where XXXXXX is the bug number as listed below.
528 (3.18.0.RC1: 12 Oct 2021)
529 (3.18.0: 15 Oct 2021)
533 Release 3.17.0 (19 Mar 2021)
534 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
536 3.17.0 fixes a number of bugs and adds some functional changes: support for
537 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
538 some new instructions for Arm64, S390 and POWER. There are also some tool
541 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
542 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
543 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
544 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
545 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
547 * ==================== CORE CHANGES ===================
549 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
552 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
553 debugging information. When a debuginfo file cannot be found locally,
554 Valgrind is able to query debuginfod servers for the file using its
555 build-id. See the user manual for more information about debuginfod support.
557 * ================== PLATFORM CHANGES =================
561 - Inaccuracies resulting from double-rounding in the simulation of
562 floating-point multiply-add/subtract instructions have been fixed. These
563 should now behave exactly as the hardware does.
565 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
566 ongoing. Support for the half-word variants of at least the following
567 instructions has been added:
569 FABS <Vd>.<T>, <Vn>.<T>
571 FNEG <Vd>.<T>, <Vn>.<T>
573 FSQRT <Vd>.<T>, <Vn>.<T>
578 - Implement the new instructions/features that were added to z/Architecture
579 with the vector-enhancements facility 1. Also cover the instructions from
580 the vector-packed-decimal facility that are defined outside the chapter
581 "Vector Decimal Instructions", but not the ones from that chapter itself.
583 For a detailed list of newly supported instructions see the updates to
584 `docs/internals/s390-opcodes.csv'.
586 Since the miscellaneous instruction extensions facility 2 was already
587 added in Valgrind 3.16.0, this completes the support necessary to run
588 general programs built with `--march=z14' under Valgrind. The
589 vector-packed-decimal facility is currently not exploited by the standard
590 toolchain and libraries.
594 - Various bug fixes. Fix for the sync field to limit setting just two of
595 the two bits in the L-field. Fix the write size for the stxsibx and
596 stxsihx instructions. Fix the modsw and modsd instructions.
598 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
599 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
600 conversion instructions are still missing.
602 * ==================== TOOL CHANGES ====================
604 * General tool changes
606 - All the tools and their vgpreload libraries are now installed under
607 libexec because they cannot be executed directly and should be run through
608 the valgrind executable. This should be an internal, not user visible,
609 change, but might impact valgrind packagers.
611 - The --track-fds option now respects -q, --quiet and won't output anything
612 if no file descriptors are leaked. It also won't report the standard stdin
613 (0), stdout (1) or stderr (2) descriptors as being leaked with
614 --trace-fds=yes anymore. To track whether the standard file descriptors
615 are still open at the end of the program run use --trace-fds=all.
619 - DHAT has been extended, with two new modes of operation. The new
620 --mode=copy flag triggers copy profiling, which records calls to memcpy,
621 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
622 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
623 the new dhat/dhat.h file. This is useful for learning more about hot code
624 paths. See the user manual for more information about the new modes.
626 - Because of these changes, DHAT's file format has changed. DHAT output
627 files produced with earlier versions of DHAT will not work with this
628 version of DHAT's viewer, and DHAT output files produced with this version
629 of DHAT will not work with earlier versions of DHAT's viewer.
631 * ==================== FIXED BUGS ====================
633 The following bugs have been fixed or resolved. Note that "n-i-bz"
634 stands for "not in bugzilla" -- that is, a bug that was reported to us
635 but never got a bugzilla entry. We encourage you to file bugs in
636 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
637 than mailing the developers (or mailing lists) directly -- bugs that
638 are not entered into bugzilla tend to get forgotten about or ignored.
640 To see details of a given bug, visit
641 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
642 where XXXXXX is the bug number as listed below.
644 140178 open("/proc/self/exe", ...); doesn't quite work
645 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
646 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
647 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
648 345077 linux syscall execveat support (linux 3.19)
649 361770 Missing F_ADD_SEALS
650 369029 handle linux syscalls sched_getattr and sched_setattr
651 384729 __libc_freeres inhibits cross-platform valgrind
652 388787 Support for C++17 new/delete
653 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
654 396656 Warnings while reading debug info
655 397605 ioctl FICLONE mishandled
656 401416 Compile failure with openmpi 4.0
657 408663 Suppression file for musl libc
658 404076 s390x: z14 vector instructions not implemented
659 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
660 (actually affected all x86 and nanomips regardless of host bitness)
661 413547 regression test does not check for Arm 64 features.
662 414268 Enable AArch64 feature detection and decoding for v8.x instructions
663 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
664 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
665 422261 platform selection fails for unqualified client name
666 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
667 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
668 423195 PPC ISA 3.1 support is missing, part 1
669 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
670 424012 crash with readv/writev having invalid but not NULL arg2 iovec
671 424298 amd64: Implement RDSEED
672 425232 PPC ISA 3.1 support is missing, part 2
673 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
674 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
675 426123 PPC ISA 3.1 support is missing, part 3
676 426144 Fix "condition variable has not been initialized" on Fedora 33.
677 427400 PPC ISA 3.1 support is missing, part 4
678 427401 PPC ISA 3.1 support is missing, part 5
679 427404 PPC ISA 3.1 support is missing, part 6
680 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
681 427787 Support new faccessat2 linux syscall (439)
682 427969 debuginfo section duplicates a section in the main ELF file
683 428035 drd: Unbreak the musl build
684 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
685 428716 cppcheck detects potential leak in VEX/useful/smchash.c
686 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
687 429352 PPC ISA 3.1 support is missing, part 7
688 429354 PPC ISA 3.1 support is missing, part 8
689 429692 unhandled ppc64le-linux syscall: 147 (getsid)
690 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
692 429952 Errors when building regtest with clang
693 430354 ppc stxsibx and stxsihx instructions write too much data
694 430429 valgrind.h doesn't compile on s390x with clang
695 430485 expr_is_guardable doesn't handle Iex_Qop
696 431556 Complete arm64 FADDP v8.2 instruction support
697 432102 Add support for DWARF5 as produced by GCC11
698 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
699 432381 drd: Process STACK_REGISTER client requests
700 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
701 432672 vg_regtest: test-specific environment variables not reset between tests
702 432809 VEX should support REX.W + POPF
703 432861 PPC modsw and modsd give incorrect results for 1 mod 12
704 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
705 432215 Add debuginfod functionality
706 433323 Use pkglibexecdir as vglibdir
707 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
708 433629 valgrind/README has type "abd" instead of "and"
709 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
710 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
711 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
712 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
713 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
715 (3.17.0.RC1: 13 Mar 2021)
716 (3.17.0.RC2: 17 Mar 2021)
717 (3.17.0: 19 Mar 2021)
721 Release 3.16.1 (22 June 2020)
722 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
724 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
725 fixes character encoding problems in the documentation HTML.
727 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
728 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
730 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
734 Release 3.16.0 (27 May 2020)
735 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
737 3.16.0 is a feature release with many improvements and the usual collection of
740 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
741 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
742 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
743 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
744 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
746 * ==================== CORE CHANGES ===================
748 * It is now possible to dynamically change the value of many command line
749 options while your program (or its children) are running under Valgrind.
751 To see the list of dynamically changeable options, run
752 "valgrind --help-dyn-options".
754 You can change the options from the shell by using vgdb to launch
755 the monitor command "v.clo <clo option>...".
756 The same monitor command can be used from a gdb connected
757 to the valgrind gdbserver.
758 Your program can also change the dynamically changeable options using
759 the client request VALGRIND_CLO_CHANGE(option).
761 * ================== PLATFORM CHANGES =================
763 * MIPS: preliminary support for nanoMIPS instruction set has been added.
765 * ==================== TOOL CHANGES ====================
769 - The implicit memcpy done by each call to realloc now counts towards the
770 read and write counts of resized heap blocks, making those counts higher
775 - cg_annotate's --auto and --show-percs options now default to 'yes', because
776 they are usually wanted.
780 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
781 because they are usually wanted.
783 - The command option --collect-systime has been enhanced to specify
784 the unit used to record the elapsed time spent during system calls.
785 The command option now accepts the values no|yes|msec|usec|nsec,
786 where yes is a synonym of msec. When giving the value nsec, the
787 system cpu time of system calls is also recorded.
791 - Several memcheck options are now dynamically changeable.
792 Use valgrind --help-dyn-options to list them.
794 - The release 3.15 introduced a backward incompatible change for
795 some suppression entries related to preadv and pwritev syscalls.
796 When reading a suppression entry using the unsupported 3.14 format,
797 valgrind will now produce a warning to say the suppression entry will not
798 work, and suggest the needed change.
800 - Significantly fewer false positive errors on optimised code generated by
801 Clang and GCC. In particular, Memcheck now deals better with the
802 situation where the compiler will transform C-level "A && B" into "B && A"
803 under certain circumstances (in which the transformation is valid).
804 Handling of integer equality/non-equality checks on partially defined
805 values is also improved on some architectures.
809 - The exprimental Stack and Global Array Checking tool has been removed.
810 It only ever worked on x86 and amd64, and even on those it had a
811 high false positive rate and was slow. An alternative for detecting
812 stack and global array overruns is using the AddressSanitizer (ASAN)
813 facility of the GCC and Clang compilers, which require you to rebuild
814 your code with -fsanitize=address.
816 * ==================== OTHER CHANGES ====================
818 * New and modified GDB server monitor features:
820 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
822 - The gdbserver monitor commands that require an address and an optional
823 length argument now accepts the alternate 'C like' syntax "address[length]".
824 For example, the memcheck command "monitor who_points_at 0x12345678 120"
825 can now also be given as "monitor who_points_at 0x12345678[120]".
827 * ==================== FIXED BUGS ====================
829 The following bugs have been fixed or resolved. Note that "n-i-bz"
830 stands for "not in bugzilla" -- that is, a bug that was reported to us
831 but never got a bugzilla entry. We encourage you to file bugs in
832 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
833 than mailing the developers (or mailing lists) directly -- bugs that
834 are not entered into bugzilla tend to get forgotten about or ignored.
836 To see details of a given bug, visit
837 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
838 where XXXXXX is the bug number as listed below.
840 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
841 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
842 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
843 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
844 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
845 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
846 400872 Add nanoMIPS support to Valgrind
847 403212 drd/tests/trylock hangs on FreeBSD
848 404406 s390x: z14 miscellaneous instructions not implemented
849 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
850 406561 mcinfcallWSRU gdbserver_test fails on ppc64
851 406824 Unsupported baseline
852 407218 Add support for the copy_file_range syscall
853 407307 Intercept stpcpy also in ld.so for arm64
854 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
856 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
857 408009 Expose rdrand and f16c even on avx if host cpu supports them
858 408091 Missing pkey syscalls
859 408414 Add support for missing for preadv2 and pwritev2 syscalls
860 409141 Valgrind hangs when SIGKILLed
861 409206 Support for Linux PPS and PTP ioctls
862 409367 exit_group() after signal to thread waiting in futex() causes hangs
863 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
864 409780 References to non-existent configure.in
865 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
866 410599 Non-deterministic behaviour of pth_self_kill_15_other test
867 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
868 411134 Allow the user to change a set of command line options during execution
869 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
870 412344 Problem setting mips flags with specific paths
871 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
872 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
873 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
874 413603 callgrind_annotate/cg_annotate truncate function names at '#'
875 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
876 415136 ARMv8.1 Compare-and-Swap instructions are not supported
877 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
878 416239 valgrind crashes when handling clock_adjtime
879 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
880 416286 DRD reports "conflicting load" error on std::mutex::lock()
881 416301 s390x: "compare and signal" not supported
882 416387 finit_module and bpf syscalls are unhandled on arm64
883 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
884 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
885 416753 new 32bit time syscalls for 2038+
886 417075 pwritev(vector[...]) suppression ignored
887 417075 is not fixed, but incompatible supp entries are detected
888 and a warning is produced for these.
889 417187 [MIPS] Conditional branch problem since 'grail' changes
890 417238 Test memcheck/tests/vbit-test fails on mips64 BE
891 417266 Make memcheck/tests/linux/sigqueue usable with musl
892 417281 s390x: /bin/true segfaults with "grail" enabled
893 417427 commit to fix vki_siginfo_t definition created numerous regression
895 417452 s390_insn_store_emit: dst->tag for HRcVec128
896 417578 Add suppressions for glibc DTV leaks
897 417906 clone with CLONE_VFORK and no CLONE_VM fails
898 418004 Grail code additions break ppc64.
899 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
900 418997 s390x: Support Iex_ITE for float and vector types
901 419503 s390x: Avoid modifying registers returned from isel functions
902 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
903 421570 std_mutex fails on Arm v8.1 h/w
904 434035 vgdb might crash if valgrind is killed
905 n-i-bz Fix minor one time leaks in dhat.
906 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
907 n-i-bz Add support for the Linux io_uring system calls
908 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
909 n-i-bz Fix non-glibc build of test suite with s390x_features
910 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
911 423195 PPC ISA 3.1 support is missing, part 1
913 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
914 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
915 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
919 Release 3.15.0 (12 April 2019)
920 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
922 3.15.0 is a feature release with many improvements and the usual collection of
925 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
926 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
927 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
928 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
929 support for X86/macOS 10.13 and AMD64/macOS 10.13.
931 * ==================== CORE CHANGES ===================
933 * The XTree Massif output format now makes use of the information obtained
934 when specifying --read-inline-info=yes.
936 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
938 * ==================== TOOL CHANGES ====================
942 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
943 it has been promoted from an experimental tool to a regular tool. Run it
944 with --tool=dhat instead of --tool=exp-dhat.
946 - DHAT now prints only minimal data when the program ends, instead writing
947 the bulk of the profiling data to a file. As a result, the --show-top-n
948 and --sort-by options have been removed.
950 - Profile results can be viewed with the new viewer, dh_view.html. When
951 a run ends, a short message is printed, explaining how to view the result.
953 - See the documentation for more details.
957 - cg_annotate has a new option, --show-percs, which prints percentages next
962 - callgrind_annotate has a new option, --show-percs, which prints percentages
963 next to all event counts.
965 - callgrind_annotate now inserts commas in call counts, and
966 sort the caller/callee lists in the call tree.
970 - The default value for --read-inline-info is now "yes" on
971 Linux/Android/Solaris. It is still "no" on other OS.
975 - The option --xtree-leak=yes (to output leak result in xtree format)
976 automatically activates the option --show-leak-kinds=all, as xtree
977 visualisation tools such as kcachegrind can in any case select what kind
978 of leak to visualise.
980 - There has been further work to avoid false positives. In particular,
981 integer equality on partially defined inputs (C == and !=) is now handled
984 * ==================== OTHER CHANGES ====================
986 * The new option --show-error-list=no|yes displays, at the end of the run, the
987 list of detected errors and the used suppressions. Prior to this change,
988 showing this information could only be done by specifying "-v -v", but that
989 also produced a lot of other possibly-non-useful messages. The option -s is
990 equivalent to --show-error-list=yes.
992 * ==================== FIXED BUGS ====================
994 The following bugs have been fixed or resolved. Note that "n-i-bz"
995 stands for "not in bugzilla" -- that is, a bug that was reported to us
996 but never got a bugzilla entry. We encourage you to file bugs in
997 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
998 than mailing the developers (or mailing lists) directly -- bugs that
999 are not entered into bugzilla tend to get forgotten about or ignored.
1001 To see details of a given bug, visit
1002 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1003 where XXXXXX is the bug number as listed below.
1005 385411 s390x: z13 vector floating-point instructions not implemented
1006 397187 z13 vector register support for vgdb gdbserver
1007 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
1008 398870 Please add support for instruction vcvtps2ph
1009 399287 amd64 front end: Illegal Instruction vcmptrueps
1010 399301 Use inlined frames in Massif XTree output.
1011 399322 Improve callgrind_annotate output
1012 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
1013 400164 helgrind test encounters mips x-compiler warnings and assembler error
1014 400490 s390x: VRs allocated as if separate from FPRs
1015 400491 s390x: Operand of LOCH treated as unsigned integer
1016 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
1017 options, which specify a mips64 processor
1018 401112 LLVM 5.0 generates comparison against partially initialized data
1019 401277 More bugs in z13 support
1020 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
1021 401578 drd: crashes sometimes on fork()
1022 401627 memcheck errors with glibc avx2 optimized wcsncmp
1023 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
1024 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
1025 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
1027 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
1028 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
1029 402123 invalid assembler opcodes for mips32r2
1030 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
1031 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
1032 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
1033 402351 mips64 libvexmultiarch_test fails on s390x
1034 402369 Overhaul DHAT
1035 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
1036 402480 Do not use %rsp in clobber list
1037 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
1038 402515 Implement new option --show-error-list=no|yes / -s
1039 402519 POWER 3.0 addex instruction incorrectly implemented
1040 402781 Redo the cache used to process indirect branch targets
1041 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
1042 403552 s390x: wrong facility bit checked for vector facility
1043 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
1044 404638 Add VG_(replaceIndexXA)
1045 404843 s390x: backtrace sometimes ends prematurely
1046 404888 autotools cleanup series
1047 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
1048 405182 Valgrind fails to build with Clang
1049 405205 filter_libc: remove the line holding the futex syscall error entirely
1050 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
1051 the upper and lower 32-bits of the 64-bit result
1052 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
1053 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
1054 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
1056 405403 s390x disassembler cannot be used on x86
1057 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
1058 405458 MIPS mkFormVEC arguments swapped?
1059 405716 drd: Fix an integer overflow in the stack margin calculation
1060 405722 Support arm64 core dump
1061 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
1062 of the 64-bit destination field.
1063 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
1064 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
1065 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
1067 406256 PPC64, vector floating point instructions don't handle subnormal
1068 according to VSCR[NJ] bit setting.
1069 406352 cachegrind/callgrind fails ann tests because of missing a.c
1070 406354 dhat is broken on x86 (32bit)
1071 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
1072 406357 gdbserver_tests fails because of gdb output change
1073 406360 memcheck/tests/libstdc++.supp needs more supression variants
1074 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
1075 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
1076 407340 PPC64, does not support the vlogefp, vexptefp instructions.
1077 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
1078 n-i-bz Fix callgrind_annotate non deterministic order for equal total
1079 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
1080 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
1081 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
1083 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
1084 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
1085 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
1089 Release 3.14.0 (9 October 2018)
1090 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1092 3.14.0 is a feature release with many improvements and the usual collection of
1095 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1096 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1097 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1098 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
1099 support for X86/macOS 10.13, AMD64/macOS 10.13.
1101 * ==================== CORE CHANGES ===================
1103 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
1104 debug info for unloaded code. This allows saved stack traces (e.g. for
1105 memory leaks) to include file/line info for code that has been dlclose'd (or
1106 similar). See the user manual for more information and known limitations.
1108 * Ability to specify suppressions based on source file name and line number.
