1 Release 3.22.0 (?? Oct 2023)
2 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
5 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
6 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
7 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
8 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
9 AMD64/macOS 10.13 and nanoMIPS/Linux.
11 * ==================== CORE CHANGES ===================
13 * A new configure option --with-gdbscripts-dir lets you install
14 the gdb valgrind python monitor scripts in a specific location.
15 For example an distro could use it to install the scripts in a
16 safe load location --with-gdbscripts-dir=%{_datadir}/gdb/auto-load
17 It is also possible to configure --without-gdb-scripts-dir so no
18 .debug_gdb_scripts section is added to the vgpreload library and
19 no valgrind-monitor python scripts are installed at all.
21 * ================== PLATFORM CHANGES =================
24 * ==================== TOOL CHANGES ===================
27 * ==================== FIXED BUGS ====================
29 The following bugs have been fixed or resolved. Note that "n-i-bz"
30 stands for "not in bugzilla" -- that is, a bug that was reported to us
31 but never got a bugzilla entry. We encourage you to file bugs in
32 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
33 than mailing the developers (or mailing lists) directly -- bugs that
34 are not entered into bugzilla tend to get forgotten about or ignored.
36 460192 Add epoll_pwait2
37 469049 link failure on ppc64 (big endian) valgrind 3.20
38 469146 massif --ignore-fn does not ignore inlined functions
39 469768 Make it possible to install gdb scripts in a different location
40 470520 Multiple realloc zero errors crash in MC_(eq_Error)
41 470713 Failure on the Yosys project: valgrind: m_libcfile.c:1802
42 (Bool vgPlain_realpath(const HChar *, HChar *)): Assertion 'resolved' failed
44 To see details of a given bug, visit
45 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
46 where XXXXXX is the bug number as listed above.
48 Release 3.21.0 (28 Apr 2023)
49 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
51 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
52 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
53 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
54 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
55 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
56 AMD64/macOS 10.13 and nanoMIPS/Linux.
58 * ==================== CORE CHANGES ===================
60 * When GDB is used to debug a program running under valgrind using
61 the valgrind gdbserver, GDB will automatically load some
62 python code provided in valgrind defining GDB front end commands
63 corresponding to the valgrind monitor commands.
64 These GDB front end commands accept the same format as
65 the monitor commands directly sent to the Valgrind gdbserver.
66 These GDB front end commands provide a better integration
67 in the GDB command line interface, so as to use for example
68 GDB auto-completion, command specific help, searching for
69 a command or command help matching a regexp, ...
70 For relevant monitor commands, GDB will evaluate arguments
71 to make the use of monitor commands easier.
72 For example, instead of having to print the address of a variable
73 to pass it to a subsequent monitor command, the GDB front end
74 command will evaluate the address argument. It is for example
76 (gdb) memcheck who_points_at &some_struct sizeof(some_struct)
79 $2 = (some_struct_type *) 0x1130a0 <some_struct>
80 (gdb) p sizeof(some_struct)
82 (gdb) monitor who_point_at 0x1130a0 40
84 * The vgdb utility now supports extended-remote protocol when
85 invoked with --multi. In this mode the GDB run command is
86 supported. Which means you don't need to run gdb and valgrind
87 from different terminals. So for example to start your program
88 in gdb and run it under valgrind you can do:
90 (gdb) set remote exec-file prog
92 (gdb) target extended-remote | vgdb --multi
95 * The behaviour of realloc with a size of zero can now
96 be changed for tools that intercept malloc. Those
97 tools are memcheck, helgrind, drd, massif and dhat.
98 Realloc implementations generally do one of two things
99 - free the memory like free() and return NULL
100 (GNU libc and ptmalloc).
101 - either free the memory and then allocate a
102 minimum sized block or just return the
103 original pointer. Return NULL if the
104 allocation of the minimum sized block fails
105 (jemalloc, musl, snmalloc, Solaris, macOS).
106 When Valgrind is configured and built it will
107 try to match the OS and libc behaviour. However
108 if you are using a non-default library to replace
109 malloc and family (e.g., musl on a glibc Linux or
110 tcmalloc on FreeBSD) then you can use a command line
111 option to change the behaviour of Valgrind:
112 --realloc-zero-bytes-frees=yes|no [yes on Linux glibc, no otherwise]
114 * ================== PLATFORM CHANGES =================
116 * Make the address space limit on FreeBSD amd64 128Gbytes
117 (the same as Linux and Solaris, it was 32Gbytes)
119 * ==================== TOOL CHANGES ===================
122 - When doing a delta leak_search, it is now possible to only
123 output the new loss records compared to the previous leak search.
124 This is available in the memcheck monitor command 'leak_search'
125 by specifying the "new" keyword or in your program by using
126 the client request VALGRIND_DO_NEW_LEAK_CHECK.
127 Whenever a "delta" leak search is done (i.e. when specifying
128 "new" or "increased" or "changed" in the monitor command),
129 the new loss records have a "new" marker.
130 - Valgrind now contains python code that defines GDB memcheck
131 front end monitor commands. See CORE CHANGES.
132 - Performs checks for the use of realloc with a size of zero.
133 This is non-portable and a source of errors. If memcheck
134 detects such a usage it will generate an error
135 realloc() with size 0
136 followed by the usual callstacks.
137 A switch has been added to allow this to be turned off:
138 --show-realloc-size-zero=yes|no [yes]
141 - The option ---history-backtrace-size=<number> allows to configure
142 the number of entries to record in the stack traces of "old"
143 accesses. Previously, this number was hardcoded to 8.
144 - Valgrind now contains python code that defines GDB helgrind
145 front end monitor commands. See CORE CHANGES.
148 - `--cache-sim=no` is now the default. The cache simulation is old and
149 unlikely to match any real modern machine. This means only the `Ir`
150 event are gathered by default, but that is by far the most useful
152 - `cg_annotate`, `cg_diff`, and `cg_merge` have been rewritten in
153 Python. As a result, they all have more flexible command line
154 argument handling, e.g. supporting `--show-percs` and
155 `--no-show-percs` forms as well as the existing `--show-percs=yes`
156 and `--show-percs=no`.
157 - `cg_annotate` has some functional changes.
158 - It's much faster, e.g. 3-4x on common cases.
159 - It now supports diffing (with `--diff`, `--mod-filename`, and
160 `--mod-funcname`) and merging (by passing multiple data files).
161 - It now provides more information at the file and function level.
162 There are now "File:function" and "Function:file" sections. These
163 are very useful for programs that use inlining a lot.
164 - Support for user-annotated files and the `-I`/`--include` option
165 has been removed, because it was of little use and blocked other
167 - The `--auto` option is renamed `--annotate`, though the old
168 `--auto=yes`/`--auto=no` forms are still supported.
169 - `cg_diff` and `cg_merge` are now deprecated, because `cg_annotate`
170 now does a better job of diffing and merging.
171 - The Cachegrind output file format has changed very slightly, but in
172 ways nobody is likely to notice.
175 - Valgrind now contains python code that defines GDB callgrind
176 front end monitor commands. See CORE CHANGES.
179 - Valgrind now contains python code that defines GDB massif
180 front end monitor commands. See CORE CHANGES.
183 - A new kind of user request has been added which allows you to
184 override the 1024 byte limit on access count histograms for blocks
185 of memory. The client request is DHAT_HISTOGRAM_MEMORY.
187 * ==================== FIXED BUGS ====================
189 The following bugs have been fixed or resolved. Note that "n-i-bz"
190 stands for "not in bugzilla" -- that is, a bug that was reported to us
191 but never got a bugzilla entry. We encourage you to file bugs in
192 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
193 than mailing the developers (or mailing lists) directly -- bugs that
194 are not entered into bugzilla tend to get forgotten about or ignored.
196 170510 Don't warn about ioctl of size 0 without direction hint
197 241072 List tools in --help output
198 327548 false positive while destroying mutex
199 382034 Testcases build fixes for musl
200 351857 confusing error message about valid command line option
201 374596 inconsistent RDTSCP support on x86_64
202 392331 Spurious lock not held error from inside pthread_cond_timedwait
203 397083 Likely false positive "uninitialised value(s)" for __wmemchr_avx2 and __wmemcmp_avx2_movbe
204 400793 pthread_rwlock_timedwrlock false positive
205 419054 Unhandled syscall getcpu on arm32
206 433873 openat2 syscall unimplemented on Linux
207 434057 Add stdio mode to valgrind's gdbserver
208 435441 valgrind fails to interpose malloc on musl 1.2.2 due to weak symbol name and no libc soname
209 436413 Warn about realloc of size zero
210 439685 compiler warning in callgrind/main.c
211 444110 priv/guest_ppc_toIR.c:36198:31: warning: duplicated 'if' condition.
212 444487 hginfo test detects an extra lock inside data symbol "_rtld_local"
213 444488 Use glibc.pthread.stack_cache_size tunable
214 444568 drd/tests/pth_barrier_thr_cr fails on Fedora 38
215 445743 "The impossible happened: mutex is locked simultaneously by two threads"
216 while using mutexes with priority inheritance and signals
217 449309 Missing loopback device ioctl(s)
218 459476 vgdb: allow address reuse to avoid "address already in use" errorsuse" errors
219 460356 s390: Sqrt32Fx4 -- cannot reduce tree
220 462830 WARNING: unhandled amd64-freebsd syscall: 474
221 463027 broken check for MPX instruction support in assembler
222 464103 Enhancement: add a client request to DHAT to mark memory to be histogrammed
223 464476 Firefox fails to start under Valgrind
224 464609 Valgrind memcheck should support Linux pidfd_open
225 464680 Show issues caused by memory policies like selinux deny_execmem
226 464859 Build failures with GCC-13 (drd tsan_unittest)
227 464969 D language demangling
228 465435 m_libcfile.c:66 (vgPlain_safe_fd): Assertion 'newfd >= VG_(fd_hard_limit)' failed.
229 466104 aligned_alloc problems, part 1
230 467036 Add time cost statistics for Regtest
231 467482 Build failure on aarch64 Alpine
232 467714 fdleak_* and rlimit tests fail when parent process has more than
233 64 descriptors opened
234 467839 Gdbserver: Improve compatibility of library directory name
235 468401 [PATCH] Add a style file for clang-format
236 468556 Build failure for vgdb
237 468606 build: remove "Valgrind relies on GCC" check/output
238 469097 ppc64(be) doesn't support SCV syscall instruction
239 n-i-bz FreeBSD rfork syscall fail with EINVAL or ENOSYS rather than VG_(unimplemented)
241 To see details of a given bug, visit
242 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
243 where XXXXXX is the bug number as listed above.
245 * ==================== KNOWN ISSUES ===================
247 * configure --enable-lto=yes is know to not work in all setups.
248 See bug 469049. Workaround: Build without LTO.
250 (3.21.0.RC1: 14 Apr 2023)
251 (3.21.0.RC2: 21 Apr 2023)
253 Release 3.20.0 (24 Oct 2022)
254 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
256 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
257 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
258 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
259 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
260 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
261 AMD64/macOS 10.13 and nanoMIPS/Linux.
263 * ==================== CORE CHANGES ===================
265 * The option "--vgdb-stop-at=event1,event2,..." accepts the new value abexit.
266 This indicates to invoke gdbserver when your program exits abnormally
267 (i.e. with a non zero exit code).
268 * Fix Rust v0 name demangling.
269 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
270 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
271 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with
273 * The option --enable-debuginfod=<no|yes> [default: yes] has been added on
275 * More DWARF5 support as generated by clang14.
277 * ==================== FIXED BUGS ====================
279 The following bugs have been fixed or resolved. Note that "n-i-bz"
280 stands for "not in bugzilla" -- that is, a bug that was reported to us
281 but never got a bugzilla entry. We encourage you to file bugs in
282 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
283 than mailing the developers (or mailing lists) directly -- bugs that
284 are not entered into bugzilla tend to get forgotten about or ignored.
286 131186 writev reports error in (vector[...])
287 434764 iconv_open causes ld.so v2.28+ to use optimised strncmp
288 446754 Improve error codes from alloc functions under memcheck
289 452274 memcheck crashes with Assertion 'sci->status.what == SsIdle' failed
290 452779 Valgrind fails to build on FreeBSD 13.0 with llvm-devel (15.0.0)
291 453055 shared_timed_mutex drd test fails with "Lock shared failed" message
292 453602 Missing command line option to enable/disable debuginfod
293 452802 Handle lld 9+ split RW PT_LOAD segments correctly
294 454040 s390x: False-positive memcheck:cond in memmem on arch13 systems
295 456171 [PATCH] FreeBSD: Don't record address errors when accessing the 'kern.ps_strings' sysctl struct
296 n-i-bz Implement vgdb invoker on FreeBSD
297 458845 PowerPC: The L field for the dcbf and sync instruction should be
299 458915 Remove register cache to fix 458915 gdbserver causes wrong syscall return
300 459031 Documentation on --error-exitcode incomplete
301 459477 XERROR messages lacks ending '\n' in vgdb
302 462007 Implicit int in none/tests/faultstatus.c
304 To see details of a given bug, visit
305 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
306 where XXXXXX is the bug number as listed above.