1110 * Majorly overhauled register allocator. No end-user changes, but the JIT
1111 generates code a bit more quickly now.
1113 * ================== PLATFORM CHANGES =================
1115 * Preliminary support for macOS 10.13 has been added.
1117 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
1119 * mips: support for MIPS SIMD architecture (MSA) has been added.
1121 * mips: support for MIPS N32 ABI has been added.
1123 * s390: partial support for vector instructions (integer and string) has been
1126 * ==================== TOOL CHANGES ====================
1128 * Helgrind: Addition of a flag
1129 --delta-stacktrace=no|yes [yes on linux amd64/x86]
1130 which specifies how full history stack traces should be computed.
1131 Setting this to =yes can speed up Helgrind by 25% when using
1132 --history-level=full.
1134 * Memcheck: reduced false positive rate for optimised code created by Clang 6
1135 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
1136 blocks more carefully to determine where it can avoid expensive definedness
1137 checks without loss of precision. This is controlled by the flag
1138 --expensive-definedness-checks=no|auto|yes [auto].
1140 * ==================== OTHER CHANGES ====================
1142 * Valgrind is now buildable with link-time optimisation (LTO). A new
1143 configure option --enable-lto=yes allows building Valgrind with LTO. If the
1144 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
1145 Note that if you are doing Valgrind development, --enable-lto=yes massively
1146 slows down the build process.
1148 * ==================== FIXED BUGS ====================
1150 The following bugs have been fixed or resolved. Note that "n-i-bz"
1151 stands for "not in bugzilla" -- that is, a bug that was reported to us
1152 but never got a bugzilla entry. We encourage you to file bugs in
1153 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1154 than mailing the developers (or mailing lists) directly -- bugs that
1155 are not entered into bugzilla tend to get forgotten about or ignored.
1157 To see details of a given bug, visit
1158 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1159 where XXXXXX is the bug number as listed below.
1161 79362 Debug info is lost for .so files when they are dlclose'd
1162 208052 strlcpy error when n = 0
1163 255603 exp-sgcheck Assertion '!already_present' failed
1164 338252 building valgrind with -flto (link time optimisation) fails
1165 345763 MIPS N32 ABI support
1166 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
1167 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
1168 372347 Replacement problem of the additional c++14/c++17 new/delete operators
1169 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1170 376257 helgrind history full speed up using a cached stack
1171 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
1173 379748 Fix missing pselect syscall (OS X 10.11)
1174 379754 Fix missing syscall ulock_wait (OS X 10.12)
1175 380397 s390x: __GI_strcspn() replacemenet needed
1176 381162 possible array overrun in VEX register allocator
1177 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
1178 381274 powerpc too chatty even with --sigill-diagnostics=no
1179 381289 epoll_pwait can have a NULL sigmask
1180 381553 VEX register allocator v3
1181 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
1182 381769 Use ucontext_t instead of struct ucontext
1183 381805 arm32 needs ld.so index hardwire for new glibc security fixes
1184 382256 gz compiler flag test doesn't work for gold
1185 382407 vg_perf needs "--terse" command line option
1186 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
1187 382563 MIPS MSA ASE support
1188 382998 xml-socket doesn't work
1189 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
1190 383723 Fix missing kevent_qos syscall (macOS 10.11)
1191 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
1192 384096 Mention AddrCheck at Memcheck's command line option [..]
1193 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
1194 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
1195 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
1196 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
1197 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
1198 384337 performance improvements to VEX register allocator v2 and v3
1199 384526 reduce number of spill insns generated by VEX register allocator v3
1200 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
1201 384631 Sanitise client args as printed with -v
1202 384633 Add a simple progress-reporting facility
1203 384987 VEX regalloc: allocate caller-save registers for short lived vregs
1204 385055 PPC VEX temporary storage exhausted
1205 385182 PPC64 is missing support for the DSCR
1206 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
1207 385207 PPC64, generate_store_FPRF() generates too many Iops
1208 385208 PPC64, xxperm instruction exhausts temporary memory
1209 385210 PPC64, vpermr instruction could exhaust temporary memory
1210 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
1211 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
1212 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
1213 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
1214 385408 s390x: z13 vector "support" instructions not implemented
1215 385409 s390x: z13 vector integer instructions not implemented
1216 385410 s390x: z13 vector string instructions not implemented
1217 385412 s390x: new non-vector z13 instructions not implemented
1218 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
1219 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
1220 385939 Optionally exit on the first error
1221 386318 valgrind.org/info/tools.html is missing SGCheck
1222 386425 running valgrind + wine on armv7l gives illegal opcode
1223 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
1224 387410 MIPSr6 support
1225 387664 Memcheck: make expensive-definedness-checks be the default
1226 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
1227 387766 asm shifts cause false positive "Conditional jump or move depends
1228 on uninitialised value"
1229 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
1230 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
1231 388786 Support bpf syscall in amd64 Linux
1232 388862 Add replacements for wmemchr and wcsnlen on Linux
1233 389065 valgrind meets gcc flag -Wlogical-op
1234 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
1235 390471 suppression by specification of source-file line number
1236 390723 make xtree dump files world wide readable, similar to log files
1237 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
1238 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
1239 392118 unhandled amd64-linux syscall: 332 (statx)
1240 392449 callgrind not clearing the number of calls properly
1241 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
1242 lxssp, stxssp and stxvl instructions.
1243 393023 callgrind_control risks using the wrong vgdb
1244 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
1245 393099 posix_memalign() invalid write if alignment == 0
1246 393146 failing assert "is_DebugInfo_active(di)"
1247 395709 PPC64 is missing support for the xvnegsp instruction
1248 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
1250 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
1251 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
1252 396839 s390x: Trap instructions not implemented
1253 396887 arch_prctl should return EINVAL on unknown option
1254 == 397286 crash before launching binary (Unsupported arch_prctl option)
1255 == 397393 valgrind: the 'impossible' happened: (Archlinux)
1256 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
1257 396906 compile tests failure on mips32-linux: broken inline asm in tests on
1259 397012 glibc ld.so uses arch_prctl on i386
1260 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
1261 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
1262 397424 glibc 2.27 and gdb_server tests
1263 398028 Assertion `cfsi_fits` failing in simple C program
1264 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
1266 n-i-bz Fix missing workq_ops operations (macOS)
1267 n-i-bz fix bug in strspn replacement
1268 n-i-bz Add support for the Linux BLKFLSBUF ioctl
1269 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
1270 n-i-bz Fix possible stack trashing by semctl syscall wrapping
1271 n-i-bz Add support for the Linux membarrier() system call
1272 n-i-bz x86 front end: recognise and handle UD2 correctly
1273 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
1274 correctly aligned before entering the handler.
1276 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
1277 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
1278 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
1282 Release 3.13.0 (15 June 2017)
1283 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1285 3.13.0 is a feature release with many improvements and the usual collection of
1288 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1289 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1290 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1291 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
1293 * ==================== CORE CHANGES ===================
1295 * The translation cache size has been increased to keep up with the demands of
1296 large applications. The maximum number of sectors has increased from 24 to
1297 48. The default number of sectors has increased from 16 to 32 on all
1298 targets except Android, where the increase is from 6 to 12.
1300 * The amount of memory that Valgrind can use has been increased from 64GB to
1301 128GB. In particular this means your application can allocate up to about
1302 60GB when running on Memcheck.
1304 * Valgrind's default load address has been changed from 0x3800'0000 to
1305 0x5800'0000, so as to make it possible to load larger executables. This
1306 should make it possible to load executables of size at least 1200MB.
1308 * A massive spaceleak caused by reading compressed debuginfo files has been
1309 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
1312 * The C++ demangler has been updated.
1314 * Support for demangling Rust symbols has been added.
1316 * A new representation of stack traces, the "XTree", has been added. An XTree
1317 is a tree of stacktraces with data associated with the stacktraces. This is
1318 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
1319 consumption of your program. Reporting is controlled by the new options
1320 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
1322 A report can also be produced on demand using the gdbserver monitor command
1323 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
1324 format' and 'massif format. The existing visualisers for these formats (e.g.
1325 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
1326 analyse these reports.
1328 Memcheck can also produce XTree leak reports using the Callgrind file
1329 format. For more details, see the user manual.
1331 * ================== PLATFORM CHANGES =================
1333 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
1335 * amd64: fixes for JIT failure problems on long AVX2 code blocks
1337 * amd64 and x86: support for CET prefixes has been added
1339 * arm32: a few missing ARMv8 instructions have been implemented
1341 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
1342 Store-Conditional instructions has been added. This is to deal with
1343 processor implementations that implement the LL/SC specifications strictly
1344 and as a result cause Valgrind to hang in certain situations. The
1345 alternative implementation is automatically enabled at startup, as required.
1346 You can use the option --sim-hints=fallback-llsc to force-enable it if you
1349 * Support for OSX 10.12 has been improved.
1351 * On Linux, clone handling has been improved to honour CLONE_VFORK that
1352 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
1353 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
1354 CLONE_VM exact semantics will (still) not work.
1356 * The TileGX/Linux port has been removed because it appears to be both unused
1359 * ==================== TOOL CHANGES ====================
1363 - Memcheck should give fewer false positives when running optimised
1364 Clang/LLVM generated code.
1366 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1368 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
1369 to produce the end of execution leak report in a xtree callgrind format
1372 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
1373 the leak report in an xtree file.
1377 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1379 - For some workloads (typically, for big applications), Massif memory
1380 consumption and CPU consumption has decreased significantly.
1384 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1386 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
1387 for Ada gnat compiled applications.
1389 * ==================== OTHER CHANGES ====================
1391 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
1392 append the inner guest stacktrace to the inner host stacktrace. This helps
1393 to investigate the errors reported by the outer, when they are caused by the
1394 inner guest program (such as an inner regtest). See README_DEVELOPERS for
1397 * To allow fast detection of callgrind files by desktop environments and file
1398 managers, the format was extended to have an optional first line that
1399 uniquely identifies the format ("# callgrind format"). Callgrind creates
1400 this line now, as does the new xtree functionality.
1402 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
1403 have a new %n format letter that is replaced by a sequence number.
1405 * "--version -v" now shows the SVN revision numbers from which Valgrind was
1408 * ==================== FIXED BUGS ====================
1410 The following bugs have been fixed or resolved. Note that "n-i-bz"
1411 stands for "not in bugzilla" -- that is, a bug that was reported to us
1412 but never got a bugzilla entry. We encourage you to file bugs in
1413 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1414 than mailing the developers (or mailing lists) directly -- bugs that
1415 are not entered into bugzilla tend to get forgotten about or ignored.
1417 To see details of a given bug, visit
1418 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1419 where XXXXXX is the bug number as listed below.
1421 162848 --log-file output isn't split when a program forks
1422 340777 Illegal instruction on mips (ar71xx)
1423 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
1424 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
1425 to a different stack.
1426 344139 x86 stack-seg overrides, needed by the Wine people
1427 344524 store conditional of guest applications always fail - observed on
1429 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
1430 352395 Please provide SVN revision info in --version -v
1431 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
1432 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
1433 INVALID_POSIX_THREADID' failed
1434 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
1435 barrier implementation
1436 358697 valgrind.h: Some code remains even when defining NVALGRIND
1437 359202 Add musl libc configure/compile
1438 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
1439 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
1440 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
1441 362223 assertion failed when .valgrindrc is a directory instead of a file
1442 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
1443 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
1444 368507 can't malloc chunks larger than about 34GB
1445 368529 Android arm target link error, missing atexit and pthread_atfork
1446 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
1447 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
1448 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
1449 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
1450 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
1451 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
1452 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
1453 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
1454 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
1455 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
1456 370028 Reduce the number of compiler warnings on MIPS platforms
1457 370635 arm64 missing syscall getcpu
1458 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
1459 371227 Clean AArch64 syscall table
1460 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
1461 371471 Valgrind complains about non legit memory leaks on placement new (C++)
1462 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
1463 371503 disInstr(arm64): unhandled instruction 0xF89F0000
1464 371869 support '%' in symbol Z-encoding
1465 371916 execution tree xtree concept
1466 372120 c++ demangler demangles symbols which are not c++
1467 372185 Support of valgrind on ARMv8 with 32 bit executable
1468 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1469 372195 Power PC, xxsel instruction is not always recognized.
1470 372504 Hanging on exit_group
1471 372600 process loops forever when fatal signals are arriving quickly
1472 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1473 373046 Stacks registered by core are never deregistered
1474 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1475 373086 Implement additional Xen hypercalls
1476 373192 Calling posix_spawn in glibc 2.24 completely broken
1477 373488 Support for fanotify API on ARM64 architecture
1478 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1479 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1480 373938 const IRExpr arguments for matchIRExpr()
1481 374719 some spelling fixes
1482 374963 increase valgrind's load address to prevent mmap failure
1483 375514 valgrind_get_tls_addr() does not work in case of static TLS
1484 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1485 for ML_(find_rx_mapping)()
1486 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1487 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1488 == 377159 "vex: the `impossible' happened" still present
1489 == 375150 Assertion 'tres.status == VexTransOK' failed
1490 == 378068 valgrind crashes on AVX2 function in FFmpeg
1491 376142 Segfaults on MIPS Cavium Octeon boards
1492 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1493 376455 Solaris: unhandled syscall lgrpsys(180)
1494 376518 Solaris: unhandled fast trap getlgrp(6)
1495 376611 ppc64 and arm64 don't know about prlimit64 syscall
1496 376729 PPC64, remove R2 from the clobber list
1498 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1499 to be wrongly marked as addressable
1500 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1501 PIE enabled by default
1502 377376 memcheck/tests/linux/getregset fails with glibc2.24
1503 377427 PPC64, lxv instruction failing on odd destination register
1504 377478 PPC64: ISA 3.0 setup fixes
1505 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1506 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1507 and 2 args for FUTEX_TRYLOCK_PI
1508 377717 Fix massive space leak when reading compressed debuginfo sections
1509 377891 Update Xen 4.6 domctl wrappers
1510 377930 fcntl syscall wrapper is missing flock structure check
1511 378524 libvexmultiarch_test regression on s390x and ppc64
1512 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1513 378673 Update libiberty demangler
1514 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1515 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1516 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1517 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1518 (task_register_dyld_image_infos)
1519 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1520 (task_register_dyld_shared_cache_image_info)
1521 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1522 379473 MIPS: add support for rdhwr cycle counter register
1523 379504 remove TileGX/Linux port
1524 379525 Support more x86 nop opcodes
1525 379838 disAMode(x86): not an addr!
1526 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1527 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1528 379895 clock_gettime does not execute POST syscall wrapper
1529 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1530 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1531 380200 xtree generated callgrind files refer to files without directory name
1532 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1533 380397 s390x: __GI_strcspn() replacement needed
1534 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1536 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1537 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1538 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1542 Release 3.12.0 (20 October 2016)
1543 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1545 3.12.0 is a feature release with many improvements and the usual
1546 collection of bug fixes.
1548 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1549 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1550 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1551 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1552 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1553 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1555 * ================== PLATFORM CHANGES =================
1557 * POWER: Support for ISA 3.0 has been added
1559 * mips: support for O32 FPXX ABI has been added.
1560 * mips: improved recognition of different processors
1561 * mips: determination of page size now done at run time
1563 * amd64: Partial support for AMD FMA4 instructions.
1565 * arm, arm64: Support for v8 crypto and CRC instructions.
1567 * Improvements and robustification of the Solaris port.
1569 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1571 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1572 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1573 where possible. Valgrind's support for 32-bit x86 has stagnated in
1574 recent years and has fallen far behind that for 64-bit x86
1575 instructions. By contrast 64-bit x86 is well supported, up to and
1578 * ==================== TOOL CHANGES ====================
1582 - Added meta mempool support for describing a custom allocator which:
1583 - Auto-frees all chunks assuming that destroying a pool destroys all
1585 - Uses itself to allocate other memory blocks
1587 - New flag --ignore-range-below-sp to ignore memory accesses below
1588 the stack pointer, if you really have to. The related flag
1589 --workaround-gcc296-bugs=yes is now deprecated. Use
1590 --ignore-range-below-sp=1024-1 as a replacement.
1594 - Improved thread startup time significantly on non-Linux platforms.
1598 - Added collection of the metric "tot-blocks-allocd"
1600 * ==================== OTHER CHANGES ====================
1602 * Replacement/wrapping of malloc/new related functions is now done not just
1603 for system libraries by default, but for any globally defined malloc/new
1604 related function (both in shared libraries and statically linked alternative
1605 malloc implementations). The dynamic (runtime) linker is excluded, though.
1606 To only intercept malloc/new related functions in
1607 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1608 "nouserintercepts" can be any non-existing library name).
1609 This new functionality is not implemented for MacOS X.
1611 * The maximum number of callers in a suppression entry is now equal to
1612 the maximum size for --num-callers (500).
1613 Note that --gen-suppressions=yes|all similarly generates suppressions
1614 containing up to --num-callers frames.
1616 * New and modified GDB server monitor features:
1618 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1619 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1622 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1623 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1626 * Valgrind is able to read compressed debuginfo sections in two formats:
1627 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1628 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1630 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1631 for the most common use case (x86_64-linux, Memcheck) has been
1634 * Improved performance for programs that do a lot of discarding of
1635 instruction address ranges of 8KB or less.
1637 * The C++ symbol demangler has been updated.
1639 * More robustness against invalid syscall parameters on Linux.
1641 * ==================== FIXED BUGS ====================
1643 The following bugs have been fixed or resolved. Note that "n-i-bz"
1644 stands for "not in bugzilla" -- that is, a bug that was reported to us
1645 but never got a bugzilla entry. We encourage you to file bugs in
1646 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1647 than mailing the developers (or mailing lists) directly -- bugs that
1648 are not entered into bugzilla tend to get forgotten about or ignored.
1650 To see details of a given bug, visit
1651 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1652 where XXXXXX is the bug number as listed below.
1654 191069 Exiting due to signal not reported in XML output
1655 199468 Suppressions: stack size limited to 25
1656 while --num-callers allows more frames
1657 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1658 278744 cvtps2pd with redundant RexW
1659 303877 valgrind doesn't support compressed debuginfo sections.
1660 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1661 348345 Assertion fails for negative lineno
1662 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1663 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1664 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1665 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1666 352197 mips: mmap2() not wrapped correctly for page size > 4096
1667 353083 arm64 doesn't implement various xattr system calls
1668 353084 arm64 doesn't support sigpending system call
1669 353137 www: update info for Supported Platforms
1670 353138 www: update "The Valgrind Developers" page
1671 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1674 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1675 353398 WARNING: unhandled amd64-solaris syscall: 207
1676 353660 XML in auxwhat tag not escaping reserved symbols properly
1677 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1678 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1679 353802 ELF debug info reader confused with multiple .rodata sections
1680 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1681 353917 unhandled amd64-solaris syscall fchdir(120)
1682 353920 unhandled amd64-solaris syscall: 170
1683 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1684 354392 unhandled amd64-solaris syscall: 171
1685 354797 Vbit test does not include Iops for Power 8 instruction support
1686 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1690 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1691 355188 valgrind should intercept all malloc related global functions
1692 355454 do not intercept malloc related symbols from the runtime linker
1693 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1694 356044 Dwarf line info reader misinterprets is_stmt register
1695 356112 mips: replace addi with addiu
1696 356393 valgrind (vex) crashes because isZeroU happened
1699 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1700 356678 arm64-linux: unhandled syscall 232 (mincore)
1701 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1702 356823 Unsupported ARM instruction: stlex
1703 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1704 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1705 357673 crash if I try to run valgrind with a binary link with libcurl
1706 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1707 357871 pthread_spin_destroy not properly wrapped
1708 357887 Calls to VG_(fclose) do not close the file descriptor
1709 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1710 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1711 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1712 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1713 359181 Buffer Overflow during Demangling
1714 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1715 359289 s390x: popcnt (B9E1) not implemented
1716 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1717 359503 Add missing syscalls for aarch64 (arm64)
1718 359645 "You need libc6-dbg" help message could be more helpful
1719 359703 s390: wire up separate socketcalls system calls
1720 359724 getsockname might crash - deref_UInt should call safe_to_deref
1721 359733 amd64 implement ld.so strchr/index override like x86
1722 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1723 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1725 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1726 359871 Incorrect mask handling in ppoll
1727 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1728 360008 Contents of Power vr registers contents is not printed correctly when
1729 the --vgdb-shadow-registers=yes option is used
1730 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1731 non-zero shadow bits
1732 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1733 360425 arm64 unsupported instruction ldpsw
1735 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1736 360571 Error about the Android Runtime reading below the stack pointer on ARM
1737 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1738 360749 kludge for multiple .rodata sections on Solaris no longer needed
1739 360752 raise the number of reserved fds in m_main.c from 10 to 12
1740 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1741 361226 s390x: risbgn (EC59) not implemented
1742 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1743 361354 ppc64[le]: wire up separate socketcalls system calls
1744 361615 Inconsistent termination for multithreaded process terminated by signal
1745 361926 Unhandled Solaris syscall: sysfs(84)
1746 362009 V dumps core on unimplemented functionality before threads are created
1747 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1748 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1749 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1750 362953 Request for an update to the Valgrind Developers page
1751 363680 add renameat2() support
1752 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1753 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1754 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1755 364058 clarify in manual limitations of array overruns detections
1756 364413 pselect sycallwrapper mishandles NULL sigmask
1757 364728 Power PC, missing support for several HW registers in
1758 get_otrack_shadow_offset_wrk()
1759 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1760 365273 Invalid write to stack location reported after signal handler runs
1761 365912 ppc64BE segfault during jm-insns test (RELRO)
1762 366079 FPXX Support for MIPS32 Valgrind
1763 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1764 366344 Multiple unhandled instruction for Aarch64
1765 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1766 367995 Integration of memcheck with custom memory allocator
1767 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1768 368412 False positive result for altivec capability check
1769 368416 Add tc06_two_races_xml.exp output for ppc64
1770 368419 Perf Events ioctls not implemented
1771 368461 mmapunmap test fails on ppc64
1772 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1773 369000 AMD64 fma4 instructions unsupported.
1774 369169 ppc64 fails jm_int_isa_2_07 test
1775 369175 jm_vec_isa_2_07 test crashes on ppc64
1776 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1777 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1778 369359 msghdr_foreachfield can crash when handling bad iovec
1779 369360 Bad sigprocmask old or new sets can crash valgrind
1780 369361 vmsplice syscall wrapper crashes on bad iovec
1781 369362 Bad sigaction arguments crash valgrind
1782 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1783 369402 Bad set/get_thread_area pointer crashes valgrind
1784 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1785 369446 valgrind crashes on unknown fcntl command
1786 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1787 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1788 370265 ISA 3.0 HW cap stuff needs updating
1789 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1790 372195 Power PC, xxsel instruction is not always recognized
1792 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1793 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1794 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1795 n-i-bz Fix ppoll_alarm exclusion on OS X
1796 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1797 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1798 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1799 n-i-bz amd64: memcheck false positive with shr %edx
1800 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1801 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1802 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1803 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1805 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1806 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1807 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1811 Release 3.11.0 (22 September 2015)
1812 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1814 3.11.0 is a feature release with many improvements and the usual
1815 collection of bug fixes.
1817 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1818 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1819 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1820 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1821 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1822 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1824 * ================== PLATFORM CHANGES =================
1826 * Support for Solaris/x86 and Solaris/amd64 has been added.
1828 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1830 * Preliminary support for the Tilera TileGX architecture has been added.
1832 * s390x: It is now required for the host to have the "long displacement"
1833 facility. The oldest supported machine model is z990.
1835 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1836 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1838 * The JIT's register allocator is significantly faster, making the JIT
1839 as a whole somewhat faster, so JIT-intensive activities, for example
1840 program startup, are modestly faster, around 5%.
1842 * There have been changes to the default settings of several command
1843 line flags, as detailed below.
1845 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1846 capable hosts, the simulated CPUID will now indicate AVX2 support.
1848 * ==================== TOOL CHANGES ====================
1852 - The default value for --leak-check-heuristics has been changed from
1853 "none" to "all". This helps to reduce the number of possibly
1854 lost blocks, in particular for C++ applications.
1856 - The default value for --keep-stacktraces has been changed from
1857 "malloc-then-free" to "malloc-and-free". This has a small cost in
1858 memory (one word per malloc-ed block) but allows Memcheck to show the
1859 3 stacktraces of a dangling reference: where the block was allocated,
1860 where it was freed, and where it is acccessed after being freed.
1862 - The default value for --partial-loads-ok has been changed from "no" to
1863 "yes", so as to avoid false positive errors resulting from some kinds
1864 of vectorised loops.
1866 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1867 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1868 than get_vbits when you need to associate byte data value with
1869 their corresponding validity bits.
1871 - The 'block_list' monitor command has been enhanced:
1872 o it can print a range of loss records
1873 o it now accepts an optional argument 'limited <max_blocks>'
1874 to control the number of blocks printed.
1875 o if a block has been found using a heuristic, then
1876 'block_list' now shows the heuristic after the block size.
1877 o the loss records/blocks to print can be limited to the blocks
1878 found via specified heuristics.
1880 - The C helper functions used to instrument loads on
1881 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1882 replaced by handwritten assembly sequences. This gives speedups
1883 in the region of 0% to 7% for those targets only.
1885 - A new command line option, --expensive-definedness-checks=yes|no,
1886 has been added. This is useful for avoiding occasional invalid
1887 uninitialised-value errors in optimised code. Watch out for
1888 runtime degradation, as this can be up to 25%. As always, though,
1889 the slowdown is highly application specific. The default setting
1894 - A new monitor command 'all_snapshots <filename>' dumps all
1895 snapshots taken so far.
1899 - Significant memory reduction and moderate speedups for
1900 --history-level=full for applications accessing a lot of memory
1901 with many different stacktraces.
1903 - The default value for --conflict-cache-size=N has been doubled to
1904 2000000. Users that were not using the default value should
1905 preferably also double the value they give.
1907 The default was changed due to the changes in the "full history"
1908 implementation. Doubling the value gives on average a slightly more
1909 complete history and uses similar memory (or significantly less memory
1910 in the worst case) than the previous implementation.
1912 - The Helgrind monitor command 'info locks' now accepts an optional
1913 argument 'lock_addr', which shows information about the lock at the
1916 - When using --history-level=full, the new Helgrind monitor command
1917 'accesshistory <addr> [<len>]' will show the recorded accesses for
1918 <len> (or 1) bytes at <addr>.
1920 * ==================== OTHER CHANGES ====================
1922 * The default value for the --smc-check option has been changed from
1923 "stack" to "all-non-file" on targets that provide automatic D-I
1924 cache coherence (x86, amd64 and s390x). The result is to provide,
1925 by default, transparent support for JIT generated and self-modifying
1926 code on all targets.
1928 * Mac OS X only: the default value for the --dsymutil option has been
1929 changed from "no" to "yes", since any serious usage on Mac OS X
1930 always required it to be "yes".
1932 * The command line options --db-attach and --db-command have been removed.
1933 They were deprecated in 3.10.0.
1935 * When a process dies due to a signal, Valgrind now shows the signal
1936 and the stacktrace at default verbosity (i.e. verbosity 1).
1938 * The address description logic used by Memcheck and Helgrind now
1939 describes addresses in anonymous segments, file mmap-ed segments,
1940 shared memory segments and the brk data segment.
1942 * The new option --error-markers=<begin>,<end> can be used to mark the
1943 begin/end of errors in textual output mode, to facilitate
1944 searching/extracting errors in output files that mix valgrind errors
1945 with program output.
1947 * The new option --max-threads=<number> can be used to change the number
1948 of threads valgrind can handle. The default is 500 threads which
1949 should be more than enough for most applications.
1951 * The new option --valgrind-stacksize=<number> can be used to change the
1952 size of the private thread stacks used by Valgrind. This is useful
1953 for reducing memory use or increasing the stack size if Valgrind
1954 segfaults due to stack overflow.
1956 * The new option --avg-transtab-entry-size=<number> can be used to specify
1957 the expected instrumented block size, either to reduce memory use or
1958 to avoid excessive retranslation.
1960 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1962 * New and modified GDB server monitor features:
1964 - When a signal is reported in GDB, you can now use the GDB convenience
1965 variable $_siginfo to examine detailed signal information.
1967 - Valgrind's gdbserver now allows the user to change the signal
1968 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1969 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1970 continue without passing the signal to the process.
1972 - With GDB >= 7.10, the command 'target remote'
1973 will automatically load the executable file of the process running
1974 under Valgrind. This means you do not need to specify the executable
1975 file yourself, GDB will discover it itself. See GDB documentation about
1976 'qXfer:exec-file:read' packet for more info.
1978 * ==================== FIXED BUGS ====================
1980 The following bugs have been fixed or resolved. Note that "n-i-bz"
1981 stands for "not in bugzilla" -- that is, a bug that was reported to us
1982 but never got a bugzilla entry. We encourage you to file bugs in
1983 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1984 than mailing the developers (or mailing lists) directly -- bugs that
1985 are not entered into bugzilla tend to get forgotten about or ignored.
1987 To see details of a given bug, visit
1988 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1989 where XXXXXX is the bug number as listed below.
1991 116002 VG_(printf): Problems with justification of strings and integers
1992 155125 avoid cutting away file:lineno after long function name
1993 197259 Unsupported arch_prtctl PR_SET_GS option
1994 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
1995 201216 Fix Valgrind does not support pthread_sigmask() on OS X
1996 201435 Fix Darwin: -v does not show kernel version
1997 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
1998 211256 Fixed an outdated comment regarding the default platform.
1999 211529 Incomplete call stacks for code compiled by newer versions of MSVC
2000 211926 Avoid compilation warnings in valgrind.h with -pedantic
2001 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
2003 226609 Crediting upstream authors in man page
2004 231257 Valgrind omits path when executing script from shebang line
2005 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
2006 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
2007 269360 s390x: Fix addressing mode selection for compare-and-swap
2008 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
2010 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
2011 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
2012 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
2013 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
2014 330147 libmpiwrap PMPI_Get_count returns undefined value
2015 333051 mmap of huge pages fails due to incorrect alignment
2017 334802 valgrind does not always explain why a given option is bad
2018 335618 mov.w rN, pc/sp (ARM32)
2019 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
2024 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
2025 338602 AVX2 bit in CPUID missing
2026 338606 Strange message for scripts with invalid interpreter
2027 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
2028 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
2029 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
2031 339156 gdbsrv not called for fatal signal
2032 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
2033 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
2034 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
2035 339442 Fix testsuite build failure on OS X 10.9
2036 339542 Enable compilation with Intel's ICC compiler
2037 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
2038 339688 Mac-specific ASM does not support .version directive (cpuid,
2039 tronical and pushfpopf tests)
2040 339745 Valgrind crash when check Marmalade app (partial fix)
2041 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
2042 339778 Linux/TileGx platform support for Valgrind
2043 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
2044 339789 Fix none/tests/execve test on Mac OS X 10.9
2045 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
2046 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
2047 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
2048 340392 Allow user to select more accurate definedness checking in memcheck
2049 to avoid invalid complaints on optimised code
2050 340430 Fix some grammatical weirdness in the manual.
2051 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
2052 341419 Signal handler ucontext_t not filled out correctly on OS X
2053 341539 VG_(describe_addr) should not describe address as belonging to client
2054 segment if it is past the heap end
2055 341613 Enable building of manythreads and thread-exits tests on Mac OS X
2056 341615 Fix none/tests/darwin/access_extended test on Mac OS X
2057 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
2058 341789 aarch64: shmat fails with valgrind on ARMv8
2059 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
2060 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
2061 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
2062 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
2063 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
2064 342221 socket connect false positive uninit memory for unknown af family
2065 342353 Allow dumping full massif output while valgrind is still running
2066 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
2070 342603 Add I2C_SMBUS ioctl support
2071 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
2072 342683 Mark memory past the initial brk limit as unaddressable
2073 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
2074 342795 Internal glibc __GI_mempcpy call should be intercepted
2075 342841 s390x: Support instructions fiebr(a) and fidbr(a)
2076 343012 Unhandled syscall 319 (memfd_create)
2077 343069 Patch updating v4l2 API support
2078 343173 helgrind crash during stack unwind
2079 343219 fix GET_STARTREGS for arm
2080 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
2081 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
2082 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
2083 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
2084 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
2085 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
2086 343597 ppc64le: incorrect use of offseof macro
2087 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
2088 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
2089 343732 Unhandled syscall 144 (setgid) on aarch64
2090 343733 Unhandled syscall 187 (msgctl and related) on aarch64
2091 343802 s390x: False positive "conditional jump or move depends on [..]
2092 343902 --vgdb=yes doesn't break when --xml=yes is used
2093 343967 Don't warn about setuid/setgid/setcap executable for directories
2094 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
2095 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
2096 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
2097 344054 www - update info for Solaris/illumos
2098 344416 'make regtest' does not work cleanly on OS X
2099 344235 Remove duplicate include of pub_core_aspacemgr.h
2100 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
2101 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
2102 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
2103 344314 callgrind_annotate ... warnings about commands containing newlines
2104 344318 socketcall should wrap recvmmsg and sendmmsg
2105 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
2106 344416 Fix 'make regtest' does not work cleanly on OS X
2107 344499 Fix compilation for Linux kernel >= 4.0.0
2108 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
2109 unix:349 (__pthread_fchdir)
2110 344559 Garbage collection of unused segment names in address space manager
2111 344560 Fix stack traces missing penultimate frame on OS X
2112 344621 Fix memcheck/tests/err_disable4 test on OS X
2113 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
2114 344702 Fix missing libobjc suppressions on OS X 10.10
2116 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
2117 344939 Fix memcheck/tests/xml1 on OS X 10.10
2118 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
2119 345079 Fix build problems in VEX/useful/test_main.c
2120 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
2121 345177 arm64: prfm (reg) not implemented
2122 345215 Performance improvements for the register allocator
2123 345248 add support for Solaris OS in valgrind
2124 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
2125 345394 Fix memcheck/tests/strchr on OS X
2126 345637 Fix memcheck/tests/sendmsg on OS X
2127 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
2128 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
2129 345887 Fix an assertion in the address space manager
2130 345928 amd64: callstack only contains current function for small stacks
2131 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
2132 345987 MIPS64: Implement cavium LHX instruction
2133 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
2134 346185 Fix typo saving altivec register v24
2135 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
2136 and LibVEX_GuestPPC64_get_CR()
2137 346270 Regression tests none/tests/jm_vec/isa_2_07 and
2138 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
2139 346307 fuse filesystem syscall deadlocks
2140 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
2141 346411 MIPS: SysRes::_valEx handling is incorrect
2142 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
2143 346474 PPC64 Power 8, spr TEXASRU register not supported
2144 346487 Compiler generates "note" about a future ABI change for PPC64
2145 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
2146 and causing spurious "invalid read of size 8" warnings
2147 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
2148 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
2149 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
2150 347322 Power PC regression test cleanup
2151 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
2153 347389 unhandled syscall: 373 (Linux ARM syncfs)
2154 347686 Patch set to cleanup PPC64 regtests
2155 347978 Remove bash dependencies where not needed
2156 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
2157 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
2159 348102 Patch updating v4l2 API support
2160 348247 amd64 front end: jno jumps wrongly when overflow is not set
2161 348269 Improve mmap MAP_HUGETLB support.
2162 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
2163 348345 Assertion fails for negative lineno
2164 348377 Unsupported ARM instruction: yield
2165 348565 Fix detection of command line option availability for clang
2166 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
2167 348728 Fix broken check for VIDIOC_G_ENC_INDEX
2168 348748 Fix redundant condition
2169 348890 Fix clang warning about unsupported --param inline-unit-growth=900
2170 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
2171 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
2172 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
2173 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
2174 349626 Implemented additional Xen hypercalls
2175 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
2176 349790 Clean up of the hardware capability checking utilities.
2177 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
2178 349874 Fix typos in source code
2179 349879 memcheck: add handwritten assembly for helperc_LOADV*
2180 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
2181 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
2182 350202 Add limited param to 'monitor block_list'
2183 350290 s390x: Support instructions fixbr(a)
2184 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
2185 350809 Fix none/tests/async-sigs for Solaris
2186 350811 Remove reference to --db-attach which has been removed.
2187 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
2188 350854 hard-to-understand code in VG_(load_ELF)()
2189 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2190 351386 Solaris: Cannot run ld.so.1 under Valgrind
2191 351474 Fix VG_(iseqsigset) as obvious
2192 351531 Typo in /include/vki/vki-xen-physdev.h header guard
2193 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
2194 351858 ldsoexec support on Solaris
2195 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
2196 352130 helgrind reports false races for printfs using mempcpy on FILE* state
2197 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
2198 352320 arm64 crash on none/tests/nestedfs
2199 352765 Vbit test fails on Power 6
2200 352768 The mbar instruction is missing from the Power PC support
2201 352769 Power PC program priority register (PPR) is not supported
2202 n-i-bz Provide implementations of certain compiler builtins to support
2203 compilers that may not provide those
2204 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
2205 n-i-bz Fix compilation on distros with glibc < 2.5
2206 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
2207 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
2208 n-i-bz mremap did not work properly on shared memory
2209 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
2210 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
2212 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
2213 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
2214 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
2218 Release 3.10.1 (25 November 2014)
2219 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2220 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
2221 and backports fixes for all reported missing AArch64 ARMv8 instructions
2222 and syscalls from the trunk. If you package or deliver 3.10.0 for others
2223 to use, you might want to consider upgrading to 3.10.1 instead.