308 (3.20.0.RC1: 20 Oct 2022)
311 Release 3.19.0 (11 Apr 2022)
312 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
314 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
315 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
316 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
317 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
318 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
319 AMD64/macOS 10.13 and nanoMIPS/Linux.
321 * ==================== CORE CHANGES ===================
323 * Fix Rust v0 name demangling.
324 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
325 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
326 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with HardenedBSD
328 * ================== PLATFORM CHANGES =================
331 - ignore the "v8.x" architecture levels, only look at actual CPU features
332 present. Fixes mismatch detected between RDMA and atomics features
333 preventing startup on some QEMU configurations.
334 - Implement LD{,A}XP and ST{,L}XP
335 - Fix incorrect code emitted for doubleword CAS.
338 - Fix sys_ipc semtimedop syscall
339 - Fix VFLRX and WFLRX instructions
340 - Fix EXRL instruction with negative offset
343 - Reimplement the vbpermq instruction support to generate less Iops and
344 avoid overflowing internal buffers.
345 - Fix checking for scv support to avoid "Facility 'SCV' unavailable (12),
346 exception" messages in dmsg.
347 - Fix setting condition code for Vector Compare quad word instructions.
348 - Fix fix lxsibzx, lxsihzx and lxsihzx instructions so they only load
349 their respective sized data.
350 - Fix the prefixed stq instruction in PC relative mode.
352 * ==================== TOOL CHANGES ===================
355 - Speed up --track-origins=yes for large (in the range of hundreds to
356 thousands of megabytes) mmap/munmaps.
358 - Several fixes for new versions of libstd++ using new posix try_lock
361 * ==================== FIXED BUGS ====================
363 The following bugs have been fixed or resolved. Note that "n-i-bz"
364 stands for "not in bugzilla" -- that is, a bug that was reported to us
365 but never got a bugzilla entry. We encourage you to file bugs in
366 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
367 than mailing the developers (or mailing lists) directly -- bugs that
368 are not entered into bugzilla tend to get forgotten about or ignored.
370 403802 leak_cpp_interior fails with some reachable blocks different than expected
371 435732 memcheck/tests/leak_cpp_interior fails with gcc11
372 444242 s390x: Valgrind crashes on EXRL with negative offset
373 444399 arm64: unhandled instruction 0xC87F2D89 (LD{,A}XP and ST{,L}XP).
375 444481 gdb_server test failures on s390x
376 444495 dhat/tests/copy fails on s390x
377 444552 memcheck/tests/sem fails on s390x with glibc 2.34
378 444571 PPC, fix the lxsibzx and lxsihzx so they only load their respective
380 444836 PPC, pstq instruction for R=1 is not storing to the correct address.
381 444925 fexecve syscall wrapper not properly implemented
382 445032 valgrind/memcheck crash with SIGSEGV when SIGVTALRM timer used and
384 445211 Fix out of tree builds
385 445300 [PATCH] Fix building tests with Musl
386 445011 SIGCHLD is sent when valgrind uses debuginfod-find
387 445354 arm64 backend: incorrect code emitted for doubleword CAS
388 445415 arm64 front end: alignment checks missing for atomic instructions
389 445504 Using C++ condition_variable results in bogus "mutex is locked simultaneously by two threads" warning
390 445607 Unhandled amd64-freebsd syscall: 247
391 445668 Inline stack frame generation is broken for Rust binaries
392 445916 Demangle Rust v0 symbols with .llvm suffix
393 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false positives
394 446138 DRD/Helgrind with std::timed_mutex::try_lock_until false positives
395 446281 Add a DRD suppression for fwrite
396 446103 Memcheck: `--track-origins=yes` causes extreme slowdowns for large mmap/munmap
397 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false
398 446251 TARGET_SIGNAL_THR added to enum target_signal
399 446823 FreeBSD - missing syscalls when using libzm4
400 447991 s390x: Valgrind indicates illegal instruction on wflrx
401 447995 Valgrind segfault on power10 due to hwcap checking code
402 449483 Powerpc: vcmpgtsq., vcmpgtuq,, vcmpequq. instructions not setting the
403 condition code correctly.
404 449672 ppc64 --track-origins=yes failures because of bad cmov addHRegUse
405 449838 sigsegv liburing the 'impossible' happened for io_uring_setup
406 450025 Powerc: ACC file not implemented as a logical overlay of the VSR
408 450437 Warn for execve syscall with argv or argv[0] being NULL
409 450536 Powerpc: valgrind throws 'facility scv unavailable exception'
410 451626 Syscall param bpf(attr->raw_tracepoint.name) points to unaddressable byte(s)
411 451827 [ppc64le] VEX temporary storage exhausted with several vbpermq instructions
412 451843 valgrind fails to start on a FreeBSD system which enforces W^X
414 To see details of a given bug, visit
415 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
416 where XXXXXX is the bug number as listed above.
418 (3.19.0.RC1: 02 Apr 2022)
419 (3.19.0.RC2: 08 Apr 2022)
422 Release 3.18.0 (15 Oct 2021)
423 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
425 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
426 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
427 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
428 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
429 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
430 AMD64/macOS 10.13 and nanoMIPS/Linux.
432 * ==================== CORE CHANGES ===================
434 * The libiberty demangler has been updated, which brings support for
435 Rust v0 name demangling. [Update: alas, due to a bug, this support
436 isn't working in 3.18.0.]
438 * __libc_freeres isn't called anymore after the program recieves a
439 fatal signal. Causing some internal glibc resources to hang around,
440 but preventing any crashes after the program has ended.
442 * The DWARF reader is now very much faster at startup when just
443 --read-inline-info=yes (the default in most cases) is given.
445 * glibc 2.34, which moved various functions from libpthread.so into
446 libc.so, is now supported.
448 * ================== PLATFORM CHANGES =================
452 - v8.2 scalar and vector FABD, FACGE, FACGT and FADD.
453 - v8.2 FP compare & conditional compare instructions.
454 - Zero variants of v8.2 FP compare instructions.
458 - Support the miscellaneous-instruction-extensions facility 3 and
459 the vector-enhancements facility 2. This enables programs
460 compiled with "-march=arch13" or "-march=z15" to be executed
465 - ISA 3.1 support is now complete
466 - ISA 3.0 support for the darn instruction added.
467 - ISA 3.0 support for the vector system call instruction scv added.
468 - ISA 3.0 support for the copy, paste and cpabort instructions added.
470 * Support for X86/FreeBSD and AMD64/FreeBSD has been added.
472 * ==================== OTHER CHANGES ====================
474 * Memcheck on amd64: minor fixes to remove some false positive
477 * ==================== FIXED BUGS ====================
479 The following bugs have been fixed or resolved. Note that "n-i-bz"
480 stands for "not in bugzilla" -- that is, a bug that was reported to us
481 but never got a bugzilla entry. We encourage you to file bugs in
482 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
483 than mailing the developers (or mailing lists) directly -- bugs that
484 are not entered into bugzilla tend to get forgotten about or ignored.
486 208531 [PATCH]: FreeBSD support for valgrind
487 368960 WARNING: unhandled amd64-linux syscall: 163 (acct)
488 407589 [Linux] Add support for C11 aligned_alloc() and GNU reallocarray()
489 423963 Error in child thread when CLONE_PIDFD is used
490 426148 crash with "impossible happened" when running BPF CO-RE programs
491 429375 PPC ISA 3.1 support is missing, part 9
492 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
493 431306 Update demangler to support Rust v0 name mangling
494 432387 s390x: z15 instructions support
495 433437 FreeBSD support, part 1
496 433438 FreeBSD support, part 2
497 433439 FreeBSD support, part 3
498 433469 FreeBSD support, part 4
499 433473 FreeBSD support, part 5
500 433477 FreeBSD support, part 6
501 433479 FreeBSD support, part 7
502 433504 FreeBSD support, part 8
503 433506 FreeBSD support, part 9
504 433507 FreeBSD support, part 10
505 433508 FreeBSD support, part 11
506 433510 FreeBSD support, part 12
507 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
508 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
509 434296 s390x: False-positive memcheck diagnostics from vector string
511 434840 PPC64 darn instruction not supported
512 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
513 435908 valgrind tries to fetch from deubginfod for files which already
514 have debug information
515 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2
516 439046 valgrind is unusably large when linked with lld
517 439090 Implement close_range(2)
518 439326 Valgrind 3.17.0 won't compile with Intel 2021 oneAPI compilers
519 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
520 440670 unhandled ppc64le-linux syscall: 252 statfs64 and 253 fstatfs64
521 440906 Fix impossible constraint issue in P10 testcase.
522 441512 Remove a unneeded / unnecessary prefix check.
523 441534 Update the expected output for test_isa_3_1_VRT.
524 442061 very slow execution under Fedora 34 (readdwarf3)
525 443031 Gcc -many change requires explicit .machine directives
526 443033 Add support for the ISA 3.0 mcrxrx instruction
527 443034 Sraw, srawi, srad, sradi, mfs
528 443178 Powerpc, test jm-mfspr expected output needs to be updated.
529 443179 Need new test for the lxvx and stxvx instructions on ISA 2.07 and
531 443180 The subnormal test and the ISA 3.0 test generate compiler warnings
532 443314 In the latest GIT version, Valgrind with "--trace-flags" crashes
534 443605 Don't call final_tidyup (__libc_freeres) on FatalSignal
536 To see details of a given bug, visit
537 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
538 where XXXXXX is the bug number as listed below.
540 (3.18.0.RC1: 12 Oct 2021)
541 (3.18.0: 15 Oct 2021)
545 Release 3.17.0 (19 Mar 2021)
546 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
548 3.17.0 fixes a number of bugs and adds some functional changes: support for
549 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
550 some new instructions for Arm64, S390 and POWER. There are also some tool
553 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
554 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
555 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
556 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
557 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
559 * ==================== CORE CHANGES ===================
561 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
564 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
565 debugging information. When a debuginfo file cannot be found locally,
566 Valgrind is able to query debuginfod servers for the file using its
567 build-id. See the user manual for more information about debuginfod support.
569 * ================== PLATFORM CHANGES =================
573 - Inaccuracies resulting from double-rounding in the simulation of
574 floating-point multiply-add/subtract instructions have been fixed. These
575 should now behave exactly as the hardware does.
577 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
578 ongoing. Support for the half-word variants of at least the following
579 instructions has been added:
581 FABS <Vd>.<T>, <Vn>.<T>
583 FNEG <Vd>.<T>, <Vn>.<T>
585 FSQRT <Vd>.<T>, <Vn>.<T>
590 - Implement the new instructions/features that were added to z/Architecture
591 with the vector-enhancements facility 1. Also cover the instructions from
592 the vector-packed-decimal facility that are defined outside the chapter
593 "Vector Decimal Instructions", but not the ones from that chapter itself.
595 For a detailed list of newly supported instructions see the updates to
596 `docs/internals/s390-opcodes.csv'.
598 Since the miscellaneous instruction extensions facility 2 was already
599 added in Valgrind 3.16.0, this completes the support necessary to run
600 general programs built with `--march=z14' under Valgrind. The
601 vector-packed-decimal facility is currently not exploited by the standard
602 toolchain and libraries.
606 - Various bug fixes. Fix for the sync field to limit setting just two of
607 the two bits in the L-field. Fix the write size for the stxsibx and
608 stxsihx instructions. Fix the modsw and modsd instructions.
610 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
611 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
612 conversion instructions are still missing.
614 * ==================== TOOL CHANGES ====================
616 * General tool changes
618 - All the tools and their vgpreload libraries are now installed under
619 libexec because they cannot be executed directly and should be run through
620 the valgrind executable. This should be an internal, not user visible,
621 change, but might impact valgrind packagers.
623 - The --track-fds option now respects -q, --quiet and won't output anything
624 if no file descriptors are leaked. It also won't report the standard stdin
625 (0), stdout (1) or stderr (2) descriptors as being leaked with
626 --trace-fds=yes anymore. To track whether the standard file descriptors
627 are still open at the end of the program run use --trace-fds=all.
631 - DHAT has been extended, with two new modes of operation. The new
632 --mode=copy flag triggers copy profiling, which records calls to memcpy,
633 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
634 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
635 the new dhat/dhat.h file. This is useful for learning more about hot code
636 paths. See the user manual for more information about the new modes.
638 - Because of these changes, DHAT's file format has changed. DHAT output
639 files produced with earlier versions of DHAT will not work with this
640 version of DHAT's viewer, and DHAT output files produced with this version
641 of DHAT will not work with earlier versions of DHAT's viewer.
643 * ==================== FIXED BUGS ====================
645 The following bugs have been fixed or resolved. Note that "n-i-bz"
646 stands for "not in bugzilla" -- that is, a bug that was reported to us
647 but never got a bugzilla entry. We encourage you to file bugs in
648 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
649 than mailing the developers (or mailing lists) directly -- bugs that
650 are not entered into bugzilla tend to get forgotten about or ignored.
652 To see details of a given bug, visit
653 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
654 where XXXXXX is the bug number as listed below.
656 140178 open("/proc/self/exe", ...); doesn't quite work
657 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
658 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
659 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
660 345077 linux syscall execveat support (linux 3.19)
661 361770 Missing F_ADD_SEALS
662 369029 handle linux syscalls sched_getattr and sched_setattr
663 384729 __libc_freeres inhibits cross-platform valgrind
664 388787 Support for C++17 new/delete
665 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
666 396656 Warnings while reading debug info
667 397605 ioctl FICLONE mishandled
668 401416 Compile failure with openmpi 4.0
669 408663 Suppression file for musl libc
670 404076 s390x: z14 vector instructions not implemented
671 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
672 (actually affected all x86 and nanomips regardless of host bitness)
673 413547 regression test does not check for Arm 64 features.
674 414268 Enable AArch64 feature detection and decoding for v8.x instructions
675 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
676 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
677 422261 platform selection fails for unqualified client name
678 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
679 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
680 423195 PPC ISA 3.1 support is missing, part 1
681 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
682 424012 crash with readv/writev having invalid but not NULL arg2 iovec
683 424298 amd64: Implement RDSEED
684 425232 PPC ISA 3.1 support is missing, part 2
685 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
686 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
687 426123 PPC ISA 3.1 support is missing, part 3
688 426144 Fix "condition variable has not been initialized" on Fedora 33.
689 427400 PPC ISA 3.1 support is missing, part 4
690 427401 PPC ISA 3.1 support is missing, part 5
691 427404 PPC ISA 3.1 support is missing, part 6
692 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
693 427787 Support new faccessat2 linux syscall (439)
694 427969 debuginfo section duplicates a section in the main ELF file
695 428035 drd: Unbreak the musl build
696 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
697 428716 cppcheck detects potential leak in VEX/useful/smchash.c
698 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
699 429352 PPC ISA 3.1 support is missing, part 7
700 429354 PPC ISA 3.1 support is missing, part 8
701 429692 unhandled ppc64le-linux syscall: 147 (getsid)
702 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
704 429952 Errors when building regtest with clang
705 430354 ppc stxsibx and stxsihx instructions write too much data
706 430429 valgrind.h doesn't compile on s390x with clang
707 430485 expr_is_guardable doesn't handle Iex_Qop
708 431556 Complete arm64 FADDP v8.2 instruction support
709 432102 Add support for DWARF5 as produced by GCC11
710 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
711 432381 drd: Process STACK_REGISTER client requests
712 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
713 432672 vg_regtest: test-specific environment variables not reset between tests
714 432809 VEX should support REX.W + POPF
715 432861 PPC modsw and modsd give incorrect results for 1 mod 12
716 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
717 432215 Add debuginfod functionality
718 433323 Use pkglibexecdir as vglibdir
719 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
720 433629 valgrind/README has type "abd" instead of "and"
721 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
722 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
723 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
724 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
725 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
727 (3.17.0.RC1: 13 Mar 2021)
728 (3.17.0.RC2: 17 Mar 2021)
729 (3.17.0: 19 Mar 2021)
733 Release 3.16.1 (22 June 2020)
734 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
736 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
737 fixes character encoding problems in the documentation HTML.
739 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
740 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
742 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
746 Release 3.16.0 (27 May 2020)
747 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
749 3.16.0 is a feature release with many improvements and the usual collection of
752 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
753 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
754 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
755 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
756 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
758 * ==================== CORE CHANGES ===================
760 * It is now possible to dynamically change the value of many command line
761 options while your program (or its children) are running under Valgrind.
763 To see the list of dynamically changeable options, run
764 "valgrind --help-dyn-options".
766 You can change the options from the shell by using vgdb to launch
767 the monitor command "v.clo <clo option>...".
768 The same monitor command can be used from a gdb connected
769 to the valgrind gdbserver.
770 Your program can also change the dynamically changeable options using
771 the client request VALGRIND_CLO_CHANGE(option).
773 * ================== PLATFORM CHANGES =================
775 * MIPS: preliminary support for nanoMIPS instruction set has been added.
777 * ==================== TOOL CHANGES ====================
781 - The implicit memcpy done by each call to realloc now counts towards the
782 read and write counts of resized heap blocks, making those counts higher
787 - cg_annotate's --auto and --show-percs options now default to 'yes', because
788 they are usually wanted.
792 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
793 because they are usually wanted.
795 - The command option --collect-systime has been enhanced to specify
796 the unit used to record the elapsed time spent during system calls.
797 The command option now accepts the values no|yes|msec|usec|nsec,
798 where yes is a synonym of msec. When giving the value nsec, the
799 system cpu time of system calls is also recorded.
803 - Several memcheck options are now dynamically changeable.
804 Use valgrind --help-dyn-options to list them.
806 - The release 3.15 introduced a backward incompatible change for
807 some suppression entries related to preadv and pwritev syscalls.
808 When reading a suppression entry using the unsupported 3.14 format,
809 valgrind will now produce a warning to say the suppression entry will not
810 work, and suggest the needed change.
812 - Significantly fewer false positive errors on optimised code generated by
813 Clang and GCC. In particular, Memcheck now deals better with the
814 situation where the compiler will transform C-level "A && B" into "B && A"
815 under certain circumstances (in which the transformation is valid).
816 Handling of integer equality/non-equality checks on partially defined
817 values is also improved on some architectures.
821 - The exprimental Stack and Global Array Checking tool has been removed.
822 It only ever worked on x86 and amd64, and even on those it had a
823 high false positive rate and was slow. An alternative for detecting
824 stack and global array overruns is using the AddressSanitizer (ASAN)
825 facility of the GCC and Clang compilers, which require you to rebuild
826 your code with -fsanitize=address.
828 * ==================== OTHER CHANGES ====================
830 * New and modified GDB server monitor features:
832 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
834 - The gdbserver monitor commands that require an address and an optional
835 length argument now accepts the alternate 'C like' syntax "address[length]".
836 For example, the memcheck command "monitor who_points_at 0x12345678 120"
837 can now also be given as "monitor who_points_at 0x12345678[120]".
839 * ==================== FIXED BUGS ====================
841 The following bugs have been fixed or resolved. Note that "n-i-bz"
842 stands for "not in bugzilla" -- that is, a bug that was reported to us
843 but never got a bugzilla entry. We encourage you to file bugs in
844 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
845 than mailing the developers (or mailing lists) directly -- bugs that
846 are not entered into bugzilla tend to get forgotten about or ignored.
848 To see details of a given bug, visit
849 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
850 where XXXXXX is the bug number as listed below.
852 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
853 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
854 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
855 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
856 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
857 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
858 400872 Add nanoMIPS support to Valgrind
859 403212 drd/tests/trylock hangs on FreeBSD
860 404406 s390x: z14 miscellaneous instructions not implemented
861 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
862 406561 mcinfcallWSRU gdbserver_test fails on ppc64
863 406824 Unsupported baseline
864 407218 Add support for the copy_file_range syscall
865 407307 Intercept stpcpy also in ld.so for arm64
866 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
868 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
869 408009 Expose rdrand and f16c even on avx if host cpu supports them
870 408091 Missing pkey syscalls
871 408414 Add support for missing for preadv2 and pwritev2 syscalls
872 409141 Valgrind hangs when SIGKILLed
873 409206 Support for Linux PPS and PTP ioctls
874 409367 exit_group() after signal to thread waiting in futex() causes hangs
875 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
876 409780 References to non-existent configure.in
877 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
878 410599 Non-deterministic behaviour of pth_self_kill_15_other test
879 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
880 411134 Allow the user to change a set of command line options during execution
881 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
882 412344 Problem setting mips flags with specific paths
883 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
884 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
885 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
886 413603 callgrind_annotate/cg_annotate truncate function names at '#'
887 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
888 415136 ARMv8.1 Compare-and-Swap instructions are not supported
889 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
890 416239 valgrind crashes when handling clock_adjtime
891 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
892 416286 DRD reports "conflicting load" error on std::mutex::lock()
893 416301 s390x: "compare and signal" not supported
894 416387 finit_module and bpf syscalls are unhandled on arm64
895 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
896 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
897 416753 new 32bit time syscalls for 2038+
898 417075 pwritev(vector[...]) suppression ignored
899 417075 is not fixed, but incompatible supp entries are detected
900 and a warning is produced for these.
901 417187 [MIPS] Conditional branch problem since 'grail' changes
902 417238 Test memcheck/tests/vbit-test fails on mips64 BE
903 417266 Make memcheck/tests/linux/sigqueue usable with musl
904 417281 s390x: /bin/true segfaults with "grail" enabled
905 417427 commit to fix vki_siginfo_t definition created numerous regression
907 417452 s390_insn_store_emit: dst->tag for HRcVec128
908 417578 Add suppressions for glibc DTV leaks
909 417906 clone with CLONE_VFORK and no CLONE_VM fails
910 418004 Grail code additions break ppc64.
911 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
912 418997 s390x: Support Iex_ITE for float and vector types
913 419503 s390x: Avoid modifying registers returned from isel functions
914 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
915 421570 std_mutex fails on Arm v8.1 h/w
916 434035 vgdb might crash if valgrind is killed
917 n-i-bz Fix minor one time leaks in dhat.
918 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
919 n-i-bz Add support for the Linux io_uring system calls
920 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
921 n-i-bz Fix non-glibc build of test suite with s390x_features
922 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
923 423195 PPC ISA 3.1 support is missing, part 1
925 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
926 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
927 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
931 Release 3.15.0 (12 April 2019)
932 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
934 3.15.0 is a feature release with many improvements and the usual collection of
937 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
938 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
939 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
940 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
941 support for X86/macOS 10.13 and AMD64/macOS 10.13.
943 * ==================== CORE CHANGES ===================
945 * The XTree Massif output format now makes use of the information obtained
946 when specifying --read-inline-info=yes.
948 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
950 * ==================== TOOL CHANGES ====================
954 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
955 it has been promoted from an experimental tool to a regular tool. Run it
956 with --tool=dhat instead of --tool=exp-dhat.
958 - DHAT now prints only minimal data when the program ends, instead writing
959 the bulk of the profiling data to a file. As a result, the --show-top-n
960 and --sort-by options have been removed.
962 - Profile results can be viewed with the new viewer, dh_view.html. When
963 a run ends, a short message is printed, explaining how to view the result.
965 - See the documentation for more details.
969 - cg_annotate has a new option, --show-percs, which prints percentages next
974 - callgrind_annotate has a new option, --show-percs, which prints percentages
975 next to all event counts.
977 - callgrind_annotate now inserts commas in call counts, and
978 sort the caller/callee lists in the call tree.
982 - The default value for --read-inline-info is now "yes" on
983 Linux/Android/Solaris. It is still "no" on other OS.
987 - The option --xtree-leak=yes (to output leak result in xtree format)
988 automatically activates the option --show-leak-kinds=all, as xtree
989 visualisation tools such as kcachegrind can in any case select what kind
990 of leak to visualise.
992 - There has been further work to avoid false positives. In particular,
993 integer equality on partially defined inputs (C == and !=) is now handled
996 * ==================== OTHER CHANGES ====================
998 * The new option --show-error-list=no|yes displays, at the end of the run, the
999 list of detected errors and the used suppressions. Prior to this change,
1000 showing this information could only be done by specifying "-v -v", but that
1001 also produced a lot of other possibly-non-useful messages. The option -s is
1002 equivalent to --show-error-list=yes.
1004 * ==================== FIXED BUGS ====================
1006 The following bugs have been fixed or resolved. Note that "n-i-bz"
1007 stands for "not in bugzilla" -- that is, a bug that was reported to us
1008 but never got a bugzilla entry. We encourage you to file bugs in
1009 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1010 than mailing the developers (or mailing lists) directly -- bugs that
1011 are not entered into bugzilla tend to get forgotten about or ignored.
1013 To see details of a given bug, visit
1014 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1015 where XXXXXX is the bug number as listed below.
1017 385411 s390x: z13 vector floating-point instructions not implemented
1018 397187 z13 vector register support for vgdb gdbserver
1019 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
1020 398870 Please add support for instruction vcvtps2ph
1021 399287 amd64 front end: Illegal Instruction vcmptrueps
1022 399301 Use inlined frames in Massif XTree output.
1023 399322 Improve callgrind_annotate output
1024 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
1025 400164 helgrind test encounters mips x-compiler warnings and assembler error
1026 400490 s390x: VRs allocated as if separate from FPRs
1027 400491 s390x: Operand of LOCH treated as unsigned integer
1028 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
1029 options, which specify a mips64 processor
1030 401112 LLVM 5.0 generates comparison against partially initialized data
1031 401277 More bugs in z13 support
1032 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
1033 401578 drd: crashes sometimes on fork()
1034 401627 memcheck errors with glibc avx2 optimized wcsncmp
1035 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
1036 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
1037 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
1039 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
1040 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
1041 402123 invalid assembler opcodes for mips32r2
1042 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
1043 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
1044 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
1045 402351 mips64 libvexmultiarch_test fails on s390x
1046 402369 Overhaul DHAT
1047 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
1048 402480 Do not use %rsp in clobber list
1049 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
1050 402515 Implement new option --show-error-list=no|yes / -s
1051 402519 POWER 3.0 addex instruction incorrectly implemented
1052 402781 Redo the cache used to process indirect branch targets
1053 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
1054 403552 s390x: wrong facility bit checked for vector facility
1055 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
1056 404638 Add VG_(replaceIndexXA)
1057 404843 s390x: backtrace sometimes ends prematurely
1058 404888 autotools cleanup series
1059 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
1060 405182 Valgrind fails to build with Clang
1061 405205 filter_libc: remove the line holding the futex syscall error entirely
1062 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
1063 the upper and lower 32-bits of the 64-bit result
1064 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
1065 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
1066 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
1068 405403 s390x disassembler cannot be used on x86
1069 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
1070 405458 MIPS mkFormVEC arguments swapped?