2225 The following bugs have been fixed or resolved. Note that "n-i-bz"
2226 stands for "not in bugzilla" -- that is, a bug that was reported to us
2227 but never got a bugzilla entry. We encourage you to file bugs in
2228 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2229 than mailing the developers (or mailing lists) directly -- bugs that
2230 are not entered into bugzilla tend to get forgotten about or ignored.
2232 To see details of a given bug, visit
2233 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2234 where XXXXXX is the bug number as listed below.
2236 335440 arm64: ld1 (single structure) is not implemented
2237 335713 arm64: unhanded instruction: prfm (immediate)
2238 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
2239 339182 ppc64: AvSplat ought to load destination vector register with [..]
2240 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
2241 339433 ppc64 lxvw4x instruction uses four 32-byte loads
2242 339645 Use correct tag names in sys_getdents/64 wrappers
2243 339706 Fix false positive for ioctl(TIOCSIG) on linux
2244 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
2245 339853 arm64 times syscall unknown
2246 339855 arm64 unhandled getsid/setsid syscalls
2247 339858 arm64 dmb sy not implemented
2248 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
2249 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
2250 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
2252 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
2253 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
2254 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
2255 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
2256 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
2257 340509 arm64: unhandled instruction fcvtas
2258 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
2259 340632 arm64: unhandled instruction fcvtas
2260 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
2261 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
2262 340788 warning: unhandled syscall: 318 (getrandom)
2263 340807 disInstr(arm): unhandled instruction: 0xEE989B20
2264 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
2265 340922 arm64: unhandled getgroups/setgroups syscalls
2266 350251 Fix typo in VEX utility program (test_main.c).
2267 350407 arm64: unhandled instruction ucvtf (vector, integer)
2268 350809 none/tests/async-sigs breaks when run under cron on Solaris
2269 350811 update README.solaris after r15445
2270 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
2271 350854 strange code in VG_(load_ELF)()
2272 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2273 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
2274 n-i-bz Add missing ]] to terminate CDATA.
2275 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
2276 n-i-bz Enable sys_fadvise64_64 on arm32.
2277 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
2278 n-i-bz Add test cases for all known arm64 load/store instructions.
2279 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
2280 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
2281 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
2282 n-i-bz arm64: Implement "fcvtpu w, s".
2283 n-i-bz arm64: implement ADDP and various others
2284 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
2285 n-i-bz arm64: enable FCVT{A,N}S X,S.
2287 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
2291 Release 3.10.0 (10 September 2014)
2292 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2294 3.10.0 is a feature release with many improvements and the usual
2295 collection of bug fixes.
2297 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
2298 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
2299 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
2300 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
2301 significantly improved relative to the 3.9.0 release.
2303 * ================== PLATFORM CHANGES =================
2305 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
2306 is mostly complete, and is usable, but some SIMD instructions are as
2309 * Support for little-endian variant of the 64-bit POWER architecture.
2311 * Support for Android on MIPS32.
2313 * Support for 64bit FPU on MIPS32 platforms.
2315 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
2317 * Configuration for and running on Android targets has changed.
2318 See README.android in the source tree for details.
2320 * ================== DEPRECATED FEATURES =================
2322 * --db-attach is now deprecated and will be removed in the next
2323 valgrind feature release. The built-in GDB server capabilities are
2324 superior and should be used instead. Learn more here:
2325 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
2327 * ==================== TOOL CHANGES ====================
2331 - Client code can now selectively disable and re-enable reporting of
2332 invalid address errors in specific ranges using the new client
2333 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
2334 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
2336 - Leak checker: there is a new leak check heuristic called
2337 "length64". This is used to detect interior pointers pointing 8
2338 bytes inside a block, on the assumption that the first 8 bytes
2339 holds the value "block size - 8". This is used by
2340 sqlite3MemMalloc, for example.
2342 - Checking of system call parameters: if a syscall parameter
2343 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
2344 several fields not initialised, an error is now reported for each
2345 field. Previously, an error was reported only for the first
2346 uninitialised field.
2348 - Mismatched alloc/free checking: a new flag
2349 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
2350 such checks if necessary.
2354 - Improvements to error messages:
2356 o Race condition error message involving heap allocated blocks also
2357 show the thread number that allocated the raced-on block.
2359 o All locks referenced by an error message are now announced.
2360 Previously, some error messages only showed the lock addresses.
2362 o The message indicating where a lock was first observed now also
2363 describes the address/location of the lock.
2365 - Helgrind now understands the Ada task termination rules and
2366 creates a happens-before relationship between a terminated task
2367 and its master. This avoids some false positives and avoids a big
2368 memory leak when a lot of Ada tasks are created and terminated.
2369 The interceptions are only activated with forthcoming releases of
2370 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
2372 - A new GDB server monitor command "info locks" giving the list of
2373 locks, their location, and their status.
2377 - callgrind_control now supports the --vgdb-prefix argument,
2378 which is needed if valgrind was started with this same argument.
2380 * ==================== OTHER CHANGES ====================
2382 * Unwinding through inlined function calls. Stack unwinding can now
2383 make use of Dwarf3 inlined-unwind information if it is available.
2384 The practical effect is that inlined calls become visible in stack
2385 traces. The suppression matching machinery has been adjusted
2386 accordingly. This is controlled by the new option
2387 --read-inline-info=yes|no. Currently this is enabled by default
2388 only on Linux and Android targets and only for the tools Memcheck,
2391 * Valgrind can now read EXIDX unwind information on 32-bit ARM
2392 targets. If an object contains both CFI and EXIDX unwind
2393 information, Valgrind will prefer the CFI over the EXIDX. This
2394 facilitates unwinding through system libraries on arm-android
2397 * Address description logic has been improved and is now common
2398 between Memcheck and Helgrind, resulting in better address
2399 descriptions for some kinds of error messages.
2401 * Error messages about dubious arguments (eg, to malloc or calloc) are
2402 output like other errors. This means that they can be suppressed
2403 and they have a stack trace.
2405 * The C++ demangler has been updated for better C++11 support.
2407 * New and modified GDB server monitor features:
2409 - Thread local variables/storage (__thread) can now be displayed.
2411 - The GDB server monitor command "v.info location <address>"
2412 displays information about an address. The information produced
2413 depends on the tool and on the options given to valgrind.
2414 Possibly, the following are described: global variables, local
2415 (stack) variables, allocated or freed blocks, ...
2417 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
2418 ask the GDB server to stop at the start of program execution, at
2419 the end of the program execution and on Valgrind internal errors.
2421 - A new monitor command "v.info stats" shows various Valgrind core
2422 and tool statistics.
2424 - A new monitor command "v.set hostvisibility" allows the GDB server
2425 to provide access to Valgrind internal host status/memory.
2427 * A new option "--aspace-minaddr=<address>" can in some situations
2428 allow the use of more memory by decreasing the address above which
2429 Valgrind maps memory. It can also be used to solve address
2430 conflicts with system libraries by increasing the default value.
2431 See user manual for details.
2433 * The amount of memory used by Valgrind to store debug info (unwind
2434 info, line number information and symbol data) has been
2435 significantly reduced, even though Valgrind now reads more
2436 information in order to support unwinding of inlined function calls.
2438 * Dwarf3 handling with --read-var-info=yes has been improved:
2440 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
2442 - Code compiled with
2443 -ffunction-sections -fdata-sections -Wl,--gc-sections
2444 no longer causes assertion failures.
2446 * Improved checking for the --sim-hints= and --kernel-variant=
2447 options. Unknown strings are now detected and reported to the user
2450 * The semantics of stack start/end boundaries in the valgrind.h
2451 VALGRIND_STACK_REGISTER client request has been clarified and
2452 documented. The convention is that start and end are respectively
2453 the lowest and highest addressable bytes of the stack.
2455 * ==================== FIXED BUGS ====================
2457 The following bugs have been fixed or resolved. Note that "n-i-bz"
2458 stands for "not in bugzilla" -- that is, a bug that was reported to us
2459 but never got a bugzilla entry. We encourage you to file bugs in
2460 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2461 than mailing the developers (or mailing lists) directly -- bugs that
2462 are not entered into bugzilla tend to get forgotten about or ignored.
2464 To see details of a given bug, visit
2465 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2466 where XXXXXX is the bug number as listed below.
2468 175819 Support for ipv6 socket reporting with --track-fds
2469 232510 make distcheck fails
2470 249435 Analyzing wine programs with callgrind triggers a crash
2471 278972 support for inlined function calls in stacktraces and suppression
2473 291310 FXSAVE instruction marks memory as undefined on amd64
2474 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2475 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2476 315199 vgcore file for threaded app does not show which thread crashed
2477 315952 tun/tap ioctls are not supported
2478 323178 Unhandled instruction: PLDW register (ARM)
2479 323179 Unhandled instruction: PLDW immediate (ARM)
2480 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2481 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2482 325124 [MIPSEL] Compilation error
2483 325477 Phase 4 support for IBM Power ISA 2.07
2484 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2485 325628 Phase 5 support for IBM Power ISA 2.07
2486 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2487 325751 Missing the two privileged Power PC Transactional Memory Instructions
2488 325816 Phase 6 support for IBM Power ISA 2.07
2489 325856 Make SGCheck fail gracefully on unsupported platforms
2490 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2491 326436 DRD: False positive in libstdc++ std::list::push_back
2492 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2493 326462 Refactor vgdb to isolate invoker stuff into separate module
2494 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2495 326623 DRD: false positive conflict report in a field assignment
2496 326724 Valgrind does not compile on OSX 1.9 Mavericks
2497 326816 Intercept for __strncpy_sse2_unaligned missing?
2498 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2499 326983 Clear direction flag after tests on amd64.
2500 327212 Do not prepend the current directory to absolute path names.
2501 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2502 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2503 327284 s390x: Fix translation of the risbg instruction
2504 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2505 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2506 327916 DW_TAG_typedef may have no name
2507 327943 s390x: add a redirection for the 'index' function
2508 328100 XABORT not implemented
2509 328205 Implement additional Xen hypercalls
2510 328454 add support Backtraces with ARM unwind tables (EXIDX)
2511 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2512 328711 valgrind.1 manpage "memcheck options" section is badly generated
2513 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2514 329612 Incorrect handling of AT_BASE for image execution
2515 329694 clang warns about using uninitialized variable
2516 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2517 330228 mmap must align to VKI_SHMLBA on mips32
2518 330257 LLVM does not support `-mno-dynamic-no-pic` option
2519 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2520 330459 --track-fds=yes doesn't track eventfds
2521 330469 Add clock_adjtime syscall support
2522 330594 Missing sysalls on PowerPC / uClibc
2523 330622 Add test to regression suite for POWER instruction: dcbzl
2524 330939 Support for AMD's syscall instruction on x86
2526 330941 Typo in PRE(poll) syscall wrapper
2527 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2528 331254 Fix expected output for memcheck/tests/dw4
2529 331255 Fix race condition in test none/tests/coolo_sigaction
2530 331257 Fix type of jump buffer in test none/tests/faultstatus
2531 331305 configure uses bash specific syntax
2532 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2533 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2534 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2535 331829 Unexpected ioctl opcode sign extension
2536 331830 ppc64: WARNING: unhandled syscall: 96/97
2537 331839 drd/tests/sem_open specifies invalid semaphore name
2538 331847 outcome of drd/tests/thread_name is nondeterministic
2539 332037 Valgrind cannot handle Thumb "add pc, reg"
2540 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2541 consistency checks enabled
2542 332263 intercepts for pthread_rwlock_timedrdlock and
2543 pthread_rwlock_timedwrlock are incorrect
2544 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2546 332276 Implement additional Xen hypercalls
2547 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2548 332765 Fix ms_print to create temporary files in a proper directory
2549 333072 drd: Add semaphore annotations
2550 333145 Tests for missaligned PC+#imm access for arm
2551 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2552 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2553 333248 WARNING: unhandled syscall: unix:443
2554 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2555 333501 cachegrind: assertion: Cache set count is not a power of two.
2558 333666 Recognize MPX instructions and bnd prefix.
2559 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2560 333817 Valgrind reports the memory areas written to by the SG_IO
2562 334049 lzcnt fails silently (x86_32)
2563 334384 Valgrind does not have support Little Endian support for
2565 334585 recvmmsg unhandled (+patch) (arm)
2566 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2567 334727 Build fails with -Werror=format-security
2568 334788 clarify doc about --log-file initial program directory
2569 334834 PPC64 Little Endian support, patch 2
2570 334836 PPC64 Little Endian support, patch 3 testcase fixes
2571 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2572 335034 Unhandled ioctl: HCIGETDEVLIST
2573 335155 vgdb, fix error print statement.
2574 335262 arm64: movi 8bit version is not supported
2575 335263 arm64: dmb instruction is not implemented
2576 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2577 335496 arm64: sbc/abc instructions are not implemented
2578 335554 arm64: unhandled instruction: abs
2579 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2580 335735 arm64: unhandled instruction: cnt
2581 335736 arm64: unhandled instruction: uaddlv
2582 335848 arm64: unhandled instruction: {s,u}cvtf
2583 335902 arm64: unhandled instruction: sli
2584 335903 arm64: unhandled instruction: umull (vector)
2585 336055 arm64: unhandled instruction: mov (element)
2586 336062 arm64: unhandled instruction: shrn{,2}
2587 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2588 336189 arm64: unhandled Instruction: mvn
2589 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2590 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2591 336772 Make moans about unknown ioctls more informative
2592 336957 Add a section about the Solaris/illumos port on the webpage
2593 337094 ifunc wrapper is broken on ppc64
2594 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2595 337528 leak check heuristic for block prefixed by length as 64bit number
2596 337740 Implement additional Xen hypercalls
2597 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2598 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2599 337871 deprecate --db-attach
2600 338023 Add support for all V4L2/media ioctls
2601 338024 inlined functions are not shown if DW_AT_ranges is used
2602 338106 Add support for 'kcmp' syscall
2603 338115 DRD: computed conflict set differs from actual after fork
2604 338160 implement display of thread local storage in gdbsrv
2605 338205 configure.ac and check for -Wno-tautological-compare
2606 338300 coredumps are missing one byte of every segment
2607 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2608 338499 --sim-hints parsing broken due to wrong order in tokens
2609 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2610 338681 Unable to unwind through clone thread created on i386-linux
2611 338698 race condition between gdbsrv and vgdb on startup
2612 338703 helgrind on arm-linux gets false positives in dynamic loader
2613 338791 alt dwz files can be relative of debug/main file
2614 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2615 338932 build V-trunk with gcc-trunk
2616 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2617 345079 Fix build problems in VEX/useful/test_main.c
2618 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2619 n-i-bz s390x: Fix memory corruption for multithreaded applications
2620 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2621 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2622 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2623 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2624 n-i-bz Add command-line option --default-suppressions
2625 n-i-bz Add support for BLKDISCARDZEROES ioctl
2626 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2627 n-i-bz Add support for sys_pivot_root and sys_unshare
2629 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2630 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2631 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2635 Release 3.9.0 (31 October 2013)
2636 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2637 3.9.0 is a feature release with many improvements and the usual
2638 collection of bug fixes.
2640 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2641 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2642 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2643 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2645 * ================== PLATFORM CHANGES =================
2647 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2648 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2650 * Support for MIPS DSP ASE on MIPS32 platforms.
2652 * Support for s390x Decimal Floating Point instructions on hosts that
2653 have the DFP facility installed.
2655 * Support for POWER8 (Power ISA 2.07) instructions
2657 * Support for Intel AVX2 instructions. This is available only on 64
2660 * Initial support for Intel Transactional Synchronization Extensions,
2663 * Initial support for Hardware Transactional Memory on POWER.
2665 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2666 run large GUI apps tolerably well.
2668 * ==================== TOOL CHANGES ====================
2672 - Improvements in handling of vectorised code, leading to
2673 significantly fewer false error reports. You need to use the flag
2674 --partial-loads-ok=yes to get the benefits of these changes.
2676 - Better control over the leak checker. It is now possible to
2677 specify which leak kinds (definite/indirect/possible/reachable)
2678 should be displayed, which should be regarded as errors, and which
2679 should be suppressed by a given leak suppression. This is done
2680 using the options --show-leak-kinds=kind1,kind2,..,
2681 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2682 "match-leak-kinds:" line in suppression entries, respectively.
2684 Note that generated leak suppressions contain this new line and
2685 are therefore more specific than in previous releases. To get the
2686 same behaviour as previous releases, remove the "match-leak-kinds:"
2687 line from generated suppressions before using them.
2689 - Reduced "possible leak" reports from the leak checker by the use
2690 of better heuristics. The available heuristics provide detection
2691 of valid interior pointers to std::stdstring, to new[] allocated
2692 arrays with elements having destructors and to interior pointers
2693 pointing to an inner part of a C++ object using multiple
2694 inheritance. They can be selected individually using the
2695 option --leak-check-heuristics=heur1,heur2,...
2697 - Better control of stacktrace acquisition for heap-allocated
2698 blocks. Using the --keep-stacktraces option, it is possible to
2699 control independently whether a stack trace is acquired for each
2700 allocation and deallocation. This can be used to create better
2701 "use after free" errors or to decrease Valgrind's resource
2702 consumption by recording less information.
2704 - Better reporting of leak suppression usage. The list of used
2705 suppressions (shown when the -v option is given) now shows, for
2706 each leak suppressions, how many blocks and bytes it suppressed
2707 during the last leak search.
2711 - False errors resulting from the use of statically initialised
2712 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2715 - False errors resulting from the use of pthread_cond_waits that
2716 timeout, have been removed.
2718 * ==================== OTHER CHANGES ====================
2720 * Some attempt to tune Valgrind's space requirements to the expected
2721 capabilities of the target:
2723 - The default size of the translation cache has been reduced from 8
2724 sectors to 6 on Android platforms, since each sector occupies
2725 about 40MB when using Memcheck.
2727 - The default size of the translation cache has been increased to 16
2728 sectors on all other platforms, reflecting the fact that large
2729 applications require instrumentation and storage of huge amounts
2730 of code. For similar reasons, the number of memory mapped
2731 segments that can be tracked has been increased by a factor of 6.
2733 - In all cases, the maximum number of sectors in the translation
2734 cache can be controlled by the new flag --num-transtab-sectors.
2736 * Changes in how debug info (line numbers, etc) is read:
2738 - Valgrind no longer temporarily mmaps the entire object to read
2739 from it. Instead, reading is done through a small fixed sized
2740 buffer. This avoids virtual memory usage spikes when Valgrind
2741 reads debuginfo from large shared objects.
2743 - A new experimental remote debug info server. Valgrind can read
2744 debug info from a different machine (typically, a build host)
2745 where debuginfo objects are stored. This can save a lot of time
2746 and hassle when running Valgrind on resource-constrained targets
2747 (phones, tablets) when the full debuginfo objects are stored
2748 somewhere else. This is enabled by the --debuginfo-server=
2751 - Consistency checking between main and debug objects can be
2752 disabled using the --allow-mismatched-debuginfo option.
2754 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2755 scanning can recover stack traces in some cases when the normal
2756 unwind mechanisms fail. Stack scanning is best described as "a
2757 nasty, dangerous and misleading hack" and so is disabled by default.
2758 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2761 * Detection and merging of recursive stack frame cycles. When your
2762 program has recursive algorithms, this limits the memory used by
2763 Valgrind for recorded stack traces and avoids recording
2764 uninteresting repeated calls. This is controlled by the command
2765 line option --merge-recursive-frame and by the monitor command
2766 "v.set merge-recursive-frames".
2768 * File name and line numbers for used suppressions. The list of used
2769 suppressions (shown when the -v option is given) now shows, for each
2770 used suppression, the file name and line number where the suppression
2773 * New and modified GDB server monitor features:
2775 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2776 that can be used to execute gdbserver monitor commands from the
2779 - A new monitor command, "v.info open_fds", that gives the list of
2780 open file descriptors and additional details.
2782 - An optional message in the "v.info n_errs_found" monitor command,
2783 for example "v.info n_errs_found test 1234 finished", allowing a
2784 comment string to be added to the process output, perhaps for the
2785 purpose of separating errors of different tests or test phases.
2787 - A new monitor command "v.info execontext" that shows information
2788 about the stack traces recorded by Valgrind.
2790 - A new monitor command "v.do expensive_sanity_check_general" to run
2791 some internal consistency checks.
2793 * New flag --sigill-diagnostics to control whether a diagnostic
2794 message is printed when the JIT encounters an instruction it can't
2795 translate. The actual behavior -- delivery of SIGILL to the
2796 application -- is unchanged.
2798 * The maximum amount of memory that Valgrind can use on 64 bit targets
2799 has been increased from 32GB to 64GB. This should make it possible
2800 to run applications on Memcheck that natively require up to about 35GB.
2802 * ==================== FIXED BUGS ====================
2804 The following bugs have been fixed or resolved. Note that "n-i-bz"
2805 stands for "not in bugzilla" -- that is, a bug that was reported to us
2806 but never got a bugzilla entry. We encourage you to file bugs in
2807 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2808 than mailing the developers (or mailing lists) directly -- bugs that
2809 are not entered into bugzilla tend to get forgotten about or ignored.
2811 To see details of a given bug, visit
2812 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2813 where XXXXXX is the bug number as listed below.
2815 123837 system call: 4th argument is optional, depending on cmd
2816 135425 memcheck should tell you where Freed blocks were Mallocd
2817 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2818 207815 Adds some of the drm ioctls to syswrap-linux.c
2819 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2820 252955 Impossible to compile with ccache
2821 253519 Memcheck reports auxv pointer accesses as invalid reads.
2822 263034 Crash when loading some PPC64 binaries
2823 269599 Increase deepest backtrace
2824 274695 s390x: Support "compare to/from logical" instructions (z196)
2825 275800 s390x: Autodetect cache info (part 2)
2826 280271 Valgrind reports possible memory leaks on still-reachable std::string
2827 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2828 289578 Backtraces with ARM unwind tables (stack scan flags)
2829 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2830 304832 ppc32: build failure
2831 305431 Use find_buildid shdr fallback for separate .debug files
2832 305728 Add support for AVX2 instructions
2833 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2834 306035 s390x: Fix IR generation for LAAG and friends
2835 306054 s390x: Condition code computation for convert-to-int/logical
2836 306098 s390x: alternate opcode form for convert to/from fixed
2837 306587 Fix cache line detection from auxiliary vector for PPC.
2838 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2839 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2840 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2841 307101 sys_capget second argument can be NULL
2842 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2843 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2844 307113 s390x: DFP support
2845 307141 valgrind does't work in mips-linux system
2846 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2847 307285 x86_amd64 feature test for avx in test suite is wrong
2848 307290 memcheck overlap testcase needs memcpy version filter
2849 307463 Please add "&limit=0" to the "all open bugs" link
2850 307465 --show-possibly-lost=no should reduce the error count / exit code
2851 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2852 307729 pkgconfig support broken valgrind.pc
2853 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2854 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2855 308089 Unhandled syscall on ppc64: prctl
2856 308135 PPC32 MPC8xx has 16 bytes cache size
2857 308321 testsuite memcheck filter interferes with gdb_filter
2859 308341 vgdb should report process exit (or fatal signal)
2860 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2861 308495 Remove build dependency on installed Xen headers
2862 308573 Internal error on 64-bit instruction executed in 32-bit mode
2864 308627 pmovmskb validity bit propagation is imprecise
2865 308644 vgdb command for having the info for the track-fds option
2866 308711 give more info about aspacemgr and arenas in out_of_memory
2867 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2868 308718 ARM implement SMLALBB family of instructions
2869 308886 Missing support for PTRACE_SET/GETREGSET
2870 308930 syscall name_to_handle_at (303 on amd64) not handled
2871 309229 V-bit tester does not report number of tests generated
2872 309323 print unrecognized instuction on MIPS
2873 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2874 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2875 309430 Self hosting ppc64 encounters a vassert error on operand type
2876 309600 valgrind is a bit confused about 0-sized sections
2877 309823 Generate errors for still reachable blocks
2878 309921 PCMPISTRI validity bit propagation is imprecise
2879 309922 none/tests/ppc64/test_dfp5 sometimes fails
2880 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2881 310424 --read-var-info does not properly describe static variables
2882 310792 search additional path for debug symbols
2883 310931 s390x: Message-security assist (MSA) instruction extension [..]
2884 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2885 311318 ARM: "128-bit constant is not implemented" error message
2886 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2887 311690 V crashes because it redirects branches inside of a redirected function
2888 311880 x86_64: make regtest hangs at shell_valid1
2889 311922 WARNING: unhandled syscall: 170
2891 312171 ppc: insn selection for DFP
2892 312571 Rounding mode call wrong for the DFP Iops [..]
2893 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2894 312913 Dangling pointers error should also report the alloc stack trace
2895 312980 Building on Mountain Lion generates some compiler warnings
2896 313267 Adding MIPS64/Linux port to Valgrind
2899 313811 Buffer overflow in assert_fail
2900 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2901 314269 ppc: dead code in insn selection
2902 314718 ARM: implement integer divide instruction (sdiv and udiv)
2903 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2904 315441 sendmsg syscall should ignore unset msghdr msg_flags
2905 315534 msgrcv inside a thread causes valgrind to hang (block)
2906 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2907 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2908 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2909 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2910 316144 valgrind.1 manpage contains unknown ??? strings [..]
2911 316145 callgrind command line options in manpage reference (unknown) [..]
2912 316145 callgrind command line options in manpage reference [..]
2913 316181 drd: Fixed a 4x slowdown for certain applications
2914 316503 Valgrind does not support SSE4 "movntdqa" instruction
2915 316535 Use of |signed int| instead of |size_t| in valgrind messages
2916 316696 fluidanimate program of parsec 2.1 stuck
2917 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2918 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2919 317186 "Impossible happens" when occurs VCVT instruction on ARM
2920 317318 Support for Threading Building Blocks "scalable_malloc"
2921 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2922 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2923 317463 bmi testcase IR SANITY CHECK FAILURE
2924 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2925 318050 libmpiwrap fails to compile with out-of-source build
2926 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2927 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2928 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2929 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2930 318932 Add missing PPC64 and PPC32 system call support
2931 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2932 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2933 319494 VEX Makefile-gcc standalone build update after r2702
2934 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2935 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2936 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2937 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2938 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2939 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2940 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2941 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2942 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2943 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2944 320895 add fanotify support (patch included)
2945 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2946 321065 Valgrind updates for Xen 4.3
2947 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2948 321363 Unhandled instruction: SSAX (ARM + Thumb)
2949 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2950 321466 Unhandled instruction: SHASX (ARM + Thumb)
2951 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2952 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2953 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2954 321620 Unhandled instruction: UASX (ARM + Thumb)
2955 321621 Unhandled instruction: USAX (ARM + Thumb)
2956 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2957 321693 Unhandled instruction: LDRSBT (Thumb)
2958 321694 Unhandled instruction: UQASX (ARM + Thumb)
2959 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2960 321697 Unhandled instruction: UHASX (ARM + Thumb)
2961 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2962 321704 Unhandled instruction: REVSH (ARM + Thumb)
2963 321730 Add cg_diff and cg_merge man pages
2964 321738 Add vgdb and valgrind-listener man pages
2966 321891 Unhandled instruction: LDRHT (Thumb)
2967 321960 pthread_create() then alloca() causing invalid stack write errors
2968 321969 ppc32 and ppc64 don't support [lf]setxattr
2969 322254 Show threadname together with tid if set by application
2970 322294 Add initial support for IBM Power ISA 2.07
2971 322368 Assertion failure in wqthread_hijack under OS X 10.8
2972 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2973 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2974 322851 0bXXX binary literal syntax is not standard
2975 323035 Unhandled instruction: LDRSHT(Thumb)
2976 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2977 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2978 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2979 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2980 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2981 323437 Phase 2 support for IBM Power ISA 2.07
2982 323713 Support mmxext (integer sse) subset on i386 (athlon)
2983 323803 Transactional memory instructions are not supported for Power
2984 323893 SSE3 not available on amd cpus in valgrind
2985 323905 Probable false positive from Valgrind/drd on close()
2986 323912 valgrind.h header isn't compatible for mingw64
2987 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
2988 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
2989 324181 mmap does not handle MAP_32BIT
2990 324227 memcheck false positive leak when a thread calls exit+block [..]
2991 324421 Support for fanotify API on ARM architecture
2992 324514 gdbserver monitor cmd output behaviour consistency [..]
2993 324518 ppc64: Emulation of dcbt instructions does not handle [..]
2994 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
2995 324582 When access is made to freed memory, report both allocation [..]
2996 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
2997 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
2998 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
2999 324834 Unhandled instructions in Microsoft C run-time for x86_64
3000 324894 Phase 3 support for IBM Power ISA 2.07
3001 326091 drd: Avoid false race reports from optimized strlen() impls
3002 326113 valgrind libvex hwcaps error on AMD64
3003 n-i-bz Some wrong command line options could be ignored
3004 n-i-bz patch to allow fair-sched on android
3005 n-i-bz report error for vgdb snapshot requested before execution
3006 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
3008 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
3012 Release 3.8.1 (19 September 2012)
3013 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3014 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
3015 that occur moderately frequently in real use cases, adds support for
3016 some missing instructions on ARM, and fixes a deadlock condition on
3017 MacOSX. If you package or deliver 3.8.0 for others to use, you might
3018 want to consider upgrading to 3.8.1 instead.
3020 The following bugs have been fixed or resolved. Note that "n-i-bz"
3021 stands for "not in bugzilla" -- that is, a bug that was reported to us
3022 but never got a bugzilla entry. We encourage you to file bugs in
3023 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3024 than mailing the developers (or mailing lists) directly -- bugs that
3025 are not entered into bugzilla tend to get forgotten about or ignored.
3027 To see details of a given bug, visit
3028 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3029 where XXXXXX is the bug number as listed below.
3032 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
3033 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
3034 298281 wcslen causes false(?) uninitialised value warnings
3035 301281 valgrind hangs on OS X when the process calls system()
3036 304035 disInstr(arm): unhandled instruction 0xE1023053
3037 304867 implement MOVBE instruction in x86 mode
3038 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
3039 305042 amd64: implement 0F 7F encoding of movq between two registers
3040 305199 ARM: implement QDADD and QDSUB
3041 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
3042 305513 killed by fatal signal: SIGSEGV
3043 305690 DRD reporting invalid semaphore when sem_trywait fails
3044 305926 Invalid alignment checks for some AVX instructions
3045 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
3046 306310 3.8.0 release tarball missing some files
3047 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
3048 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
3049 n-i-bz shmat of a segment > 4Gb does not work
3050 n-i-bz simulate_control_c script wrong USR1 signal number on mips
3051 n-i-bz vgdb ptrace calls wrong on mips [...]
3052 n-i-bz Fixes for more MPI false positives
3053 n-i-bz exp-sgcheck's memcpy causes programs to segfault
3054 n-i-bz OSX build w/ clang: asserts at startup
3055 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
3056 n-i-bz fix a couple of union tag-vs-field mixups
3057 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
3059 The following bugs were fixed in 3.8.0 but not listed in this NEWS
3062 254088 Valgrind should know about UD2 instruction
3065 304754 NEWS blows TeX's little mind
3067 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
3071 Release 3.8.0 (10 August 2012)
3072 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3073 3.8.0 is a feature release with many improvements and the usual
3074 collection of bug fixes.
3076 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3077 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
3078 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
3079 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
3080 There is initial support for MacOSX 10.8, but it is not usable for
3081 serious work at present.
3083 * ================== PLATFORM CHANGES =================
3085 * Support for MIPS32 platforms running Linux. Valgrind has been
3086 tested on MIPS32 and MIPS32r2 platforms running different Debian
3087 Squeeze and MeeGo distributions. Both little-endian and big-endian
3088 cores are supported. The tools Memcheck, Massif and Lackey have
3089 been tested and are known to work. See README.mips for more details.
3091 * Preliminary support for Android running on x86.
3093 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
3095 * Support for Intel AVX instructions and for AES instructions. This
3096 support is available only for 64 bit code.
3098 * Support for POWER Decimal Floating Point instructions.
3100 * ==================== TOOL CHANGES ====================
3102 * Non-libc malloc implementations are now supported. This is useful
3103 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
3104 Using the new option --soname-synonyms, such tools can be informed
3105 that the malloc implementation is either linked statically into the
3106 executable, or is present in some other shared library different
3107 from libc.so. This makes it possible to process statically linked
3108 programs, and programs using other malloc libraries, for example
3109 TCMalloc or JEMalloc.
3111 * For tools that provide their own replacement for malloc et al, the
3112 option --redzone-size=<number> allows users to specify the size of
3113 the padding blocks (redzones) added before and after each client
3114 allocated block. Smaller redzones decrease the memory needed by
3115 Valgrind. Bigger redzones increase the chance to detect blocks
3116 overrun or underrun. Prior to this change, the redzone size was
3117 hardwired to 16 bytes in Memcheck.
3121 - The leak_check GDB server monitor command now can
3122 control the maximum nr of loss records to output.
3124 - Reduction of memory use for applications allocating
3125 many blocks and/or having many partially defined bytes.
3127 - Addition of GDB server monitor command 'block_list' that lists
3128 the addresses/sizes of the blocks of a leak search loss record.
3130 - Addition of GDB server monitor command 'who_points_at' that lists
3131 the locations pointing at a block.
3133 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
3134 detect an invalid access of these redzones, by marking them
3135 noaccess. Similarly, if a redzone size is given for a memory
3136 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
3137 This still allows to find some bugs if the user has forgotten to
3138 mark the pool superblock noaccess.
3140 - Performance of memory leak check has been improved, especially in
3141 cases where there are many leaked blocks and/or many suppression
3142 rules used to suppress leak reports.
3144 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
3145 more precise analysis, which is important for LLVM/Clang
3146 generated code. This is at the cost of somewhat reduced
3147 performance. Note there is no change to analysis precision or
3148 costs on Linux targets.
3152 - Added even more facilities that can help finding the cause of a data
3153 race, namely the command-line option --ptrace-addr and the macro
3154 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
3156 - Fixed a subtle bug that could cause false positive data race reports.
3158 * ==================== OTHER CHANGES ====================
3160 * The C++ demangler has been updated so as to work well with C++
3161 compiled by up to at least g++ 4.6.
3163 * Tool developers can make replacement/wrapping more flexible thanks
3164 to the new option --soname-synonyms. This was reported above, but
3165 in fact is very general and applies to all function
3166 replacement/wrapping, not just to malloc-family functions.
3168 * Round-robin scheduling of threads can be selected, using the new
3169 option --fair-sched= yes. Prior to this change, the pipe-based
3170 thread serialisation mechanism (which is still the default) could
3171 give very unfair scheduling. --fair-sched=yes improves
3172 responsiveness of interactive multithreaded applications, and
3173 improves repeatability of results from the thread checkers Helgrind
3176 * For tool developers: support to run Valgrind on Valgrind has been
3177 improved. We can now routinely Valgrind on Helgrind or Memcheck.
3179 * gdbserver now shows the float shadow registers as integer
3180 rather than float values, as the shadow values are mostly
3181 used as bit patterns.
3183 * Increased limit for the --num-callers command line flag to 500.
3185 * Performance improvements for error matching when there are many
3186 suppression records in use.
3188 * Improved support for DWARF4 debugging information (bug 284184).
3190 * Initial support for DWZ compressed Dwarf debug info.
3192 * Improved control over the IR optimiser's handling of the tradeoff
3193 between performance and precision of exceptions. Specifically,
3194 --vex-iropt-precise-memory-exns has been removed and replaced by
3195 --vex-iropt-register-updates, with extended functionality. This
3196 allows the Valgrind gdbserver to always show up to date register
3199 * Modest performance gains through the use of translation chaining for
3202 * ==================== FIXED BUGS ====================
3204 The following bugs have been fixed or resolved. Note that "n-i-bz"
3205 stands for "not in bugzilla" -- that is, a bug that was reported to us
3206 but never got a bugzilla entry. We encourage you to file bugs in
3207 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3208 than mailing the developers (or mailing lists) directly -- bugs that
3209 are not entered into bugzilla tend to get forgotten about or ignored.
3211 To see details of a given bug, visit
3212 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3213 where XXXXXX is the bug number as listed below.
3215 197914 Building valgrind from svn now requires automake-1.10
3216 203877 increase to 16Mb maximum allowed alignment for memalign et al
3217 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
3218 247386 make perf does not run all performance tests
3219 270006 Valgrind scheduler unfair
3220 270777 Adding MIPS/Linux port to Valgrind
3221 270796 s390x: Removed broken support for the TS insn
3222 271438 Fix configure for proper SSE4.2 detection
3223 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
3224 273475 Add support for AVX instructions
3225 274078 improved configure logic for mpicc
3226 276993 fix mremap 'no thrash checks'
3227 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
3228 281482 memcheck incorrect byte allocation count in realloc() for silly argument
3229 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
3230 283413 Fix wrong sanity check
3231 283671 Robustize alignment computation in LibVEX_Alloc
3232 283961 Adding support for some HCI IOCTLs
3233 284124 parse_type_DIE: confused by: DWARF 4
3234 284864 == 273475 (Add support for AVX instructions)
3235 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
3236 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
3237 285725 == 273475 (Add support for AVX instructions)
3238 286261 add wrapper for linux I2C_RDWR ioctl
3239 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
3240 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
3241 286384 configure fails "checking for a supported version of gcc"
3242 286497 == 273475 (Add support for AVX instructions)
3243 286596 == 273475 (Add support for AVX instructions)
3244 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
3245 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
3246 287260 Incorrect conditional jump or move depends on uninitialised value(s)
3247 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
3248 287307 == 273475 (Add support for AVX instructions)
3249 287858 VG_(strerror): unknown error
3250 288298 (MacOSX) unhandled syscall shm_unlink
3251 288995 == 273475 (Add support for AVX instructions)
3252 289470 Loading of large Mach-O thin binaries fails.