1071 405716 drd: Fix an integer overflow in the stack margin calculation
1072 405722 Support arm64 core dump
1073 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
1074 of the 64-bit destination field.
1075 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
1076 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
1077 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
1079 406256 PPC64, vector floating point instructions don't handle subnormal
1080 according to VSCR[NJ] bit setting.
1081 406352 cachegrind/callgrind fails ann tests because of missing a.c
1082 406354 dhat is broken on x86 (32bit)
1083 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
1084 406357 gdbserver_tests fails because of gdb output change
1085 406360 memcheck/tests/libstdc++.supp needs more supression variants
1086 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
1087 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
1088 407340 PPC64, does not support the vlogefp, vexptefp instructions.
1089 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
1090 n-i-bz Fix callgrind_annotate non deterministic order for equal total
1091 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
1092 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
1093 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
1095 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
1096 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
1097 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
1101 Release 3.14.0 (9 October 2018)
1102 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1104 3.14.0 is a feature release with many improvements and the usual collection of
1107 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1108 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1109 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1110 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
1111 support for X86/macOS 10.13, AMD64/macOS 10.13.
1113 * ==================== CORE CHANGES ===================
1115 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
1116 debug info for unloaded code. This allows saved stack traces (e.g. for
1117 memory leaks) to include file/line info for code that has been dlclose'd (or
1118 similar). See the user manual for more information and known limitations.
1120 * Ability to specify suppressions based on source file name and line number.
1122 * Majorly overhauled register allocator. No end-user changes, but the JIT
1123 generates code a bit more quickly now.
1125 * ================== PLATFORM CHANGES =================
1127 * Preliminary support for macOS 10.13 has been added.
1129 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
1131 * mips: support for MIPS SIMD architecture (MSA) has been added.
1133 * mips: support for MIPS N32 ABI has been added.
1135 * s390: partial support for vector instructions (integer and string) has been
1138 * ==================== TOOL CHANGES ====================
1140 * Helgrind: Addition of a flag
1141 --delta-stacktrace=no|yes [yes on linux amd64/x86]
1142 which specifies how full history stack traces should be computed.
1143 Setting this to =yes can speed up Helgrind by 25% when using
1144 --history-level=full.
1146 * Memcheck: reduced false positive rate for optimised code created by Clang 6
1147 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
1148 blocks more carefully to determine where it can avoid expensive definedness
1149 checks without loss of precision. This is controlled by the flag
1150 --expensive-definedness-checks=no|auto|yes [auto].
1152 * ==================== OTHER CHANGES ====================
1154 * Valgrind is now buildable with link-time optimisation (LTO). A new
1155 configure option --enable-lto=yes allows building Valgrind with LTO. If the
1156 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
1157 Note that if you are doing Valgrind development, --enable-lto=yes massively
1158 slows down the build process.
1160 * ==================== FIXED BUGS ====================
1162 The following bugs have been fixed or resolved. Note that "n-i-bz"
1163 stands for "not in bugzilla" -- that is, a bug that was reported to us
1164 but never got a bugzilla entry. We encourage you to file bugs in
1165 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1166 than mailing the developers (or mailing lists) directly -- bugs that
1167 are not entered into bugzilla tend to get forgotten about or ignored.
1169 To see details of a given bug, visit
1170 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1171 where XXXXXX is the bug number as listed below.
1173 79362 Debug info is lost for .so files when they are dlclose'd
1174 208052 strlcpy error when n = 0
1175 255603 exp-sgcheck Assertion '!already_present' failed
1176 338252 building valgrind with -flto (link time optimisation) fails
1177 345763 MIPS N32 ABI support
1178 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
1179 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
1180 372347 Replacement problem of the additional c++14/c++17 new/delete operators
1181 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1182 376257 helgrind history full speed up using a cached stack
1183 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
1185 379748 Fix missing pselect syscall (OS X 10.11)
1186 379754 Fix missing syscall ulock_wait (OS X 10.12)
1187 380397 s390x: __GI_strcspn() replacemenet needed
1188 381162 possible array overrun in VEX register allocator
1189 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
1190 381274 powerpc too chatty even with --sigill-diagnostics=no
1191 381289 epoll_pwait can have a NULL sigmask
1192 381553 VEX register allocator v3
1193 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
1194 381769 Use ucontext_t instead of struct ucontext
1195 381805 arm32 needs ld.so index hardwire for new glibc security fixes
1196 382256 gz compiler flag test doesn't work for gold
1197 382407 vg_perf needs "--terse" command line option
1198 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
1199 382563 MIPS MSA ASE support
1200 382998 xml-socket doesn't work
1201 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
1202 383723 Fix missing kevent_qos syscall (macOS 10.11)
1203 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
1204 384096 Mention AddrCheck at Memcheck's command line option [..]
1205 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
1206 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
1207 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
1208 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
1209 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
1210 384337 performance improvements to VEX register allocator v2 and v3
1211 384526 reduce number of spill insns generated by VEX register allocator v3
1212 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
1213 384631 Sanitise client args as printed with -v
1214 384633 Add a simple progress-reporting facility
1215 384987 VEX regalloc: allocate caller-save registers for short lived vregs
1216 385055 PPC VEX temporary storage exhausted
1217 385182 PPC64 is missing support for the DSCR
1218 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
1219 385207 PPC64, generate_store_FPRF() generates too many Iops
1220 385208 PPC64, xxperm instruction exhausts temporary memory
1221 385210 PPC64, vpermr instruction could exhaust temporary memory
1222 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
1223 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
1224 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
1225 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
1226 385408 s390x: z13 vector "support" instructions not implemented
1227 385409 s390x: z13 vector integer instructions not implemented
1228 385410 s390x: z13 vector string instructions not implemented
1229 385412 s390x: new non-vector z13 instructions not implemented
1230 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
1231 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
1232 385939 Optionally exit on the first error
1233 386318 valgrind.org/info/tools.html is missing SGCheck
1234 386425 running valgrind + wine on armv7l gives illegal opcode
1235 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
1236 387410 MIPSr6 support
1237 387664 Memcheck: make expensive-definedness-checks be the default
1238 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
1239 387766 asm shifts cause false positive "Conditional jump or move depends
1240 on uninitialised value"
1241 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
1242 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
1243 388786 Support bpf syscall in amd64 Linux
1244 388862 Add replacements for wmemchr and wcsnlen on Linux
1245 389065 valgrind meets gcc flag -Wlogical-op
1246 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
1247 390471 suppression by specification of source-file line number
1248 390723 make xtree dump files world wide readable, similar to log files
1249 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
1250 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
1251 392118 unhandled amd64-linux syscall: 332 (statx)
1252 392449 callgrind not clearing the number of calls properly
1253 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
1254 lxssp, stxssp and stxvl instructions.
1255 393023 callgrind_control risks using the wrong vgdb
1256 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
1257 393099 posix_memalign() invalid write if alignment == 0
1258 393146 failing assert "is_DebugInfo_active(di)"
1259 395709 PPC64 is missing support for the xvnegsp instruction
1260 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
1262 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
1263 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
1264 396839 s390x: Trap instructions not implemented
1265 396887 arch_prctl should return EINVAL on unknown option
1266 == 397286 crash before launching binary (Unsupported arch_prctl option)
1267 == 397393 valgrind: the 'impossible' happened: (Archlinux)
1268 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
1269 396906 compile tests failure on mips32-linux: broken inline asm in tests on
1271 397012 glibc ld.so uses arch_prctl on i386
1272 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
1273 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
1274 397424 glibc 2.27 and gdb_server tests
1275 398028 Assertion `cfsi_fits` failing in simple C program
1276 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
1278 n-i-bz Fix missing workq_ops operations (macOS)
1279 n-i-bz fix bug in strspn replacement
1280 n-i-bz Add support for the Linux BLKFLSBUF ioctl
1281 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
1282 n-i-bz Fix possible stack trashing by semctl syscall wrapping
1283 n-i-bz Add support for the Linux membarrier() system call
1284 n-i-bz x86 front end: recognise and handle UD2 correctly
1285 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
1286 correctly aligned before entering the handler.
1288 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
1289 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
1290 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
1294 Release 3.13.0 (15 June 2017)
1295 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1297 3.13.0 is a feature release with many improvements and the usual collection of
1300 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1301 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1302 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1303 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
1305 * ==================== CORE CHANGES ===================
1307 * The translation cache size has been increased to keep up with the demands of
1308 large applications. The maximum number of sectors has increased from 24 to
1309 48. The default number of sectors has increased from 16 to 32 on all
1310 targets except Android, where the increase is from 6 to 12.
1312 * The amount of memory that Valgrind can use has been increased from 64GB to
1313 128GB. In particular this means your application can allocate up to about
1314 60GB when running on Memcheck.
1316 * Valgrind's default load address has been changed from 0x3800'0000 to
1317 0x5800'0000, so as to make it possible to load larger executables. This
1318 should make it possible to load executables of size at least 1200MB.
1320 * A massive spaceleak caused by reading compressed debuginfo files has been
1321 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
1324 * The C++ demangler has been updated.
1326 * Support for demangling Rust symbols has been added.
1328 * A new representation of stack traces, the "XTree", has been added. An XTree
1329 is a tree of stacktraces with data associated with the stacktraces. This is
1330 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
1331 consumption of your program. Reporting is controlled by the new options
1332 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
1334 A report can also be produced on demand using the gdbserver monitor command
1335 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
1336 format' and 'massif format. The existing visualisers for these formats (e.g.
1337 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
1338 analyse these reports.
1340 Memcheck can also produce XTree leak reports using the Callgrind file
1341 format. For more details, see the user manual.
1343 * ================== PLATFORM CHANGES =================
1345 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
1347 * amd64: fixes for JIT failure problems on long AVX2 code blocks
1349 * amd64 and x86: support for CET prefixes has been added
1351 * arm32: a few missing ARMv8 instructions have been implemented
1353 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
1354 Store-Conditional instructions has been added. This is to deal with
1355 processor implementations that implement the LL/SC specifications strictly
1356 and as a result cause Valgrind to hang in certain situations. The
1357 alternative implementation is automatically enabled at startup, as required.
1358 You can use the option --sim-hints=fallback-llsc to force-enable it if you
1361 * Support for OSX 10.12 has been improved.
1363 * On Linux, clone handling has been improved to honour CLONE_VFORK that
1364 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
1365 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
1366 CLONE_VM exact semantics will (still) not work.
1368 * The TileGX/Linux port has been removed because it appears to be both unused
1371 * ==================== TOOL CHANGES ====================
1375 - Memcheck should give fewer false positives when running optimised
1376 Clang/LLVM generated code.
1378 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1380 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
1381 to produce the end of execution leak report in a xtree callgrind format
1384 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
1385 the leak report in an xtree file.
1389 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1391 - For some workloads (typically, for big applications), Massif memory
1392 consumption and CPU consumption has decreased significantly.
1396 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1398 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
1399 for Ada gnat compiled applications.
1401 * ==================== OTHER CHANGES ====================
1403 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
1404 append the inner guest stacktrace to the inner host stacktrace. This helps
1405 to investigate the errors reported by the outer, when they are caused by the
1406 inner guest program (such as an inner regtest). See README_DEVELOPERS for
1409 * To allow fast detection of callgrind files by desktop environments and file
1410 managers, the format was extended to have an optional first line that
1411 uniquely identifies the format ("# callgrind format"). Callgrind creates
1412 this line now, as does the new xtree functionality.
1414 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
1415 have a new %n format letter that is replaced by a sequence number.
1417 * "--version -v" now shows the SVN revision numbers from which Valgrind was
1420 * ==================== FIXED BUGS ====================
1422 The following bugs have been fixed or resolved. Note that "n-i-bz"
1423 stands for "not in bugzilla" -- that is, a bug that was reported to us
1424 but never got a bugzilla entry. We encourage you to file bugs in
1425 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1426 than mailing the developers (or mailing lists) directly -- bugs that
1427 are not entered into bugzilla tend to get forgotten about or ignored.
1429 To see details of a given bug, visit
1430 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1431 where XXXXXX is the bug number as listed below.
1433 162848 --log-file output isn't split when a program forks
1434 340777 Illegal instruction on mips (ar71xx)
1435 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
1436 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
1437 to a different stack.