3253 289656 == 273475 (Add support for AVX instructions)
3254 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
3255 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
3256 289839 s390x: Provide support for unicode conversion instructions
3257 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
3258 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
3259 290655 Add support for AESKEYGENASSIST instruction
3260 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
3261 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
3262 291253 ES register not initialised in valgrind simulation
3263 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
3264 291865 s390x: Support the "Compare Double and Swap" family of instructions
3265 292300 == 273475 (Add support for AVX instructions)
3266 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
3267 292493 == 273475 (Add support for AVX instructions)
3268 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
3269 292627 Missing support for some SCSI ioctls
3270 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
3271 292841 == 273475 (Add support for AVX instructions)
3272 292993 implement the getcpu syscall on amd64-linux
3273 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
3274 293088 Add some VEX sanity checks for ppc64 unhandled instructions
3275 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
3276 293754 PCMPxSTRx not implemented for 16-bit characters
3277 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
3278 293808 CLFLUSH not supported by latest VEX for amd64
3279 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
3280 294048 MPSADBW instruction not implemented
3281 294055 regtest none/tests/shell fails when locale is not set to C
3282 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
3283 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
3284 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
3285 294260 disInstr_AMD64: disInstr miscalculated next %rip
3286 294523 --partial-loads-ok=yes causes false negatives
3287 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
3288 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
3289 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
3290 295089 can not annotate source for both helgrind and drd
3291 295221 POWER Processor decimal floating point instruction support missing
3292 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
3293 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
3294 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
3295 295617 ARM - Add some missing syscalls
3296 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
3297 296229 Linux user input device ioctls missing wrappers
3298 296318 ELF Debug info improvements (more than one rx/rw mapping)
3299 296422 Add translation chaining support
3300 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
3301 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
3302 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
3303 297078 gdbserver signal handling problems [..]
3304 297147 drd false positives on newly allocated memory
3305 297329 disallow decoding of IBM Power DFP insns on some machines
3306 297497 POWER Processor decimal floating point instruction support missing
3307 297701 Another alias for strncasecmp_l in libc-2.13.so
3308 297911 'invalid write' not reported when using APIs for custom mem allocators.
3309 297976 s390x: revisit EX implementation
3310 297991 Valgrind interferes with mmap()+ftell()
3311 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
3312 297993 Fix compilation of valgrind with gcc -g3.
3313 298080 POWER Processor DFP support missing, part 3
3314 298227 == 273475 (Add support for AVX instructions)
3315 298335 == 273475 (Add support for AVX instructions)
3316 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
3317 298394 s390x: Don't bail out on an unknown machine model. [..]
3318 298421 accept4() syscall (366) support is missing for ARM
3319 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
3320 298732 valgrind installation problem in ubuntu with kernel version 3.x
3321 298862 POWER Processor DFP instruction support missing, part 4
3322 298864 DWARF reader mis-parses DW_FORM_ref_addr
3323 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
3324 299053 Support DWARF4 DW_AT_high_pc constant form
3325 299104 == 273475 (Add support for AVX instructions)
3326 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
3327 299629 dup3() syscall (358) support is missing for ARM
3328 299694 POWER Processor DFP instruction support missing, part 5
3329 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
3330 299803 == 273475 (Add support for AVX instructions)
3331 299804 == 273475 (Add support for AVX instructions)
3332 299805 == 273475 (Add support for AVX instructions)
3333 300140 ARM - Missing (T1) SMMUL
3334 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
3335 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
3336 300414 FCOM and FCOMP unimplemented for amd64 guest
3337 301204 infinite loop in canonicaliseSymtab with ifunc symbol
3338 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
3339 301265 add x86 support to Android build
3340 301984 configure script doesn't detect certain versions of clang
3341 302205 Fix compiler warnings for POWER VEX code and POWER test cases
3342 302287 Unhandled movbe instruction on Atom processors
3343 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
3344 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
3345 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
3346 302656 == 273475 (Add support for AVX instructions)
3347 302709 valgrind for ARM needs extra tls support for android emulator [..]
3348 302827 add wrapper for CDROM_GET_CAPABILITY
3349 302901 Valgrind crashes with dwz optimized debuginfo
3350 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
3351 303116 Add support for the POWER instruction popcntb
3352 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
3353 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
3354 303466 == 273475 (Add support for AVX instructions)
3355 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
3356 303963 strstr() function produces wrong results under valgrind callgrind
3357 304054 CALL_FN_xx macros need to enforce stack alignment
3358 304561 tee system call not supported
3359 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
3360 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
3361 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
3362 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
3363 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
3364 n-i-bz s390x: Shadow registers can now be examined using vgdb
3366 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
3367 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
3371 Release 3.7.0 (5 November 2011)
3372 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3373 3.7.0 is a feature release with many significant improvements and the
3374 usual collection of bug fixes.
3376 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3377 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
3378 Support for recent distros and toolchain components (glibc 2.14, gcc
3379 4.6, MacOSX 10.7) has been added.
3381 * ================== PLATFORM CHANGES =================
3383 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
3384 analyse 64-bit programs running on z/Architecture. Most user space
3385 instructions up to and including z10 are supported. Valgrind has
3386 been tested extensively on z9, z10, and z196 machines running SLES
3387 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
3388 known to work well. Callgrind, Helgrind, and DRD work reasonably
3389 well on z9 and later models. See README.s390 for more details.
3391 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
3392 64-bit processes are supported. Some complex threaded applications
3393 (Firefox) are observed to hang when run as 32 bit applications,
3394 whereas 64-bit versions run OK. The cause is unknown. Memcheck
3395 will likely report some false errors. In general, expect some rough
3396 spots. This release also supports MacOSX 10.6, but drops support
3399 * Preliminary support for Android (on ARM). Valgrind can now run
3400 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
3401 README.android for more details, plus instructions on how to get
3404 * Support for the IBM Power ISA 2.06 (Power7 instructions)
3406 * General correctness and performance improvements for ARM/Linux, and,
3407 by extension, ARM/Android.
3409 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
3410 instruction set support is under development but is not available in
3413 * Support for AIX5 has been removed.
3415 * ==================== TOOL CHANGES ====================
3417 * Memcheck: some incremental changes:
3419 - reduction of memory use in some circumstances
3421 - improved handling of freed memory, which in some circumstances
3422 can cause detection of use-after-free that would previously have
3425 - fix of a longstanding bug that could cause false negatives (missed
3426 errors) in programs doing vector saturated narrowing instructions.
3428 * Helgrind: performance improvements and major memory use reductions,
3429 particularly for large, long running applications which perform many
3430 synchronisation (lock, unlock, etc) events. Plus many smaller
3433 - display of locksets for both threads involved in a race
3435 - general improvements in formatting/clarity of error messages
3437 - addition of facilities and documentation regarding annotation
3438 of thread safe reference counted C++ classes
3440 - new flag --check-stack-refs=no|yes [yes], to disable race checking
3441 on thread stacks (a performance hack)
3443 - new flag --free-is-write=no|yes [no], to enable detection of races
3444 where one thread accesses heap memory but another one frees it,
3445 without any coordinating synchronisation event
3447 * DRD: enabled XML output; added support for delayed thread deletion
3448 in order to detect races that occur close to the end of a thread
3449 (--join-list-vol); fixed a memory leak triggered by repeated client
3450 memory allocatation and deallocation; improved Darwin support.
3452 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
3454 * exp-sgcheck: this tool has been reduced in scope so as to improve
3455 performance and remove checking that Memcheck does better.
3456 Specifically, the ability to check for overruns for stack and global
3457 arrays is unchanged, but the ability to check for overruns of heap
3458 blocks has been removed. The tool has accordingly been renamed to
3459 exp-sgcheck ("Stack and Global Array Checking").
3461 * ==================== OTHER CHANGES ====================
3463 * GDB server: Valgrind now has an embedded GDB server. That means it
3464 is possible to control a Valgrind run from GDB, doing all the usual
3465 things that GDB can do (single stepping, breakpoints, examining
3466 data, etc). Tool-specific functionality is also available. For
3467 example, it is possible to query the definedness state of variables
3468 or memory from within GDB when running Memcheck; arbitrarily large
3469 memory watchpoints are supported, etc. To use the GDB server, start
3470 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3473 * Improved support for unfriendly self-modifying code: a new option
3474 --smc-check=all-non-file is available. This adds the relevant
3475 consistency checks only to code that originates in non-file-backed
3476 mappings. In effect this confines the consistency checking only to
3477 code that is or might be JIT generated, and avoids checks on code
3478 that must have been compiled ahead of time. This significantly
3479 improves performance on applications that generate code at run time.
3481 * It is now possible to build a working Valgrind using Clang-2.9 on
3484 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3485 These enable and disable error reporting on a per-thread, and
3486 nestable, basis. This is useful for hiding errors in particularly
3487 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3488 now uses this facility.
3490 * Added the --mod-funcname option to cg_diff.
3492 * ==================== FIXED BUGS ====================
3494 The following bugs have been fixed or resolved. Note that "n-i-bz"
3495 stands for "not in bugzilla" -- that is, a bug that was reported to us
3496 but never got a bugzilla entry. We encourage you to file bugs in
3497 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3498 mailing the developers (or mailing lists) directly -- bugs that are
3499 not entered into bugzilla tend to get forgotten about or ignored.
3501 To see details of a given bug, visit
3502 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3503 where XXXXXX is the bug number as listed below.
3505 79311 malloc silly arg warning does not give stack trace
3506 210935 port valgrind.h (not valgrind) to win32 to support client requests
3507 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3508 243404 Port to zSeries
3509 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3510 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3511 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3512 253206 Some fixes for the faultstatus testcase
3513 255223 capget testcase fails when running as root
3514 256703 xlc_dbl_u32.c testcase broken
3515 256726 Helgrind tests have broken inline asm
3516 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3517 264800 testcase compile failure on zseries
3518 265762 make public VEX headers compilable by G++ 3.x
3519 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3520 266753 configure script does not give the user the option to not use QtCore
3521 266931 gen_insn_test.pl is broken
3522 266961 ld-linux.so.2 i?86-linux strlen issues
3523 266990 setns instruction causes false positive
3524 267020 Make directory for temporary files configurable at run-time.
3525 267342 == 267997 (segmentation fault on Mac OS 10.6)
3526 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3527 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3528 267488 regtest: darwin support for 64-bit build
3529 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3530 267630 Add support for IBM Power ISA 2.06 -- stage 1
3531 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3532 267819 Add client request for informing the core about reallocation
3533 267925 laog data structure quadratic for a single sequence of lock
3534 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3535 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3536 268513 missed optimizations in fold_Expr
3537 268619 s390x: fpr - gpr transfer facility
3538 268620 s390x: reconsider "long displacement" requirement
3539 268621 s390x: improve IR generation for XC
3540 268715 s390x: FLOGR is not universally available
3541 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3542 268930 s390x: MHY is not universally available
3543 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3544 269079 Support ptrace system call on ARM
3545 269144 missing "Bad option" error message
3546 269209 conditional load and store facility (z196)
3547 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3548 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3549 269736 s390x: minor code generation tweaks
3550 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3551 269863 s390x: remove unused function parameters
3552 269864 s390x: tweak s390_emit_load_cc
3553 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3554 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3555 270115 s390x: rewrite some testcases
3556 270309 == 267997 (valgrind crash on startup)
3557 270320 add support for Linux FIOQSIZE ioctl() call
3558 270326 segfault while trying to sanitize the environment passed to execle
3559 270794 IBM POWER7 support patch causes regression in none/tests
3560 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3561 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3562 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3563 270959 s390x: invalid use of R0 as base register
3564 271042 VSX configure check fails when it should not
3565 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3566 271259 s390x: fix code confusion
3567 271337 == 267997 (Valgrind segfaults on MacOS X)
3568 271385 s390x: Implement Ist_MBE
3569 271501 s390x: misc cleanups
3570 271504 s390x: promote likely and unlikely
3571 271579 ppc: using wrong enum type
3572 271615 unhandled instruction "popcnt" (arch=amd10h)
3573 271730 Fix bug when checking ioctls: duplicate check
3574 271776 s390x: provide STFLE instruction support
3575 271779 s390x: provide clock instructions like STCK
3576 271799 Darwin: ioctls without an arg report a memory error
3577 271820 arm: fix type confusion
3578 271917 pthread_cond_timedwait failure leads to not-locked false positive
3579 272067 s390x: fix DISP20 macro
3580 272615 A typo in debug output in mc_leakcheck.c
3581 272661 callgrind_annotate chokes when run from paths containing regex chars
3582 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3583 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3584 272967 make documentation build-system more robust
3585 272986 Fix gcc-4.6 warnings with valgrind.h
3586 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3587 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3588 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3589 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3590 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3591 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3592 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3593 273778 exp-ptrcheck: unhandled sysno == 259
3594 274089 exp-ptrcheck: unhandled sysno == 208
3595 274378 s390x: Various dispatcher tweaks
3596 274447 WARNING: unhandled syscall: 340
3597 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3598 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3599 274926 valgrind does not build against linux-3
3600 275148 configure FAIL with glibc-2.14
3601 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3602 275168 Make Valgrind work for MacOSX 10.7 Lion
3603 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3604 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3605 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3606 275308 Fix implementation for ppc64 fres instruc
3607 275339 s390x: fix testcase compile warnings
3608 275517 s390x: Provide support for CKSM instruction
3609 275710 s390x: get rid of redundant address mode calculation
3610 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3611 275852 == 250101 (valgrind uses all swap space and is killed)
3612 276784 Add support for IBM Power ISA 2.06 -- stage 3
3613 276987 gdbsrv: fix tests following recent commits
3614 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3615 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3616 277471 Unhandled syscall: 340
3617 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3618 277653 ARM: support Thumb2 PLD instruction
3619 277663 ARM: NEON float VMUL by scalar incorrect
3620 277689 ARM: tests for VSTn with register post-index are broken
3621 277694 ARM: BLX LR instruction broken in ARM mode
3622 277780 ARM: VMOV.F32 (immediate) instruction is broken
3623 278057 fuse filesystem syscall deadlocks
3624 278078 Unimplemented syscall 280 on ppc32
3625 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3626 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3627 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3628 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3629 279027 Support for MVCL and CLCL instruction
3630 279027 s390x: Provide support for CLCL and MVCL instructions
3631 279062 Remove a redundant check in the insn selector for ppc.
3632 279071 JDK creates PTEST with redundant REX.W prefix
3633 279212 gdbsrv: add monitor cmd v.info scheduler.
3634 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3635 279698 memcheck discards valid-bits for packuswb
3636 279795 memcheck reports uninitialised values for mincore on amd64
3637 279994 Add support for IBM Power ISA 2.06 -- stage 3
3638 280083 mempolicy syscall check errors
3639 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3640 280710 s390x: config files for nightly builds
3641 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3642 280965 Valgrind breaks fcntl locks when program does mmap
3643 281138 WARNING: unhandled syscall: 340
3644 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3645 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3646 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3647 281468 s390x: handle do_clone and gcc clones in call traces
3648 281488 ARM: VFP register corruption
3649 281828 == 275284 (false memmove warning: "Source and destination overlap")
3650 281883 s390x: Fix system call wrapper for "clone".
3651 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3652 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3653 282238 SLES10: make check fails
3654 282979 strcasestr needs replacement with recent(>=2.12) glibc
3655 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3656 283243 Regression in ppc64 memcheck tests
3657 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3658 283427 re-connect epoll_pwait syscall on ARM linux
3659 283600 gdbsrv: android: port vgdb.c
3660 283709 none/tests/faultstatus needs to account for page size
3661 284305 filter_gdb needs enhancement to work on ppc64
3662 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3663 284472 Thumb2 ROR.W encoding T2 not implemented
3664 284621 XML-escape process command line in XML output
3665 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3666 that have non-power-of-2 sizes (also AMDs)
3667 n-i-bz don't be spooked by libraries mashed by elfhack
3668 n-i-bz don't be spooked by libxul.so linked with gold
3669 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3671 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3672 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3673 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3677 Release 3.6.1 (16 February 2011)
3678 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3679 3.6.1 is a bug fix release. It adds support for some SSE4
3680 instructions that were omitted in 3.6.0 due to lack of time. Initial
3681 support for glibc-2.13 has been added. A number of bugs causing
3682 crashing or assertion failures have been fixed.
3684 The following bugs have been fixed or resolved. Note that "n-i-bz"
3685 stands for "not in bugzilla" -- that is, a bug that was reported to us
3686 but never got a bugzilla entry. We encourage you to file bugs in
3687 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3688 mailing the developers (or mailing lists) directly -- bugs that are
3689 not entered into bugzilla tend to get forgotten about or ignored.
3691 To see details of a given bug, visit
3692 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3693 where XXXXXX is the bug number as listed below.
3695 188572 Valgrind on Mac should suppress setenv() mem leak
3696 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3697 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3698 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3699 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3700 254420 memory pool tracking broken
3701 254957 Test code failing to compile due to changes in memcheck.h
3702 255009 helgrind/drd: crash on chmod with invalid parameter
3703 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3704 255355 helgrind/drd: crash on threaded programs doing fork
3706 255418 (SSE4.x) rint call compiled with ICC
3707 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3708 255888 closing valgrindoutput tag outputted to log-stream on error
3709 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3710 255966 Slowness when using mempool annotations
3711 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3712 256600 super-optimized strcasecmp() false positive
3713 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3714 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3715 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3716 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3717 257276 Missing case in memcheck --track-origins=yes
3718 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3719 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3720 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3721 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3722 263099 callgrind_annotate counts Ir improperly [...]
3723 263877 undefined coprocessor instruction on ARMv7
3724 265964 configure FAIL with glibc-2.13
3725 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3726 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3727 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3728 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3729 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3731 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3735 Release 3.6.0 (21 October 2010)
3736 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3737 3.6.0 is a feature release with many significant improvements and the
3738 usual collection of bug fixes.
3740 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3741 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3742 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3744 -------------------------
3746 Here are some highlights. Details are shown further down:
3748 * Support for ARM/Linux.
3750 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3752 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3754 * Support for the SSE4.2 instruction set.