1438 344139 x86 stack-seg overrides, needed by the Wine people
1439 344524 store conditional of guest applications always fail - observed on
1441 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
1442 352395 Please provide SVN revision info in --version -v
1443 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
1444 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
1445 INVALID_POSIX_THREADID' failed
1446 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
1447 barrier implementation
1448 358697 valgrind.h: Some code remains even when defining NVALGRIND
1449 359202 Add musl libc configure/compile
1450 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
1451 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
1452 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
1453 362223 assertion failed when .valgrindrc is a directory instead of a file
1454 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
1455 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
1456 368507 can't malloc chunks larger than about 34GB
1457 368529 Android arm target link error, missing atexit and pthread_atfork
1458 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
1459 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
1460 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
1461 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
1462 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
1463 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
1464 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
1465 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
1466 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
1467 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
1468 370028 Reduce the number of compiler warnings on MIPS platforms
1469 370635 arm64 missing syscall getcpu
1470 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
1471 371227 Clean AArch64 syscall table
1472 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
1473 371471 Valgrind complains about non legit memory leaks on placement new (C++)
1474 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
1475 371503 disInstr(arm64): unhandled instruction 0xF89F0000
1476 371869 support '%' in symbol Z-encoding
1477 371916 execution tree xtree concept
1478 372120 c++ demangler demangles symbols which are not c++
1479 372185 Support of valgrind on ARMv8 with 32 bit executable
1480 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1481 372195 Power PC, xxsel instruction is not always recognized.
1482 372504 Hanging on exit_group
1483 372600 process loops forever when fatal signals are arriving quickly
1484 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1485 373046 Stacks registered by core are never deregistered
1486 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1487 373086 Implement additional Xen hypercalls
1488 373192 Calling posix_spawn in glibc 2.24 completely broken
1489 373488 Support for fanotify API on ARM64 architecture
1490 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1491 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1492 373938 const IRExpr arguments for matchIRExpr()
1493 374719 some spelling fixes
1494 374963 increase valgrind's load address to prevent mmap failure
1495 375514 valgrind_get_tls_addr() does not work in case of static TLS
1496 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1497 for ML_(find_rx_mapping)()
1498 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1499 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1500 == 377159 "vex: the `impossible' happened" still present
1501 == 375150 Assertion 'tres.status == VexTransOK' failed
1502 == 378068 valgrind crashes on AVX2 function in FFmpeg
1503 376142 Segfaults on MIPS Cavium Octeon boards
1504 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1505 376455 Solaris: unhandled syscall lgrpsys(180)
1506 376518 Solaris: unhandled fast trap getlgrp(6)
1507 376611 ppc64 and arm64 don't know about prlimit64 syscall
1508 376729 PPC64, remove R2 from the clobber list
1510 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1511 to be wrongly marked as addressable
1512 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1513 PIE enabled by default
1514 377376 memcheck/tests/linux/getregset fails with glibc2.24
1515 377427 PPC64, lxv instruction failing on odd destination register
1516 377478 PPC64: ISA 3.0 setup fixes
1517 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1518 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1519 and 2 args for FUTEX_TRYLOCK_PI
1520 377717 Fix massive space leak when reading compressed debuginfo sections
1521 377891 Update Xen 4.6 domctl wrappers
1522 377930 fcntl syscall wrapper is missing flock structure check
1523 378524 libvexmultiarch_test regression on s390x and ppc64
1524 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1525 378673 Update libiberty demangler
1526 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1527 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1528 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1529 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1530 (task_register_dyld_image_infos)
1531 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1532 (task_register_dyld_shared_cache_image_info)
1533 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1534 379473 MIPS: add support for rdhwr cycle counter register
1535 379504 remove TileGX/Linux port
1536 379525 Support more x86 nop opcodes
1537 379838 disAMode(x86): not an addr!
1538 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1539 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1540 379895 clock_gettime does not execute POST syscall wrapper
1541 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1542 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1543 380200 xtree generated callgrind files refer to files without directory name
1544 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1545 380397 s390x: __GI_strcspn() replacement needed
1546 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1548 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1549 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1550 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1554 Release 3.12.0 (20 October 2016)
1555 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1557 3.12.0 is a feature release with many improvements and the usual
1558 collection of bug fixes.
1560 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1561 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1562 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1563 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1564 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1565 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1567 * ================== PLATFORM CHANGES =================
1569 * POWER: Support for ISA 3.0 has been added
1571 * mips: support for O32 FPXX ABI has been added.
1572 * mips: improved recognition of different processors
1573 * mips: determination of page size now done at run time
1575 * amd64: Partial support for AMD FMA4 instructions.
1577 * arm, arm64: Support for v8 crypto and CRC instructions.
1579 * Improvements and robustification of the Solaris port.
1581 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1583 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1584 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1585 where possible. Valgrind's support for 32-bit x86 has stagnated in
1586 recent years and has fallen far behind that for 64-bit x86
1587 instructions. By contrast 64-bit x86 is well supported, up to and
1590 * ==================== TOOL CHANGES ====================
1594 - Added meta mempool support for describing a custom allocator which:
1595 - Auto-frees all chunks assuming that destroying a pool destroys all
1597 - Uses itself to allocate other memory blocks
1599 - New flag --ignore-range-below-sp to ignore memory accesses below
1600 the stack pointer, if you really have to. The related flag
1601 --workaround-gcc296-bugs=yes is now deprecated. Use
1602 --ignore-range-below-sp=1024-1 as a replacement.
1606 - Improved thread startup time significantly on non-Linux platforms.
1610 - Added collection of the metric "tot-blocks-allocd"
1612 * ==================== OTHER CHANGES ====================
1614 * Replacement/wrapping of malloc/new related functions is now done not just
1615 for system libraries by default, but for any globally defined malloc/new
1616 related function (both in shared libraries and statically linked alternative
1617 malloc implementations). The dynamic (runtime) linker is excluded, though.
1618 To only intercept malloc/new related functions in
1619 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1620 "nouserintercepts" can be any non-existing library name).
1621 This new functionality is not implemented for MacOS X.
1623 * The maximum number of callers in a suppression entry is now equal to
1624 the maximum size for --num-callers (500).
1625 Note that --gen-suppressions=yes|all similarly generates suppressions
1626 containing up to --num-callers frames.
1628 * New and modified GDB server monitor features:
1630 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1631 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1634 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1635 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1638 * Valgrind is able to read compressed debuginfo sections in two formats:
1639 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1640 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1642 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1643 for the most common use case (x86_64-linux, Memcheck) has been
1646 * Improved performance for programs that do a lot of discarding of
1647 instruction address ranges of 8KB or less.
1649 * The C++ symbol demangler has been updated.
1651 * More robustness against invalid syscall parameters on Linux.
1653 * ==================== FIXED BUGS ====================
1655 The following bugs have been fixed or resolved. Note that "n-i-bz"
1656 stands for "not in bugzilla" -- that is, a bug that was reported to us
1657 but never got a bugzilla entry. We encourage you to file bugs in
1658 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1659 than mailing the developers (or mailing lists) directly -- bugs that
1660 are not entered into bugzilla tend to get forgotten about or ignored.
1662 To see details of a given bug, visit
1663 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1664 where XXXXXX is the bug number as listed below.
1666 191069 Exiting due to signal not reported in XML output
1667 199468 Suppressions: stack size limited to 25
1668 while --num-callers allows more frames
1669 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1670 278744 cvtps2pd with redundant RexW
1671 303877 valgrind doesn't support compressed debuginfo sections.
1672 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1673 348345 Assertion fails for negative lineno
1674 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1675 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1676 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1677 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1678 352197 mips: mmap2() not wrapped correctly for page size > 4096
1679 353083 arm64 doesn't implement various xattr system calls
1680 353084 arm64 doesn't support sigpending system call
1681 353137 www: update info for Supported Platforms
1682 353138 www: update "The Valgrind Developers" page
1683 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1686 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1687 353398 WARNING: unhandled amd64-solaris syscall: 207
1688 353660 XML in auxwhat tag not escaping reserved symbols properly
1689 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1690 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1691 353802 ELF debug info reader confused with multiple .rodata sections
1692 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1693 353917 unhandled amd64-solaris syscall fchdir(120)
1694 353920 unhandled amd64-solaris syscall: 170
1695 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1696 354392 unhandled amd64-solaris syscall: 171
1697 354797 Vbit test does not include Iops for Power 8 instruction support
1698 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1702 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1703 355188 valgrind should intercept all malloc related global functions
1704 355454 do not intercept malloc related symbols from the runtime linker
1705 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1706 356044 Dwarf line info reader misinterprets is_stmt register
1707 356112 mips: replace addi with addiu
1708 356393 valgrind (vex) crashes because isZeroU happened
1711 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1712 356678 arm64-linux: unhandled syscall 232 (mincore)
1713 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1714 356823 Unsupported ARM instruction: stlex
1715 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1716 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1717 357673 crash if I try to run valgrind with a binary link with libcurl
1718 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1719 357871 pthread_spin_destroy not properly wrapped
1720 357887 Calls to VG_(fclose) do not close the file descriptor
1721 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1722 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1723 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1724 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1725 359181 Buffer Overflow during Demangling
1726 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1727 359289 s390x: popcnt (B9E1) not implemented
1728 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1729 359503 Add missing syscalls for aarch64 (arm64)
1730 359645 "You need libc6-dbg" help message could be more helpful
1731 359703 s390: wire up separate socketcalls system calls
1732 359724 getsockname might crash - deref_UInt should call safe_to_deref
1733 359733 amd64 implement ld.so strchr/index override like x86
1734 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1735 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1737 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1738 359871 Incorrect mask handling in ppoll
1739 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1740 360008 Contents of Power vr registers contents is not printed correctly when
1741 the --vgdb-shadow-registers=yes option is used
1742 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1743 non-zero shadow bits
1744 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1745 360425 arm64 unsupported instruction ldpsw
1747 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1748 360571 Error about the Android Runtime reading below the stack pointer on ARM
1749 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1750 360749 kludge for multiple .rodata sections on Solaris no longer needed
1751 360752 raise the number of reserved fds in m_main.c from 10 to 12
1752 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1753 361226 s390x: risbgn (EC59) not implemented
1754 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1755 361354 ppc64[le]: wire up separate socketcalls system calls
1756 361615 Inconsistent termination for multithreaded process terminated by signal
1757 361926 Unhandled Solaris syscall: sysfs(84)
1758 362009 V dumps core on unimplemented functionality before threads are created
1759 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1760 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1761 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1762 362953 Request for an update to the Valgrind Developers page
1763 363680 add renameat2() support
1764 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1765 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1766 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1767 364058 clarify in manual limitations of array overruns detections
1768 364413 pselect sycallwrapper mishandles NULL sigmask
1769 364728 Power PC, missing support for several HW registers in
1770 get_otrack_shadow_offset_wrk()
1771 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1772 365273 Invalid write to stack location reported after signal handler runs
1773 365912 ppc64BE segfault during jm-insns test (RELRO)
1774 366079 FPXX Support for MIPS32 Valgrind
1775 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1776 366344 Multiple unhandled instruction for Aarch64
1777 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1778 367995 Integration of memcheck with custom memory allocator
1779 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1780 368412 False positive result for altivec capability check
1781 368416 Add tc06_two_races_xml.exp output for ppc64
1782 368419 Perf Events ioctls not implemented
1783 368461 mmapunmap test fails on ppc64
1784 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1785 369000 AMD64 fma4 instructions unsupported.
1786 369169 ppc64 fails jm_int_isa_2_07 test
1787 369175 jm_vec_isa_2_07 test crashes on ppc64
1788 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1789 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1790 369359 msghdr_foreachfield can crash when handling bad iovec
1791 369360 Bad sigprocmask old or new sets can crash valgrind
1792 369361 vmsplice syscall wrapper crashes on bad iovec
1793 369362 Bad sigaction arguments crash valgrind
1794 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1795 369402 Bad set/get_thread_area pointer crashes valgrind
1796 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1797 369446 valgrind crashes on unknown fcntl command
1798 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1799 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1800 370265 ISA 3.0 HW cap stuff needs updating
1801 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1802 372195 Power PC, xxsel instruction is not always recognized
1804 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1805 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1806 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1807 n-i-bz Fix ppoll_alarm exclusion on OS X
1808 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1809 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1810 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1811 n-i-bz amd64: memcheck false positive with shr %edx
1812 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1813 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1814 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1815 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1817 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1818 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1819 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1823 Release 3.11.0 (22 September 2015)
1824 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1826 3.11.0 is a feature release with many improvements and the usual
1827 collection of bug fixes.
1829 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1830 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1831 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1832 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1833 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1834 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1836 * ================== PLATFORM CHANGES =================
1838 * Support for Solaris/x86 and Solaris/amd64 has been added.
1840 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1842 * Preliminary support for the Tilera TileGX architecture has been added.
1844 * s390x: It is now required for the host to have the "long displacement"
1845 facility. The oldest supported machine model is z990.
1847 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1848 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1850 * The JIT's register allocator is significantly faster, making the JIT
1851 as a whole somewhat faster, so JIT-intensive activities, for example
1852 program startup, are modestly faster, around 5%.
1854 * There have been changes to the default settings of several command
1855 line flags, as detailed below.
1857 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1858 capable hosts, the simulated CPUID will now indicate AVX2 support.
1860 * ==================== TOOL CHANGES ====================
1864 - The default value for --leak-check-heuristics has been changed from
1865 "none" to "all". This helps to reduce the number of possibly
1866 lost blocks, in particular for C++ applications.
1868 - The default value for --keep-stacktraces has been changed from
1869 "malloc-then-free" to "malloc-and-free". This has a small cost in
1870 memory (one word per malloc-ed block) but allows Memcheck to show the
1871 3 stacktraces of a dangling reference: where the block was allocated,
1872 where it was freed, and where it is acccessed after being freed.