3756 * Enhancements to the Callgrind profiler, including the ability to
3757 handle CPUs with three levels of cache.
3759 * A new experimental heap profiler, DHAT.
3761 * A huge number of bug fixes and small enhancements.
3763 -------------------------
3765 Here are details of the above changes, together with descriptions of
3766 many other changes, and a list of fixed bugs.
3768 * ================== PLATFORM CHANGES =================
3770 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3771 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3772 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3774 This requires a CPU capable of running the ARMv7-A instruction set
3775 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3776 of the user space instruction set, including ARM and Thumb integer
3777 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3778 Cachegrind and Massif tools work properly; other tools work to
3781 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3782 with support for recent releases of the underlying toolchain
3783 components, notably gcc-4.5 and glibc-2.12.
3785 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3786 support also works much better on OS X 10.5, and is as solid as
3789 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3790 64-bit mode. In 32-bit mode, support is only available up to and
3791 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3792 supported in 64-bit mode, and 32-bit mode does in fact support the
3793 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3794 10.6 on 32-bit targets.
3796 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3797 and including version 2.05 is supported.
3799 * ==================== TOOL CHANGES ====================
3801 * Cachegrind has a new processing script, cg_diff, which finds the
3802 difference between two profiles. It's very useful for evaluating
3803 the performance effects of a change in a program.
3805 Related to this change, the meaning of cg_annotate's (rarely-used)
3806 --threshold option has changed; this is unlikely to affect many
3807 people, if you do use it please see the user manual for details.
3809 * Callgrind now can do branch prediction simulation, similar to
3810 Cachegrind. In addition, it optionally can count the number of
3811 executed global bus events. Both can be used for a better
3812 approximation of a "Cycle Estimation" as derived event (you need to
3813 update the event formula in KCachegrind yourself).
3815 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3816 rather than the L2 cache. This is to accommodate machines with
3817 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3818 cache configuration of such a machine it will run the simulation as
3819 if the L2 cache isn't present. This means the results are less
3820 likely to match the true result for the machine, but
3821 Cachegrind/Callgrind's results are already only approximate, and
3822 should not be considered authoritative. The results are still
3823 useful for giving a general idea about a program's locality.
3825 * Massif has a new option, --pages-as-heap, which is disabled by
3826 default. When enabled, instead of tracking allocations at the level
3827 of heap blocks (as allocated with malloc/new/new[]), it instead
3828 tracks memory allocations at the level of memory pages (as mapped by
3829 mmap, brk, etc). Each mapped page is treated as its own block.
3830 Interpreting the page-level output is harder than the heap-level
3831 output, but this option is useful if you want to account for every
3832 byte of memory used by a program.
3834 * DRD has two new command-line options: --free-is-write and
3835 --trace-alloc. The former allows to detect reading from already freed
3836 memory, and the latter allows tracing of all memory allocations and
3839 * DRD has several new annotations. Custom barrier implementations can
3840 now be annotated, as well as benign races on static variables.
3842 * DRD's happens before / happens after annotations have been made more
3843 powerful, so that they can now also be used to annotate e.g. a smart
3844 pointer implementation.
3846 * Helgrind's annotation set has also been drastically improved, so as
3847 to provide to users a general set of annotations to describe locks,
3848 semaphores, barriers and condition variables. Annotations to
3849 describe thread-safe reference counted heap objects have also been
3852 * Memcheck has a new command-line option, --show-possibly-lost, which
3853 is enabled by default. When disabled, the leak detector will not
3854 show possibly-lost blocks.
3856 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3857 has been added. DHAT keeps track of allocated heap blocks, and also
3858 inspects every memory reference to see which block (if any) is being
3859 accessed. This gives a lot of insight into block lifetimes,
3860 utilisation, turnover, liveness, and the location of hot and cold
3861 fields. You can use DHAT to do hot-field profiling.
3863 * ==================== OTHER CHANGES ====================
3865 * Improved support for unfriendly self-modifying code: the extra
3866 overhead incurred by --smc-check=all has been reduced by
3867 approximately a factor of 5 as compared with 3.5.0.
3869 * Ability to show directory names for source files in error messages.
3870 This is combined with a flexible mechanism for specifying which
3871 parts of the paths should be shown. This is enabled by the new flag
3874 * A new flag, --require-text-symbol, which will stop the run if a
3875 specified symbol is not found it a given shared object when it is
3876 loaded into the process. This makes advanced working with function
3877 intercepting and wrapping safer and more reliable.
3879 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3880 and control of Valgrind is now available for the tools Memcheck and
3881 Helgrind. XML output from Valgrind is available for Memcheck,
3882 Helgrind and exp-Ptrcheck.
3884 * More reliable stack unwinding on amd64-linux, particularly in the
3885 presence of function wrappers, and with gcc-4.5 compiled code.
3887 * Modest scalability (performance improvements) for massive
3888 long-running applications, particularly for those with huge amounts
3891 * Support for analyzing programs running under Wine with has been
3892 improved. The header files <valgrind/valgrind.h>,
3893 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3894 Windows-programs compiled with MinGW or one of the Microsoft Visual
3897 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3898 The 32-bit simulator was not affected. This did not occur often,
3899 but when it did would usually crash the program under test.
3902 * A large number of bugs were fixed. These are shown below.
3904 * A number of bugs were investigated, and were candidates for fixing,
3905 but are not fixed in 3.6.0, due to lack of developer time. They may
3906 get fixed in later releases. They are:
3908 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3909 212419 false positive "lock order violated" (A+B vs A)
3910 213685 Undefined value propagates past dependency breaking instruction
3911 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3912 237920 valgrind segfault on fork failure
3913 242137 support for code compiled by LLVM-2.8
3914 242423 Another unknown Intel cache config value
3915 243232 Inconsistent Lock Orderings report with trylock
3916 243483 ppc: callgrind triggers VEX assertion failure
3917 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3918 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3920 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3921 249435 Analyzing wine programs with callgrind triggers a crash
3922 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3923 250065 Handling large allocations
3924 250101 huge "free" memory usage due to m_mallocfree.c
3925 "superblocks fragmentation"
3926 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3927 252091 Callgrind on ARM does not detect function returns correctly
3928 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3929 254420 memory pool tracking broken
3930 n-i-bz support for adding symbols for JIT generated code
3933 The following bugs have been fixed or resolved. Note that "n-i-bz"
3934 stands for "not in bugzilla" -- that is, a bug that was reported to us
3935 but never got a bugzilla entry. We encourage you to file bugs in
3936 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3937 mailing the developers (or mailing lists) directly -- bugs that are
3938 not entered into bugzilla tend to get forgotten about or ignored.
3940 To see details of a given bug, visit
3941 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3942 where XXXXXX is the bug number as listed below.
3944 135264 dcbzl instruction missing
3946 153699 Valgrind should report unaligned reads with movdqa
3948 190429 Valgrind reports lost of errors in ld.so
3949 with x86_64 2.9.90 glibc
3950 197266 valgrind appears to choke on the xmms instruction
3952 197988 Crash when demangling very large symbol names
3953 202315 unhandled syscall: 332 (inotify_init1)
3954 203256 Add page-level profiling to Massif
3955 205093 dsymutil=yes needs quotes, locking (partial fix)
3956 205241 Snow Leopard 10.6 support (partial fix)
3957 206600 Leak checker fails to upgrade indirect blocks when their
3958 parent becomes reachable
3959 210935 port valgrind.h (not valgrind) to win32 so apps run under
3960 wine can make client requests
3961 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3962 within Linux ip-stack checksum functions
3963 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3965 213685 Undefined value propagates past dependency breaking instruction
3967 215914 Valgrind inserts bogus empty environment variable
3969 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3970 222545 shmat fails under valgind on some arm targets
3971 222560 ARM NEON support
3974 232509 Docs build fails with formatting inside <title></title> elements
3976 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3977 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3978 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3979 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3980 237485 symlink (syscall 57) is not supported on Mac OS
3981 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3983 238208 is_just_below_ESP doesn't take into account red-zone
3984 238345 valgrind passes wrong $0 when executing a shell script
3985 238679 mq_timedreceive syscall doesn't flag the reception buffer
3987 238696 fcntl command F_DUPFD_CLOEXEC not supported
3988 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3989 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3990 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
3992 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
3998 242606 unhandled syscall: setegid (in Ptrcheck)
3999 242814 Helgrind "Impossible has happened" during
4000 QApplication::initInstance();
4001 243064 Valgrind attempting to read debug information from iso
4002 243270 Make stack unwinding in Valgrind wrappers more reliable
4003 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
4004 sysno = 277 (mq_open)
4005 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
4006 244493 ARM VFP d16-d31 registers support
4007 244670 add support for audit_session_self syscall on Mac OS 10.6
4008 244921 The xml report of helgrind tool is not well format
4009 244923 In the xml report file, the <preamble> not escape the
4010 xml char, eg '<','&','>'
4011 245535 print full path names in plain text reports
4012 245925 x86-64 red zone handling problem
4013 246258 Valgrind not catching integer underruns + new [] s
4014 246311 reg/reg cmpxchg doesn't work on amd64
4015 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
4016 246888 Improve Makefile.vex.am
4017 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
4018 to [f]chmod_extended
4019 247526 IBM POWER6 (ISA 2.05) support is incomplete
4020 247561 Some leak testcases fails due to reachable addresses in
4022 247875 sizeofIRType to handle Ity_I128
4023 247894 [PATCH] unhandled syscall sys_readahead
4024 247980 Doesn't honor CFLAGS passed to configure
4025 248373 darwin10.supp is empty in the trunk
4026 248822 Linux FIBMAP ioctl has int parameter instead of long
4027 248893 [PATCH] make readdwarf.c big endianess safe to enable
4028 unwinding on big endian systems
4029 249224 Syscall 336 not supported (SYS_proc_info)
4031 249775 Incorrect scheme for detecting NEON capabilities of host CPU
4032 249943 jni JVM init fails when using valgrind
4033 249991 Valgrind incorrectly declares AESKEYGENASSIST support
4035 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
4036 250799 frexp$fenv_access_off function generates SIGILL
4037 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
4038 251251 support pclmulqdq insn
4039 251362 valgrind: ARM: attach to debugger either fails or provokes
4041 251674 Unhandled syscall 294
4044 254257 Add support for debugfiles found by build-id
4045 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
4046 254646 Wrapped functions cause stack misalignment on OS X
4047 (and possibly Linux)
4048 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
4050 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
4054 Release 3.5.0 (19 August 2009)
4055 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4056 3.5.0 is a feature release with many significant improvements and the
4057 usual collection of bug fixes. The main improvement is that Valgrind
4058 now works on Mac OS X.
4060 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
4061 and X86/Darwin. Support for recent distros and toolchain components
4062 (glibc 2.10, gcc 4.5) has been added.
4064 -------------------------
4066 Here is a short summary of the changes. Details are shown further
4069 * Support for Mac OS X (10.5.x).
4071 * Improvements and simplifications to Memcheck's leak checker.
4073 * Clarification and simplifications in various aspects of Valgrind's
4076 * XML output for Helgrind and Ptrcheck.
4078 * Performance and stability improvements for Helgrind and DRD.
4080 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
4082 * A new experimental tool, BBV, useful for computer architecture
4085 * Improved Wine support, including ability to read Windows PDB
4088 -------------------------
4090 Here are details of the above changes, followed by descriptions of
4091 many other minor changes, and a list of fixed bugs.
4094 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
4095 called "Darwin" because that is the name of the OS core, which is the
4096 level that Valgrind works at.)
4100 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
4101 because it would require work and 10.4 is only becoming less common.
4103 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
4104 fairly well. For 10.5.x, 32-bit programs are the default even on
4105 64-bit machines, so it handles most current programs.
4107 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
4108 officially supported, but simple programs at least will probably work.
4109 However, start-up is slow.
4111 - PowerPC machines are not supported.
4113 Things that don't work:
4115 - The Ptrcheck tool.
4117 - Objective-C garbage collection.
4121 - If you have Rogue Amoeba's "Instant Hijack" program installed,
4122 Valgrind will fail with a SIGTRAP at start-up. See
4123 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
4128 - You will likely find --dsymutil=yes a useful option, as error
4129 messages may be imprecise without it.
4131 - Mac OS X support is new and therefore will be less robust than the
4132 Linux support. Please report any bugs you find.
4134 - Threaded programs may run more slowly than on Linux.
4136 Many thanks to Greg Parker for developing this port over several years.
4139 * Memcheck's leak checker has been improved.
4141 - The results for --leak-check=summary now match the summary results
4142 for --leak-check=full. Previously they could differ because
4143 --leak-check=summary counted "indirectly lost" blocks and
4144 "suppressed" blocks as "definitely lost".
4146 - Blocks that are only reachable via at least one interior-pointer,
4147 but are directly pointed to by a start-pointer, were previously
4148 marked as "still reachable". They are now correctly marked as
4151 - The default value for the --leak-resolution option has been
4152 changed from "low" to "high". In general, this means that more
4153 leak reports will be produced, but each leak report will describe
4154 fewer leaked blocks.
4156 - With --leak-check=full, "definitely lost" and "possibly lost"
4157 leaks are now considered as proper errors, ie. they are counted
4158 for the "ERROR SUMMARY" and affect the behaviour of
4159 --error-exitcode. These leaks are not counted as errors if
4160 --leak-check=summary is specified, however.
4162 - Documentation for the leak checker has been improved.
4165 * Various aspects of Valgrind's text output have changed.
4167 - Valgrind's start-up message has changed. It is shorter but also
4168 includes the command being run, which makes it easier to use
4169 --trace-children=yes. An example:
4171 - Valgrind's shut-down messages have also changed. This is most
4172 noticeable with Memcheck, where the leak summary now occurs before
4173 the error summary. This change was necessary to allow leaks to be
4174 counted as proper errors (see the description of the leak checker
4175 changes above for more details). This was also necessary to fix a
4176 longstanding bug in which uses of suppressions against leaks were
4177 not "counted", leading to difficulties in maintaining suppression
4178 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
4180 - Behavior of -v has changed. In previous versions, -v printed out
4181 a mixture of marginally-user-useful information, and tool/core
4182 statistics. The statistics printing has now been moved to its own
4183 flag, --stats=yes. This means -v is less verbose and more likely
4184 to convey useful end-user information.
4186 - The format of some (non-XML) stack trace entries has changed a
4187 little. Previously there were six possible forms:
4189 0x80483BF: really (a.c:20)
4190 0x80483BF: really (in /foo/a.out)
4192 0x80483BF: (within /foo/a.out)
4193 0x80483BF: ??? (a.c:20)
4196 The third and fourth of these forms have been made more consistent
4197 with the others. The six possible forms are now:
4199 0x80483BF: really (a.c:20)
4200 0x80483BF: really (in /foo/a.out)
4201 0x80483BF: really (in ???)
4202 0x80483BF: ??? (in /foo/a.out)
4203 0x80483BF: ??? (a.c:20)
4206 Stack traces produced when --xml=yes is specified are different
4210 * Helgrind and Ptrcheck now support XML output, so they can be used
4211 from GUI tools. Also, the XML output mechanism has been
4214 - The XML format has been overhauled and generalised, so it is more
4215 suitable for error reporting tools in general. The Memcheck
4216 specific aspects of it have been removed. The new format, which
4217 is an evolution of the old format, is described in
4218 docs/internals/xml-output-protocol4.txt.
4220 - Memcheck has been updated to use the new format.
4222 - Helgrind and Ptrcheck are now able to emit output in this format.
4224 - The XML output mechanism has been overhauled. XML is now output
4225 to its own file descriptor, which means that:
4227 * Valgrind can output text and XML independently.
4229 * The longstanding problem of XML output being corrupted by
4230 unexpected un-tagged text messages is solved.
4232 As before, the destination for text output is specified using
4233 --log-file=, --log-fd= or --log-socket=.
4235 As before, XML output for a tool is enabled using --xml=yes.
4237 Because there's a new XML output channel, the XML output
4238 destination is now specified by --xml-file=, --xml-fd= or
4241 Initial feedback has shown this causes some confusion. To
4242 clarify, the two envisaged usage scenarios are:
4244 (1) Normal text output. In this case, do not specify --xml=yes
4245 nor any of --xml-file=, --xml-fd= or --xml-socket=.
4247 (2) XML output. In this case, specify --xml=yes, and one of
4248 --xml-file=, --xml-fd= or --xml-socket= to select the XML
4249 destination, one of --log-file=, --log-fd= or --log-socket=
4250 to select the destination for any remaining text messages,
4251 and, importantly, -q.
4253 -q makes Valgrind completely silent on the text channel,
4254 except in the case of critical failures, such as Valgrind
4255 itself segfaulting, or failing to read debugging information.
4256 Hence, in this scenario, it suffices to check whether or not
4257 any output appeared on the text channel. If yes, then it is
4258 likely to be a critical error which should be brought to the
4259 attention of the user. If no (the text channel produced no
4260 output) then it can be assumed that the run was successful.
4262 This allows GUIs to make the critical distinction they need to
4263 make (did the run fail or not?) without having to search or
4264 filter the text output channel in any way.
4266 It is also recommended to use --child-silent-after-fork=yes in
4270 * Improvements and changes in Helgrind:
4272 - XML output, as described above
4274 - Checks for consistent association between pthread condition
4275 variables and their associated mutexes are now performed.
4277 - pthread_spinlock functions are supported.
4279 - Modest performance improvements.
4281 - Initial (skeletal) support for describing the behaviour of
4282 non-POSIX synchronisation objects through ThreadSanitizer
4283 compatible ANNOTATE_* macros.
4285 - More controllable tradeoffs between performance and the level of
4286 detail of "previous" accesses in a race. There are now three
4289 * --history-level=full. This is the default, and was also the
4290 default in 3.4.x. It shows both stacks involved in a race, but
4291 requires a lot of memory and can be very slow in programs that
4292 do many inter-thread synchronisation events.
4294 * --history-level=none. This only shows the later stack involved
4295 in a race. This can be much faster than --history-level=full,
4296 but makes it much more difficult to find the other access
4297 involved in the race.
4299 The new intermediate setting is
4301 * --history-level=approx
4303 For the earlier (other) access, two stacks are presented. The
4304 earlier access is guaranteed to be somewhere in between the two
4305 program points denoted by those stacks. This is not as useful
4306 as showing the exact stack for the previous access (as per
4307 --history-level=full), but it is better than nothing, and it's
4308 almost as fast as --history-level=none.
4311 * New features and improvements in DRD:
4313 - The error messages printed by DRD are now easier to interpret.
4314 Instead of using two different numbers to identify each thread
4315 (Valgrind thread ID and DRD thread ID), DRD does now identify
4316 threads via a single number (the DRD thread ID). Furthermore
4317 "first observed at" information is now printed for all error
4318 messages related to synchronization objects.
4320 - Added support for named semaphores (sem_open() and sem_close()).