1874 - The default value for --partial-loads-ok has been changed from "no" to
1875 "yes", so as to avoid false positive errors resulting from some kinds
1876 of vectorised loops.
1878 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1879 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1880 than get_vbits when you need to associate byte data value with
1881 their corresponding validity bits.
1883 - The 'block_list' monitor command has been enhanced:
1884 o it can print a range of loss records
1885 o it now accepts an optional argument 'limited <max_blocks>'
1886 to control the number of blocks printed.
1887 o if a block has been found using a heuristic, then
1888 'block_list' now shows the heuristic after the block size.
1889 o the loss records/blocks to print can be limited to the blocks
1890 found via specified heuristics.
1892 - The C helper functions used to instrument loads on
1893 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1894 replaced by handwritten assembly sequences. This gives speedups
1895 in the region of 0% to 7% for those targets only.
1897 - A new command line option, --expensive-definedness-checks=yes|no,
1898 has been added. This is useful for avoiding occasional invalid
1899 uninitialised-value errors in optimised code. Watch out for
1900 runtime degradation, as this can be up to 25%. As always, though,
1901 the slowdown is highly application specific. The default setting
1906 - A new monitor command 'all_snapshots <filename>' dumps all
1907 snapshots taken so far.
1911 - Significant memory reduction and moderate speedups for
1912 --history-level=full for applications accessing a lot of memory
1913 with many different stacktraces.
1915 - The default value for --conflict-cache-size=N has been doubled to
1916 2000000. Users that were not using the default value should
1917 preferably also double the value they give.
1919 The default was changed due to the changes in the "full history"
1920 implementation. Doubling the value gives on average a slightly more
1921 complete history and uses similar memory (or significantly less memory
1922 in the worst case) than the previous implementation.
1924 - The Helgrind monitor command 'info locks' now accepts an optional
1925 argument 'lock_addr', which shows information about the lock at the
1928 - When using --history-level=full, the new Helgrind monitor command
1929 'accesshistory <addr> [<len>]' will show the recorded accesses for
1930 <len> (or 1) bytes at <addr>.
1932 * ==================== OTHER CHANGES ====================
1934 * The default value for the --smc-check option has been changed from
1935 "stack" to "all-non-file" on targets that provide automatic D-I
1936 cache coherence (x86, amd64 and s390x). The result is to provide,
1937 by default, transparent support for JIT generated and self-modifying
1938 code on all targets.
1940 * Mac OS X only: the default value for the --dsymutil option has been
1941 changed from "no" to "yes", since any serious usage on Mac OS X
1942 always required it to be "yes".
1944 * The command line options --db-attach and --db-command have been removed.
1945 They were deprecated in 3.10.0.
1947 * When a process dies due to a signal, Valgrind now shows the signal
1948 and the stacktrace at default verbosity (i.e. verbosity 1).
1950 * The address description logic used by Memcheck and Helgrind now
1951 describes addresses in anonymous segments, file mmap-ed segments,
1952 shared memory segments and the brk data segment.
1954 * The new option --error-markers=<begin>,<end> can be used to mark the
1955 begin/end of errors in textual output mode, to facilitate
1956 searching/extracting errors in output files that mix valgrind errors
1957 with program output.
1959 * The new option --max-threads=<number> can be used to change the number
1960 of threads valgrind can handle. The default is 500 threads which
1961 should be more than enough for most applications.
1963 * The new option --valgrind-stacksize=<number> can be used to change the
1964 size of the private thread stacks used by Valgrind. This is useful
1965 for reducing memory use or increasing the stack size if Valgrind
1966 segfaults due to stack overflow.
1968 * The new option --avg-transtab-entry-size=<number> can be used to specify
1969 the expected instrumented block size, either to reduce memory use or
1970 to avoid excessive retranslation.
1972 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1974 * New and modified GDB server monitor features:
1976 - When a signal is reported in GDB, you can now use the GDB convenience
1977 variable $_siginfo to examine detailed signal information.
1979 - Valgrind's gdbserver now allows the user to change the signal
1980 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1981 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1982 continue without passing the signal to the process.
1984 - With GDB >= 7.10, the command 'target remote'
1985 will automatically load the executable file of the process running
1986 under Valgrind. This means you do not need to specify the executable
1987 file yourself, GDB will discover it itself. See GDB documentation about
1988 'qXfer:exec-file:read' packet for more info.
1990 * ==================== FIXED BUGS ====================
1992 The following bugs have been fixed or resolved. Note that "n-i-bz"
1993 stands for "not in bugzilla" -- that is, a bug that was reported to us
1994 but never got a bugzilla entry. We encourage you to file bugs in
1995 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1996 than mailing the developers (or mailing lists) directly -- bugs that
1997 are not entered into bugzilla tend to get forgotten about or ignored.
1999 To see details of a given bug, visit
2000 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2001 where XXXXXX is the bug number as listed below.
2003 116002 VG_(printf): Problems with justification of strings and integers
2004 155125 avoid cutting away file:lineno after long function name
2005 197259 Unsupported arch_prtctl PR_SET_GS option
2006 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
2007 201216 Fix Valgrind does not support pthread_sigmask() on OS X
2008 201435 Fix Darwin: -v does not show kernel version
2009 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
2010 211256 Fixed an outdated comment regarding the default platform.
2011 211529 Incomplete call stacks for code compiled by newer versions of MSVC
2012 211926 Avoid compilation warnings in valgrind.h with -pedantic
2013 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
2015 226609 Crediting upstream authors in man page
2016 231257 Valgrind omits path when executing script from shebang line
2017 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
2018 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
2019 269360 s390x: Fix addressing mode selection for compare-and-swap
2020 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
2022 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
2023 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
2024 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
2025 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
2026 330147 libmpiwrap PMPI_Get_count returns undefined value
2027 333051 mmap of huge pages fails due to incorrect alignment
2029 334802 valgrind does not always explain why a given option is bad
2030 335618 mov.w rN, pc/sp (ARM32)
2031 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
2036 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
2037 338602 AVX2 bit in CPUID missing
2038 338606 Strange message for scripts with invalid interpreter
2039 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
2040 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
2041 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
2043 339156 gdbsrv not called for fatal signal
2044 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
2045 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
2046 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
2047 339442 Fix testsuite build failure on OS X 10.9
2048 339542 Enable compilation with Intel's ICC compiler
2049 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
2050 339688 Mac-specific ASM does not support .version directive (cpuid,
2051 tronical and pushfpopf tests)
2052 339745 Valgrind crash when check Marmalade app (partial fix)
2053 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
2054 339778 Linux/TileGx platform support for Valgrind
2055 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
2056 339789 Fix none/tests/execve test on Mac OS X 10.9
2057 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
2058 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
2059 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
2060 340392 Allow user to select more accurate definedness checking in memcheck
2061 to avoid invalid complaints on optimised code
2062 340430 Fix some grammatical weirdness in the manual.
2063 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
2064 341419 Signal handler ucontext_t not filled out correctly on OS X
2065 341539 VG_(describe_addr) should not describe address as belonging to client
2066 segment if it is past the heap end
2067 341613 Enable building of manythreads and thread-exits tests on Mac OS X
2068 341615 Fix none/tests/darwin/access_extended test on Mac OS X
2069 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
2070 341789 aarch64: shmat fails with valgrind on ARMv8
2071 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
2072 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
2073 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
2074 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
2075 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
2076 342221 socket connect false positive uninit memory for unknown af family
2077 342353 Allow dumping full massif output while valgrind is still running
2078 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
2082 342603 Add I2C_SMBUS ioctl support
2083 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
2084 342683 Mark memory past the initial brk limit as unaddressable
2085 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
2086 342795 Internal glibc __GI_mempcpy call should be intercepted
2087 342841 s390x: Support instructions fiebr(a) and fidbr(a)
2088 343012 Unhandled syscall 319 (memfd_create)
2089 343069 Patch updating v4l2 API support
2090 343173 helgrind crash during stack unwind
2091 343219 fix GET_STARTREGS for arm
2092 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
2093 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
2094 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
2095 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
2096 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
2097 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
2098 343597 ppc64le: incorrect use of offseof macro
2099 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
2100 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
2101 343732 Unhandled syscall 144 (setgid) on aarch64
2102 343733 Unhandled syscall 187 (msgctl and related) on aarch64
2103 343802 s390x: False positive "conditional jump or move depends on [..]
2104 343902 --vgdb=yes doesn't break when --xml=yes is used
2105 343967 Don't warn about setuid/setgid/setcap executable for directories
2106 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
2107 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
2108 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
2109 344054 www - update info for Solaris/illumos
2110 344416 'make regtest' does not work cleanly on OS X
2111 344235 Remove duplicate include of pub_core_aspacemgr.h
2112 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
2113 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
2114 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
2115 344314 callgrind_annotate ... warnings about commands containing newlines
2116 344318 socketcall should wrap recvmmsg and sendmmsg
2117 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
2118 344416 Fix 'make regtest' does not work cleanly on OS X
2119 344499 Fix compilation for Linux kernel >= 4.0.0
2120 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
2121 unix:349 (__pthread_fchdir)
2122 344559 Garbage collection of unused segment names in address space manager
2123 344560 Fix stack traces missing penultimate frame on OS X
2124 344621 Fix memcheck/tests/err_disable4 test on OS X
2125 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
2126 344702 Fix missing libobjc suppressions on OS X 10.10
2128 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
2129 344939 Fix memcheck/tests/xml1 on OS X 10.10
2130 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
2131 345079 Fix build problems in VEX/useful/test_main.c
2132 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
2133 345177 arm64: prfm (reg) not implemented
2134 345215 Performance improvements for the register allocator
2135 345248 add support for Solaris OS in valgrind
2136 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
2137 345394 Fix memcheck/tests/strchr on OS X
2138 345637 Fix memcheck/tests/sendmsg on OS X
2139 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
2140 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
2141 345887 Fix an assertion in the address space manager
2142 345928 amd64: callstack only contains current function for small stacks
2143 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
2144 345987 MIPS64: Implement cavium LHX instruction
2145 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
2146 346185 Fix typo saving altivec register v24
2147 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
2148 and LibVEX_GuestPPC64_get_CR()
2149 346270 Regression tests none/tests/jm_vec/isa_2_07 and
2150 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
2151 346307 fuse filesystem syscall deadlocks
2152 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
2153 346411 MIPS: SysRes::_valEx handling is incorrect
2154 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
2155 346474 PPC64 Power 8, spr TEXASRU register not supported
2156 346487 Compiler generates "note" about a future ABI change for PPC64
2157 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
2158 and causing spurious "invalid read of size 8" warnings
2159 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
2160 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
2161 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
2162 347322 Power PC regression test cleanup
2163 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
2165 347389 unhandled syscall: 373 (Linux ARM syncfs)
2166 347686 Patch set to cleanup PPC64 regtests
2167 347978 Remove bash dependencies where not needed
2168 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
2169 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
2171 348102 Patch updating v4l2 API support
2172 348247 amd64 front end: jno jumps wrongly when overflow is not set
2173 348269 Improve mmap MAP_HUGETLB support.
2174 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
2175 348345 Assertion fails for negative lineno
2176 348377 Unsupported ARM instruction: yield
2177 348565 Fix detection of command line option availability for clang
2178 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
2179 348728 Fix broken check for VIDIOC_G_ENC_INDEX
2180 348748 Fix redundant condition
2181 348890 Fix clang warning about unsupported --param inline-unit-growth=900
2182 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
2183 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
2184 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
2185 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
2186 349626 Implemented additional Xen hypercalls
2187 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
2188 349790 Clean up of the hardware capability checking utilities.
2189 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
2190 349874 Fix typos in source code
2191 349879 memcheck: add handwritten assembly for helperc_LOADV*
2192 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
2193 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
2194 350202 Add limited param to 'monitor block_list'
2195 350290 s390x: Support instructions fixbr(a)
2196 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
2197 350809 Fix none/tests/async-sigs for Solaris
2198 350811 Remove reference to --db-attach which has been removed.
2199 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
2200 350854 hard-to-understand code in VG_(load_ELF)()
2201 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2202 351386 Solaris: Cannot run ld.so.1 under Valgrind
2203 351474 Fix VG_(iseqsigset) as obvious
2204 351531 Typo in /include/vki/vki-xen-physdev.h header guard
2205 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
2206 351858 ldsoexec support on Solaris
2207 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
2208 352130 helgrind reports false races for printfs using mempcpy on FILE* state
2209 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
2210 352320 arm64 crash on none/tests/nestedfs
2211 352765 Vbit test fails on Power 6
2212 352768 The mbar instruction is missing from the Power PC support
2213 352769 Power PC program priority register (PPR) is not supported
2214 n-i-bz Provide implementations of certain compiler builtins to support
2215 compilers that may not provide those
2216 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
2217 n-i-bz Fix compilation on distros with glibc < 2.5
2218 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
2219 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
2220 n-i-bz mremap did not work properly on shared memory
2221 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
2222 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
2224 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
2225 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
2226 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
2230 Release 3.10.1 (25 November 2014)
2231 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2232 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
2233 and backports fixes for all reported missing AArch64 ARMv8 instructions
2234 and syscalls from the trunk. If you package or deliver 3.10.0 for others
2235 to use, you might want to consider upgrading to 3.10.1 instead.
2237 The following bugs have been fixed or resolved. Note that "n-i-bz"
2238 stands for "not in bugzilla" -- that is, a bug that was reported to us
2239 but never got a bugzilla entry. We encourage you to file bugs in
2240 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2241 than mailing the developers (or mailing lists) directly -- bugs that
2242 are not entered into bugzilla tend to get forgotten about or ignored.
2244 To see details of a given bug, visit
2245 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2246 where XXXXXX is the bug number as listed below.
2248 335440 arm64: ld1 (single structure) is not implemented
2249 335713 arm64: unhanded instruction: prfm (immediate)
2250 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
2251 339182 ppc64: AvSplat ought to load destination vector register with [..]
2252 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
2253 339433 ppc64 lxvw4x instruction uses four 32-byte loads
2254 339645 Use correct tag names in sys_getdents/64 wrappers
2255 339706 Fix false positive for ioctl(TIOCSIG) on linux
2256 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
2257 339853 arm64 times syscall unknown
2258 339855 arm64 unhandled getsid/setsid syscalls
2259 339858 arm64 dmb sy not implemented
2260 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
2261 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
2262 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
2264 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
2265 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
2266 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
2267 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
2268 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
2269 340509 arm64: unhandled instruction fcvtas
2270 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
2271 340632 arm64: unhandled instruction fcvtas
2272 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
2273 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
2274 340788 warning: unhandled syscall: 318 (getrandom)
2275 340807 disInstr(arm): unhandled instruction: 0xEE989B20
2276 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
2277 340922 arm64: unhandled getgroups/setgroups syscalls
2278 350251 Fix typo in VEX utility program (test_main.c).
2279 350407 arm64: unhandled instruction ucvtf (vector, integer)
2280 350809 none/tests/async-sigs breaks when run under cron on Solaris
2281 350811 update README.solaris after r15445
2282 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
2283 350854 strange code in VG_(load_ELF)()
2284 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2285 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
2286 n-i-bz Add missing ]] to terminate CDATA.
2287 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
2288 n-i-bz Enable sys_fadvise64_64 on arm32.
2289 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
2290 n-i-bz Add test cases for all known arm64 load/store instructions.
2291 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
2292 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
2293 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
2294 n-i-bz arm64: Implement "fcvtpu w, s".
2295 n-i-bz arm64: implement ADDP and various others
2296 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
2297 n-i-bz arm64: enable FCVT{A,N}S X,S.
2299 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
2303 Release 3.10.0 (10 September 2014)
2304 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2306 3.10.0 is a feature release with many improvements and the usual
2307 collection of bug fixes.
2309 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
2310 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
2311 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
2312 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
2313 significantly improved relative to the 3.9.0 release.
2315 * ================== PLATFORM CHANGES =================
2317 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
2318 is mostly complete, and is usable, but some SIMD instructions are as
2321 * Support for little-endian variant of the 64-bit POWER architecture.
2323 * Support for Android on MIPS32.
2325 * Support for 64bit FPU on MIPS32 platforms.
2327 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
2329 * Configuration for and running on Android targets has changed.
2330 See README.android in the source tree for details.
2332 * ================== DEPRECATED FEATURES =================
2334 * --db-attach is now deprecated and will be removed in the next
2335 valgrind feature release. The built-in GDB server capabilities are
2336 superior and should be used instead. Learn more here:
2337 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
2339 * ==================== TOOL CHANGES ====================
2343 - Client code can now selectively disable and re-enable reporting of
2344 invalid address errors in specific ranges using the new client
2345 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
2346 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
2348 - Leak checker: there is a new leak check heuristic called
2349 "length64". This is used to detect interior pointers pointing 8
2350 bytes inside a block, on the assumption that the first 8 bytes
2351 holds the value "block size - 8". This is used by
2352 sqlite3MemMalloc, for example.
2354 - Checking of system call parameters: if a syscall parameter
2355 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
2356 several fields not initialised, an error is now reported for each
2357 field. Previously, an error was reported only for the first
2358 uninitialised field.
2360 - Mismatched alloc/free checking: a new flag
2361 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
2362 such checks if necessary.
2366 - Improvements to error messages:
2368 o Race condition error message involving heap allocated blocks also
2369 show the thread number that allocated the raced-on block.
2371 o All locks referenced by an error message are now announced.
2372 Previously, some error messages only showed the lock addresses.
2374 o The message indicating where a lock was first observed now also
2375 describes the address/location of the lock.
2377 - Helgrind now understands the Ada task termination rules and
2378 creates a happens-before relationship between a terminated task
2379 and its master. This avoids some false positives and avoids a big
2380 memory leak when a lot of Ada tasks are created and terminated.
2381 The interceptions are only activated with forthcoming releases of
2382 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
2384 - A new GDB server monitor command "info locks" giving the list of
2385 locks, their location, and their status.
2389 - callgrind_control now supports the --vgdb-prefix argument,
2390 which is needed if valgrind was started with this same argument.
2392 * ==================== OTHER CHANGES ====================
2394 * Unwinding through inlined function calls. Stack unwinding can now
2395 make use of Dwarf3 inlined-unwind information if it is available.
2396 The practical effect is that inlined calls become visible in stack
2397 traces. The suppression matching machinery has been adjusted
2398 accordingly. This is controlled by the new option
2399 --read-inline-info=yes|no. Currently this is enabled by default
2400 only on Linux and Android targets and only for the tools Memcheck,
2403 * Valgrind can now read EXIDX unwind information on 32-bit ARM
2404 targets. If an object contains both CFI and EXIDX unwind
2405 information, Valgrind will prefer the CFI over the EXIDX. This
2406 facilitates unwinding through system libraries on arm-android
2409 * Address description logic has been improved and is now common
2410 between Memcheck and Helgrind, resulting in better address
2411 descriptions for some kinds of error messages.
2413 * Error messages about dubious arguments (eg, to malloc or calloc) are
2414 output like other errors. This means that they can be suppressed
2415 and they have a stack trace.
2417 * The C++ demangler has been updated for better C++11 support.
2419 * New and modified GDB server monitor features:
2421 - Thread local variables/storage (__thread) can now be displayed.
2423 - The GDB server monitor command "v.info location <address>"
2424 displays information about an address. The information produced
2425 depends on the tool and on the options given to valgrind.
2426 Possibly, the following are described: global variables, local
2427 (stack) variables, allocated or freed blocks, ...
2429 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
2430 ask the GDB server to stop at the start of program execution, at
2431 the end of the program execution and on Valgrind internal errors.
2433 - A new monitor command "v.info stats" shows various Valgrind core
2434 and tool statistics.
2436 - A new monitor command "v.set hostvisibility" allows the GDB server
2437 to provide access to Valgrind internal host status/memory.
2439 * A new option "--aspace-minaddr=<address>" can in some situations
2440 allow the use of more memory by decreasing the address above which
2441 Valgrind maps memory. It can also be used to solve address
2442 conflicts with system libraries by increasing the default value.
2443 See user manual for details.
2445 * The amount of memory used by Valgrind to store debug info (unwind
2446 info, line number information and symbol data) has been
2447 significantly reduced, even though Valgrind now reads more
2448 information in order to support unwinding of inlined function calls.
2450 * Dwarf3 handling with --read-var-info=yes has been improved:
2452 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
2454 - Code compiled with
2455 -ffunction-sections -fdata-sections -Wl,--gc-sections
2456 no longer causes assertion failures.
2458 * Improved checking for the --sim-hints= and --kernel-variant=
2459 options. Unknown strings are now detected and reported to the user
2462 * The semantics of stack start/end boundaries in the valgrind.h
2463 VALGRIND_STACK_REGISTER client request has been clarified and
2464 documented. The convention is that start and end are respectively
2465 the lowest and highest addressable bytes of the stack.
2467 * ==================== FIXED BUGS ====================
2469 The following bugs have been fixed or resolved. Note that "n-i-bz"
2470 stands for "not in bugzilla" -- that is, a bug that was reported to us
2471 but never got a bugzilla entry. We encourage you to file bugs in
2472 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2473 than mailing the developers (or mailing lists) directly -- bugs that
2474 are not entered into bugzilla tend to get forgotten about or ignored.
2476 To see details of a given bug, visit
2477 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2478 where XXXXXX is the bug number as listed below.
2480 175819 Support for ipv6 socket reporting with --track-fds
2481 232510 make distcheck fails
2482 249435 Analyzing wine programs with callgrind triggers a crash
2483 278972 support for inlined function calls in stacktraces and suppression
2485 291310 FXSAVE instruction marks memory as undefined on amd64
2486 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2487 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2488 315199 vgcore file for threaded app does not show which thread crashed
2489 315952 tun/tap ioctls are not supported
2490 323178 Unhandled instruction: PLDW register (ARM)
2491 323179 Unhandled instruction: PLDW immediate (ARM)
2492 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2493 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2494 325124 [MIPSEL] Compilation error
2495 325477 Phase 4 support for IBM Power ISA 2.07
2496 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2497 325628 Phase 5 support for IBM Power ISA 2.07
2498 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2499 325751 Missing the two privileged Power PC Transactional Memory Instructions
2500 325816 Phase 6 support for IBM Power ISA 2.07
2501 325856 Make SGCheck fail gracefully on unsupported platforms
2502 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2503 326436 DRD: False positive in libstdc++ std::list::push_back
2504 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2505 326462 Refactor vgdb to isolate invoker stuff into separate module
2506 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2507 326623 DRD: false positive conflict report in a field assignment
2508 326724 Valgrind does not compile on OSX 1.9 Mavericks
2509 326816 Intercept for __strncpy_sse2_unaligned missing?
2510 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2511 326983 Clear direction flag after tests on amd64.
2512 327212 Do not prepend the current directory to absolute path names.
2513 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2514 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2515 327284 s390x: Fix translation of the risbg instruction
2516 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2517 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2518 327916 DW_TAG_typedef may have no name
2519 327943 s390x: add a redirection for the 'index' function
2520 328100 XABORT not implemented
2521 328205 Implement additional Xen hypercalls
2522 328454 add support Backtraces with ARM unwind tables (EXIDX)
2523 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2524 328711 valgrind.1 manpage "memcheck options" section is badly generated
2525 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2526 329612 Incorrect handling of AT_BASE for image execution
2527 329694 clang warns about using uninitialized variable
2528 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2529 330228 mmap must align to VKI_SHMLBA on mips32
2530 330257 LLVM does not support `-mno-dynamic-no-pic` option
2531 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2532 330459 --track-fds=yes doesn't track eventfds
2533 330469 Add clock_adjtime syscall support
2534 330594 Missing sysalls on PowerPC / uClibc
2535 330622 Add test to regression suite for POWER instruction: dcbzl
2536 330939 Support for AMD's syscall instruction on x86
2538 330941 Typo in PRE(poll) syscall wrapper
2539 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2540 331254 Fix expected output for memcheck/tests/dw4
2541 331255 Fix race condition in test none/tests/coolo_sigaction
2542 331257 Fix type of jump buffer in test none/tests/faultstatus
2543 331305 configure uses bash specific syntax
2544 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2545 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2546 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2547 331829 Unexpected ioctl opcode sign extension
2548 331830 ppc64: WARNING: unhandled syscall: 96/97
2549 331839 drd/tests/sem_open specifies invalid semaphore name
2550 331847 outcome of drd/tests/thread_name is nondeterministic
2551 332037 Valgrind cannot handle Thumb "add pc, reg"
2552 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2553 consistency checks enabled
2554 332263 intercepts for pthread_rwlock_timedrdlock and
2555 pthread_rwlock_timedwrlock are incorrect
2556 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2558 332276 Implement additional Xen hypercalls
2559 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2560 332765 Fix ms_print to create temporary files in a proper directory
2561 333072 drd: Add semaphore annotations
2562 333145 Tests for missaligned PC+#imm access for arm
2563 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2564 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2565 333248 WARNING: unhandled syscall: unix:443
2566 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2567 333501 cachegrind: assertion: Cache set count is not a power of two.
2570 333666 Recognize MPX instructions and bnd prefix.
2571 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2572 333817 Valgrind reports the memory areas written to by the SG_IO
2574 334049 lzcnt fails silently (x86_32)
2575 334384 Valgrind does not have support Little Endian support for
2577 334585 recvmmsg unhandled (+patch) (arm)
2578 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2579 334727 Build fails with -Werror=format-security
2580 334788 clarify doc about --log-file initial program directory
2581 334834 PPC64 Little Endian support, patch 2
2582 334836 PPC64 Little Endian support, patch 3 testcase fixes
2583 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2584 335034 Unhandled ioctl: HCIGETDEVLIST
2585 335155 vgdb, fix error print statement.