4322 - Race conditions between pthread_barrier_wait() and
4323 pthread_barrier_destroy() calls are now reported.
4325 - Added support for custom allocators through the macros
4326 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
4327 in <valgrind/valgrind.h>). An alternative for these two macros is
4328 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
4331 - Added support for annotating non-POSIX synchronization objects
4332 through several new ANNOTATE_*() macros.
4334 - OpenMP: added support for the OpenMP runtime (libgomp) included
4335 with gcc versions 4.3.0 and 4.4.0.
4339 - Added two new command-line options (--first-race-only and
4340 --segment-merging-interval).
4343 * Genuinely atomic support for x86/amd64/ppc atomic instructions
4345 Valgrind will now preserve (memory-access) atomicity of LOCK-
4346 prefixed x86/amd64 instructions, and any others implying a global
4347 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
4349 This means that Valgrinded processes will "play nicely" in
4350 situations where communication with other processes, or the kernel,
4351 is done through shared memory and coordinated with such atomic
4352 instructions. Prior to this change, such arrangements usually
4353 resulted in hangs, races or other synchronisation failures, because
4354 Valgrind did not honour atomicity of such instructions.
4357 * A new experimental tool, BBV, has been added. BBV generates basic
4358 block vectors for use with the SimPoint analysis tool, which allows
4359 a program's overall behaviour to be approximated by running only a
4360 fraction of it. This is useful for computer architecture
4361 researchers. You can run BBV by specifying --tool=exp-bbv (the
4362 "exp-" prefix is short for "experimental"). BBV was written by
4366 * Modestly improved support for running Windows applications under
4367 Wine. In particular, initial support for reading Windows .PDB debug
4368 information has been added.
4371 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
4372 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
4376 * The Valgrind client requests VALGRIND_PRINTF and
4377 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
4378 the string was always printed immediately on its own line. Now, the
4379 string will be added to a buffer but not printed until a newline is
4380 encountered, or other Valgrind output is printed (note that for
4381 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
4382 "other Valgrind output"). This allows you to use multiple
4383 VALGRIND_PRINTF calls to build up a single output line, and also to
4384 print multiple output lines with a single request (by embedding
4385 multiple newlines in the string).
4388 * The graphs drawn by Massif's ms_print program have changed slightly:
4390 - The half-height chars '.' and ',' are no longer drawn, because
4391 they are confusing. The --y option can be used if the default
4392 y-resolution is not high enough.
4394 - Horizontal lines are now drawn after the top of a snapshot if
4395 there is a gap until the next snapshot. This makes it clear that
4396 the memory usage has not dropped to zero between snapshots.
4399 * Something that happened in 3.4.0, but wasn't clearly announced: the
4400 option --read-var-info=yes can be used by some tools (Memcheck,
4401 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
4402 variable type and location information. This makes those tools
4403 start up more slowly and increases memory consumption, but
4404 descriptions of data addresses in error messages become more
4408 * exp-Omega, an experimental instantaneous leak-detecting tool, was
4409 disabled in 3.4.0 due to a lack of interest and maintenance,
4410 although the source code was still in the distribution. The source
4411 code has now been removed from the distribution. For anyone
4412 interested, the removal occurred in SVN revision r10247.
4415 * Some changes have been made to the build system.
4417 - VEX/ is now integrated properly into the build system. This means
4418 that dependency tracking within VEX/ now works properly, "make
4419 install" will work without requiring "make" before it, and
4420 parallel builds (ie. 'make -j') now work (previously a
4421 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
4422 was effectively ignored).
4424 - The --with-vex configure option has been removed. It was of
4425 little use and removing it simplified the build system.
4427 - The location of some install files has changed. This should not
4428 affect most users. Those who might be affected:
4430 * For people who use Valgrind with MPI programs, the installed
4431 libmpiwrap.so library has moved from
4432 $(INSTALL)/<platform>/libmpiwrap.so to
4433 $(INSTALL)/libmpiwrap-<platform>.so.
4435 * For people who distribute standalone Valgrind tools, the
4436 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
4437 have moved to $(INSTALL)/libcoregrind-<platform>.a.
4439 These changes simplify the build system.
4441 - Previously, all the distributed suppression (*.supp) files were
4442 installed. Now, only default.supp is installed. This should not
4443 affect users as the other installed suppression files were not
4444 read; the fact that they were installed was a mistake.
4447 * KNOWN LIMITATIONS:
4449 - Memcheck is unusable with the Intel compiler suite version 11.1,
4450 when it generates code for SSE2-and-above capable targets. This
4451 is because of icc's use of highly optimised inlined strlen
4452 implementations. It causes Memcheck to report huge numbers of
4453 false errors even in simple programs. Helgrind and DRD may also
4456 Versions 11.0 and earlier may be OK, but this has not been
4460 The following bugs have been fixed or resolved. Note that "n-i-bz"
4461 stands for "not in bugzilla" -- that is, a bug that was reported to us
4462 but never got a bugzilla entry. We encourage you to file bugs in
4463 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4464 mailing the developers (or mailing lists) directly -- bugs that are
4465 not entered into bugzilla tend to get forgotten about or ignored.
4467 To see details of a given bug, visit
4468 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4469 where XXXXXX is the bug number as listed below.
4471 84303 How about a LockCheck tool?
4472 91633 dereference of null ptr in vgPlain_st_basetype
4473 97452 Valgrind doesn't report any pthreads problems
4474 100628 leak-check gets assertion failure when using
4475 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4476 108528 NPTL pthread cleanup handlers not called
4477 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4478 110128 mallinfo is not implemented...
4479 110770 VEX: Generated files not always updated when making valgrind
4480 111102 Memcheck: problems with large (memory footprint) applications
4481 115673 Vex's decoder should never assert
4482 117564 False positive: Syscall param clone(child_tidptr) contains
4483 uninitialised byte(s)
4484 119404 executing ssh from inside valgrind fails
4485 133679 Callgrind does not write path names to sources with dwarf debug
4487 135847 configure.in problem with non gnu compilers (and possible fix)
4488 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4489 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4490 136230 memcheck reports "possibly lost", should be "still reachable"
4491 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4492 137904 Valgrind reports a memory leak when using POSIX threads,
4494 139076 valgrind VT_GETSTATE error
4495 142228 complaint of elf_dynamic_do_rela in trivial usage
4496 145347 spurious warning with USBDEVFS_REAPURB
4497 148441 (wine) can't find memory leak in Wine, win32 binary
4499 148742 Leak-check fails assert on exit
4500 149878 add (proper) check for calloc integer overflow
4501 150606 Call graph is broken when using callgrind control
4502 152393 leak errors produce an exit code of 0. I need some way to
4503 cause leak errors to result in a nonzero exit code.
4504 157154 documentation (leak-resolution doc speaks about num-callers
4505 def=4) + what is a loss record
4506 159501 incorrect handling of ALSA ioctls
4507 162020 Valgrinding an empty/zero-byte file crashes valgrind
4508 162482 ppc: Valgrind crashes while reading stabs information
4509 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4510 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4511 163560 VEX/test_main.c is missing from valgrind-3.3.1
4512 164353 malloc_usable_size() doesn't return a usable size
4513 165468 Inconsistent formatting in memcheck manual -- please fix
4514 169505 main.c:286 (endOfInstr):
4515 Assertion 'ii->cost_offset == *cost_offset' failed
4516 177206 Generate default.supp during compile instead of configure
4517 177209 Configure valt_load_address based on arch+os
4518 177305 eventfd / syscall 323 patch lost
4519 179731 Tests fail to build because of inlining of non-local asm labels
4520 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4521 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4522 181594 Bogus warning for empty text segment
4523 181707 dwarf doesn't require enumerations to have name
4524 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4525 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4526 Assertion '!already_present' failed.
4527 185359 exp-ptrcheck: unhandled syscall getresuid()
4528 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4529 185816 Valgrind is unable to handle debug info for files with split
4530 debug info that are prelinked afterwards
4531 185980 [darwin] unhandled syscall: sem_open
4532 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4533 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4534 186790 Suppression pattern used for leaks are not reported
4535 186796 Symbols with length>200 in suppression files are ignored
4536 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4537 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4538 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4539 188046 bashisms in the configure script
4540 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4541 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4542 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4543 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4544 assertion fail "!lock->heldBy"
4545 188427 Add support for epoll_create1 (with patch)
4546 188530 Support for SIOCGSTAMPNS
4547 188560 Include valgrind.spec in the tarball
4548 188572 Valgrind on Mac should suppress setenv() mem leak
4549 189054 Valgrind fails to build because of duplicate non-local asm labels
4550 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4551 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4552 189763 drd assertion failure: s_threadinfo[tid].is_recording
4553 190219 unhandled syscall: 328 (x86-linux)
4554 190391 dup of 181394; see above
4555 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4556 190820 No debug information on powerpc-linux
4557 191095 PATCH: Improve usbdevfs ioctl handling
4558 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4560 191189 --xml=yes should obey --gen-suppressions=all
4561 191192 syslog() needs a suppression on macosx
4562 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4563 191761 getrlimit on MacOSX
4564 191992 multiple --fn-skip only works sometimes; dependent on order
4565 192634 V. reports "aspacem sync_check_mapping_callback:
4566 segment mismatch" on Darwin
4567 192954 __extension__ missing on 2 client requests
4568 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4569 194474 "INSTALL" file has different build instructions than "README"
4570 194671 Unhandled syscall (sem_wait?) from mac valgrind
4571 195069 memcheck: reports leak (memory still reachable) for
4573 195169 drd: (vgDrd_barrier_post_wait):
4574 Assertion 'r->sg[p->post_iteration]' failed.
4575 195268 valgrind --log-file doesn't accept ~/...
4576 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4577 195860 WARNING: unhandled syscall: unix:223
4578 196528 need a error suppression for pthread_rwlock_init under os x?
4579 197227 Support aio_* syscalls on Darwin
4580 197456 valgrind should reject --suppressions=(directory)
4581 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4582 197591 unhandled syscall 27 (mincore)
4583 197793 Merge DCAS branch to the trunk == 85756, 142103
4584 197794 Avoid duplicate filenames in Vex
4585 197898 make check fails on current SVN
4586 197901 make check fails also under exp-ptrcheck in current SVN
4587 197929 Make --leak-resolution=high the default
4588 197930 Reduce spacing between leak reports
4589 197933 Print command line of client at start-up, and shorten preamble
4590 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4591 198395 add BBV to the distribution as an experimental tool
4592 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4593 198649 callgrind_annotate doesn't cumulate counters
4594 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4595 199977 Valgrind complains about an unrecognized instruction in the
4596 atomic_incs test program
4597 200029 valgrind isn't able to read Fedora 12 debuginfo
4598 200760 darwin unhandled syscall: unix:284
4599 200827 DRD doesn't work on Mac OS X
4600 200990 VG_(read_millisecond_timer)() does not work correctly
4601 201016 Valgrind does not support pthread_kill() on Mac OS
4602 201169 Document --read-var-info
4603 201323 Pre-3.5.0 performance sanity checking
4604 201384 Review user manual for the 3.5.0 release
4605 201585 mfpvr not implemented on ppc
4606 201708 tests failing because x86 direction flag is left set
4607 201757 Valgrind doesn't handle any recent sys_futex additions
4608 204377 64-bit valgrind can not start a shell script
4609 (with #!/path/to/shell) if the shell is a 32-bit executable
4610 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4611 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4612 about memory allocation events with memory access tracing enabled
4613 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4615 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4619 Release 3.4.1 (28 February 2009)
4620 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4621 3.4.1 is a bug-fix release that fixes some regressions and assertion
4622 failures in debug info reading in 3.4.0, most notably incorrect stack
4623 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4624 other debug info problems are also fixed. A number of bugs in the
4625 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4627 In view of the fact that 3.4.0 contains user-visible regressions
4628 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4629 encouraged to ship 3.4.1 in preference to 3.4.0.
4631 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4632 bugzilla" -- that is, a bug that was reported to us but never got a
4633 bugzilla entry. We encourage you to file bugs in bugzilla
4634 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4635 developers (or mailing lists) directly -- bugs that are not entered
4636 into bugzilla tend to get forgotten about or ignored.
4638 n-i-bz Fix various bugs reading icc-11 generated debug info
4639 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4640 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4641 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4642 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4643 179618 exp-ptrcheck crashed / exit prematurely
4644 179624 helgrind: false positive races with pthread_create and
4645 recv/open/close/read
4646 134207 pkg-config output contains @VG_PLATFORM@
4647 176926 floating point exception at valgrind startup with PPC 440EPX
4648 181594 Bogus warning for empty text segment
4649 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4650 181707 Dwarf3 doesn't require enumerations to have name
4651 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4652 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4653 Assertion '!already_present' failed.
4654 185359 exp-ptrcheck unhandled syscall getresuid()
4656 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4657 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4661 Release 3.4.0 (2 January 2009)
4662 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4663 3.4.0 is a feature release with many significant improvements and the
4664 usual collection of bug fixes. This release supports X86/Linux,
4665 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4666 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4668 3.4.0 brings some significant tool improvements. Memcheck can now
4669 report the origin of uninitialised values, the thread checkers
4670 Helgrind and DRD are much improved, and we have a new experimental
4671 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4672 global arrays. In detail:
4674 * Memcheck is now able to track the origin of uninitialised values.
4675 When it reports an uninitialised value error, it will try to show
4676 the origin of the value, as either a heap or stack allocation.
4677 Origin tracking is expensive and so is not enabled by default. To
4678 use it, specify --track-origins=yes. Memcheck's speed will be
4679 essentially halved, and memory usage will be significantly
4680 increased. Nevertheless it can drastically reduce the effort
4681 required to identify the root cause of uninitialised value errors,
4682 and so is often a programmer productivity win, despite running more
4685 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4686 3.4.0, will be released shortly.
4688 * Helgrind's race detection algorithm has been completely redesigned
4689 and reimplemented, to address usability and scalability concerns:
4691 - The new algorithm has a lower false-error rate: it is much less
4692 likely to report races that do not really exist.
4694 - Helgrind will display full call stacks for both accesses involved
4695 in a race. This makes it easier to identify the root causes of
4698 - Limitations on the size of program that can run have been removed.
4700 - Performance has been modestly improved, although that is very
4703 - Direct support for Qt4 threading has been added.
4705 - pthread_barriers are now directly supported.
4707 - Helgrind works well on all supported Linux targets.
4709 * The DRD thread debugging tool has seen major improvements:
4711 - Greatly improved performance and significantly reduced memory
4714 - Support for several major threading libraries (Boost.Thread, Qt4,
4715 glib, OpenMP) has been added.
4717 - Support for atomic instructions, POSIX semaphores, barriers and
4718 reader-writer locks has been added.
4720 - Works now on PowerPC CPUs too.
4722 - Added support for printing thread stack usage at thread exit time.
4724 - Added support for debugging lock contention.
4726 - Added a manual for Drd.
4728 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4729 checks for misuses of pointers. In that sense it is a bit like
4730 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4731 detect overruns of stack and global arrays, it can detect
4732 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4733 detect accesses heap blocks that have been freed a very long time
4734 ago (millions of blocks in the past).
4736 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4737 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4738 of the main Valgrind documentation. As this is an experimental
4739 tool, we would be particularly interested in hearing about your
4740 experiences with it.
4742 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4743 longer built by default, although the code remains in the repository
4744 and the tarball. This is due to three factors: a perceived lack of
4745 users, a lack of maintenance, and concerns that it may not be
4746 possible to achieve reliable operation using the existing design.
4748 * As usual, support for the latest Linux distros and toolchain
4749 components has been added. It should work well on Fedora Core 10,
4750 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4751 state) is supported, as is glibc-2.9. The C++ demangler has been
4752 updated so as to work well with C++ compiled by even the most recent
4755 * You can now use frame-level wildcards in suppressions. This was a
4756 frequently-requested enhancement. A line "..." in a suppression now
4757 matches zero or more frames. This makes it easier to write
4758 suppressions which are precise yet insensitive to changes in
4761 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4763 * Very basic support for IBM Power6 has been added (64-bit processes only).
4765 * Valgrind is now cross-compilable. For example, it is possible to
4766 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4767 on a ppc32/64-linux target.
4769 * You can set the main thread's stack size at startup using the
4770 new --main-stacksize= flag (subject of course to ulimit settings).
4771 This is useful for running apps that need a lot of stack space.
4773 * The limitation that you can't use --trace-children=yes together
4774 with --db-attach=yes has been removed.
4776 * The following bugs have been fixed. Note that "n-i-bz" stands for
4777 "not in bugzilla" -- that is, a bug that was reported to us but
4778 never got a bugzilla entry. We encourage you to file bugs in
4779 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4780 mailing the developers (or mailing lists) directly.
4782 n-i-bz Make return types for some client requests 64-bit clean
4783 n-i-bz glibc 2.9 support
4784 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4785 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4786 n-i-bz Building in an env without gdb gives bogus gdb attach
4787 92456 Tracing the origin of uninitialised memory
4788 106497 Valgrind does not demangle some C++ template symbols
4790 151612 Suppression with "..." (frame-level wildcards in .supp files)
4791 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4792 159285 unhandled syscall:25 (stime, on x86-linux)
4793 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4794 160954 ppc build of valgrind crashes with illegal instruction (isel)
4795 160956 mallinfo implementation, w/ patch
4796 162092 Valgrind fails to start gnome-system-monitor
4797 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4798 163794 assertion failure with "--track-origins=yes"
4799 163933 sigcontext.err and .trapno must be set together
4800 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4801 164476 Missing kernel module loading system calls
4802 164669 SVN regression: mmap() drops posix file locks
4803 166581 Callgrind output corruption when program forks
4804 167288 Patch file for missing system calls on Cell BE
4805 168943 unsupported scas instruction pentium
4806 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4807 172417 x86->IR: 0x82 ...
4808 172563 amd64->IR: 0xD9 0xF5 - fprem1
4809 173099 .lds linker script generation error
4810 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4811 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4813 174908 --log-file value not expanded correctly for core file
4814 175044 Add lookup_dcookie for amd64
4815 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4817 Developer-visible changes:
4819 * Valgrind's debug-info reading machinery has been majorly overhauled.
4820 It can now correctly establish the addresses for ELF data symbols,
4821 which is something that has never worked properly before now.
4823 Also, Valgrind can now read DWARF3 type and location information for
4824 stack and global variables. This makes it possible to use the
4825 framework to build tools that rely on knowing the type and locations
4826 of stack and global variables, for example exp-Ptrcheck.
4828 Reading of such information is disabled by default, because most
4829 tools don't need it, and because it is expensive in space and time.
4830 However, you can force Valgrind to read it, using the
4831 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4832 make use of such information, if present, to provide source-level
4833 descriptions of data addresses in the error messages they create.
4835 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4836 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).