2586 335262 arm64: movi 8bit version is not supported
2587 335263 arm64: dmb instruction is not implemented
2588 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2589 335496 arm64: sbc/abc instructions are not implemented
2590 335554 arm64: unhandled instruction: abs
2591 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2592 335735 arm64: unhandled instruction: cnt
2593 335736 arm64: unhandled instruction: uaddlv
2594 335848 arm64: unhandled instruction: {s,u}cvtf
2595 335902 arm64: unhandled instruction: sli
2596 335903 arm64: unhandled instruction: umull (vector)
2597 336055 arm64: unhandled instruction: mov (element)
2598 336062 arm64: unhandled instruction: shrn{,2}
2599 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2600 336189 arm64: unhandled Instruction: mvn
2601 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2602 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2603 336772 Make moans about unknown ioctls more informative
2604 336957 Add a section about the Solaris/illumos port on the webpage
2605 337094 ifunc wrapper is broken on ppc64
2606 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2607 337528 leak check heuristic for block prefixed by length as 64bit number
2608 337740 Implement additional Xen hypercalls
2609 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2610 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2611 337871 deprecate --db-attach
2612 338023 Add support for all V4L2/media ioctls
2613 338024 inlined functions are not shown if DW_AT_ranges is used
2614 338106 Add support for 'kcmp' syscall
2615 338115 DRD: computed conflict set differs from actual after fork
2616 338160 implement display of thread local storage in gdbsrv
2617 338205 configure.ac and check for -Wno-tautological-compare
2618 338300 coredumps are missing one byte of every segment
2619 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2620 338499 --sim-hints parsing broken due to wrong order in tokens
2621 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2622 338681 Unable to unwind through clone thread created on i386-linux
2623 338698 race condition between gdbsrv and vgdb on startup
2624 338703 helgrind on arm-linux gets false positives in dynamic loader
2625 338791 alt dwz files can be relative of debug/main file
2626 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2627 338932 build V-trunk with gcc-trunk
2628 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2629 345079 Fix build problems in VEX/useful/test_main.c
2630 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2631 n-i-bz s390x: Fix memory corruption for multithreaded applications
2632 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2633 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2634 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2635 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2636 n-i-bz Add command-line option --default-suppressions
2637 n-i-bz Add support for BLKDISCARDZEROES ioctl
2638 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2639 n-i-bz Add support for sys_pivot_root and sys_unshare
2641 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2642 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2643 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2647 Release 3.9.0 (31 October 2013)
2648 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2649 3.9.0 is a feature release with many improvements and the usual
2650 collection of bug fixes.
2652 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2653 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2654 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2655 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2657 * ================== PLATFORM CHANGES =================
2659 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2660 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2662 * Support for MIPS DSP ASE on MIPS32 platforms.
2664 * Support for s390x Decimal Floating Point instructions on hosts that
2665 have the DFP facility installed.
2667 * Support for POWER8 (Power ISA 2.07) instructions
2669 * Support for Intel AVX2 instructions. This is available only on 64
2672 * Initial support for Intel Transactional Synchronization Extensions,
2675 * Initial support for Hardware Transactional Memory on POWER.
2677 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2678 run large GUI apps tolerably well.
2680 * ==================== TOOL CHANGES ====================
2684 - Improvements in handling of vectorised code, leading to
2685 significantly fewer false error reports. You need to use the flag
2686 --partial-loads-ok=yes to get the benefits of these changes.
2688 - Better control over the leak checker. It is now possible to
2689 specify which leak kinds (definite/indirect/possible/reachable)
2690 should be displayed, which should be regarded as errors, and which
2691 should be suppressed by a given leak suppression. This is done
2692 using the options --show-leak-kinds=kind1,kind2,..,
2693 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2694 "match-leak-kinds:" line in suppression entries, respectively.
2696 Note that generated leak suppressions contain this new line and
2697 are therefore more specific than in previous releases. To get the
2698 same behaviour as previous releases, remove the "match-leak-kinds:"
2699 line from generated suppressions before using them.
2701 - Reduced "possible leak" reports from the leak checker by the use
2702 of better heuristics. The available heuristics provide detection
2703 of valid interior pointers to std::stdstring, to new[] allocated
2704 arrays with elements having destructors and to interior pointers
2705 pointing to an inner part of a C++ object using multiple
2706 inheritance. They can be selected individually using the
2707 option --leak-check-heuristics=heur1,heur2,...
2709 - Better control of stacktrace acquisition for heap-allocated
2710 blocks. Using the --keep-stacktraces option, it is possible to
2711 control independently whether a stack trace is acquired for each
2712 allocation and deallocation. This can be used to create better
2713 "use after free" errors or to decrease Valgrind's resource
2714 consumption by recording less information.
2716 - Better reporting of leak suppression usage. The list of used
2717 suppressions (shown when the -v option is given) now shows, for
2718 each leak suppressions, how many blocks and bytes it suppressed
2719 during the last leak search.
2723 - False errors resulting from the use of statically initialised
2724 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2727 - False errors resulting from the use of pthread_cond_waits that
2728 timeout, have been removed.
2730 * ==================== OTHER CHANGES ====================
2732 * Some attempt to tune Valgrind's space requirements to the expected
2733 capabilities of the target:
2735 - The default size of the translation cache has been reduced from 8
2736 sectors to 6 on Android platforms, since each sector occupies
2737 about 40MB when using Memcheck.
2739 - The default size of the translation cache has been increased to 16
2740 sectors on all other platforms, reflecting the fact that large
2741 applications require instrumentation and storage of huge amounts
2742 of code. For similar reasons, the number of memory mapped
2743 segments that can be tracked has been increased by a factor of 6.
2745 - In all cases, the maximum number of sectors in the translation
2746 cache can be controlled by the new flag --num-transtab-sectors.
2748 * Changes in how debug info (line numbers, etc) is read:
2750 - Valgrind no longer temporarily mmaps the entire object to read
2751 from it. Instead, reading is done through a small fixed sized
2752 buffer. This avoids virtual memory usage spikes when Valgrind
2753 reads debuginfo from large shared objects.
2755 - A new experimental remote debug info server. Valgrind can read
2756 debug info from a different machine (typically, a build host)
2757 where debuginfo objects are stored. This can save a lot of time
2758 and hassle when running Valgrind on resource-constrained targets
2759 (phones, tablets) when the full debuginfo objects are stored
2760 somewhere else. This is enabled by the --debuginfo-server=
2763 - Consistency checking between main and debug objects can be
2764 disabled using the --allow-mismatched-debuginfo option.
2766 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2767 scanning can recover stack traces in some cases when the normal
2768 unwind mechanisms fail. Stack scanning is best described as "a
2769 nasty, dangerous and misleading hack" and so is disabled by default.
2770 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2773 * Detection and merging of recursive stack frame cycles. When your
2774 program has recursive algorithms, this limits the memory used by
2775 Valgrind for recorded stack traces and avoids recording
2776 uninteresting repeated calls. This is controlled by the command
2777 line option --merge-recursive-frame and by the monitor command
2778 "v.set merge-recursive-frames".
2780 * File name and line numbers for used suppressions. The list of used
2781 suppressions (shown when the -v option is given) now shows, for each
2782 used suppression, the file name and line number where the suppression
2785 * New and modified GDB server monitor features:
2787 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2788 that can be used to execute gdbserver monitor commands from the
2791 - A new monitor command, "v.info open_fds", that gives the list of
2792 open file descriptors and additional details.
2794 - An optional message in the "v.info n_errs_found" monitor command,
2795 for example "v.info n_errs_found test 1234 finished", allowing a
2796 comment string to be added to the process output, perhaps for the
2797 purpose of separating errors of different tests or test phases.
2799 - A new monitor command "v.info execontext" that shows information
2800 about the stack traces recorded by Valgrind.
2802 - A new monitor command "v.do expensive_sanity_check_general" to run
2803 some internal consistency checks.
2805 * New flag --sigill-diagnostics to control whether a diagnostic
2806 message is printed when the JIT encounters an instruction it can't
2807 translate. The actual behavior -- delivery of SIGILL to the
2808 application -- is unchanged.
2810 * The maximum amount of memory that Valgrind can use on 64 bit targets
2811 has been increased from 32GB to 64GB. This should make it possible
2812 to run applications on Memcheck that natively require up to about 35GB.
2814 * ==================== FIXED BUGS ====================
2816 The following bugs have been fixed or resolved. Note that "n-i-bz"
2817 stands for "not in bugzilla" -- that is, a bug that was reported to us
2818 but never got a bugzilla entry. We encourage you to file bugs in
2819 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2820 than mailing the developers (or mailing lists) directly -- bugs that
2821 are not entered into bugzilla tend to get forgotten about or ignored.
2823 To see details of a given bug, visit
2824 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2825 where XXXXXX is the bug number as listed below.
2827 123837 system call: 4th argument is optional, depending on cmd
2828 135425 memcheck should tell you where Freed blocks were Mallocd
2829 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2830 207815 Adds some of the drm ioctls to syswrap-linux.c
2831 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2832 252955 Impossible to compile with ccache
2833 253519 Memcheck reports auxv pointer accesses as invalid reads.
2834 263034 Crash when loading some PPC64 binaries
2835 269599 Increase deepest backtrace
2836 274695 s390x: Support "compare to/from logical" instructions (z196)
2837 275800 s390x: Autodetect cache info (part 2)
2838 280271 Valgrind reports possible memory leaks on still-reachable std::string
2839 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2840 289578 Backtraces with ARM unwind tables (stack scan flags)
2841 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2842 304832 ppc32: build failure
2843 305431 Use find_buildid shdr fallback for separate .debug files
2844 305728 Add support for AVX2 instructions
2845 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2846 306035 s390x: Fix IR generation for LAAG and friends
2847 306054 s390x: Condition code computation for convert-to-int/logical
2848 306098 s390x: alternate opcode form for convert to/from fixed
2849 306587 Fix cache line detection from auxiliary vector for PPC.
2850 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2851 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2852 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2853 307101 sys_capget second argument can be NULL
2854 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2855 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2856 307113 s390x: DFP support
2857 307141 valgrind does't work in mips-linux system
2858 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2859 307285 x86_amd64 feature test for avx in test suite is wrong
2860 307290 memcheck overlap testcase needs memcpy version filter
2861 307463 Please add "&limit=0" to the "all open bugs" link
2862 307465 --show-possibly-lost=no should reduce the error count / exit code
2863 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2864 307729 pkgconfig support broken valgrind.pc
2865 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2866 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2867 308089 Unhandled syscall on ppc64: prctl
2868 308135 PPC32 MPC8xx has 16 bytes cache size
2869 308321 testsuite memcheck filter interferes with gdb_filter
2871 308341 vgdb should report process exit (or fatal signal)
2872 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2873 308495 Remove build dependency on installed Xen headers
2874 308573 Internal error on 64-bit instruction executed in 32-bit mode
2876 308627 pmovmskb validity bit propagation is imprecise
2877 308644 vgdb command for having the info for the track-fds option
2878 308711 give more info about aspacemgr and arenas in out_of_memory
2879 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2880 308718 ARM implement SMLALBB family of instructions
2881 308886 Missing support for PTRACE_SET/GETREGSET
2882 308930 syscall name_to_handle_at (303 on amd64) not handled
2883 309229 V-bit tester does not report number of tests generated
2884 309323 print unrecognized instuction on MIPS
2885 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2886 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2887 309430 Self hosting ppc64 encounters a vassert error on operand type
2888 309600 valgrind is a bit confused about 0-sized sections
2889 309823 Generate errors for still reachable blocks
2890 309921 PCMPISTRI validity bit propagation is imprecise
2891 309922 none/tests/ppc64/test_dfp5 sometimes fails
2892 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2893 310424 --read-var-info does not properly describe static variables
2894 310792 search additional path for debug symbols
2895 310931 s390x: Message-security assist (MSA) instruction extension [..]
2896 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2897 311318 ARM: "128-bit constant is not implemented" error message
2898 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2899 311690 V crashes because it redirects branches inside of a redirected function
2900 311880 x86_64: make regtest hangs at shell_valid1
2901 311922 WARNING: unhandled syscall: 170
2903 312171 ppc: insn selection for DFP
2904 312571 Rounding mode call wrong for the DFP Iops [..]
2905 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2906 312913 Dangling pointers error should also report the alloc stack trace
2907 312980 Building on Mountain Lion generates some compiler warnings
2908 313267 Adding MIPS64/Linux port to Valgrind
2911 313811 Buffer overflow in assert_fail
2912 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2913 314269 ppc: dead code in insn selection
2914 314718 ARM: implement integer divide instruction (sdiv and udiv)
2915 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2916 315441 sendmsg syscall should ignore unset msghdr msg_flags
2917 315534 msgrcv inside a thread causes valgrind to hang (block)
2918 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2919 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2920 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2921 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2922 316144 valgrind.1 manpage contains unknown ??? strings [..]
2923 316145 callgrind command line options in manpage reference (unknown) [..]
2924 316145 callgrind command line options in manpage reference [..]
2925 316181 drd: Fixed a 4x slowdown for certain applications
2926 316503 Valgrind does not support SSE4 "movntdqa" instruction
2927 316535 Use of |signed int| instead of |size_t| in valgrind messages
2928 316696 fluidanimate program of parsec 2.1 stuck
2929 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2930 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2931 317186 "Impossible happens" when occurs VCVT instruction on ARM
2932 317318 Support for Threading Building Blocks "scalable_malloc"
2933 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2934 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2935 317463 bmi testcase IR SANITY CHECK FAILURE
2936 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2937 318050 libmpiwrap fails to compile with out-of-source build
2938 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2939 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2940 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2941 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2942 318932 Add missing PPC64 and PPC32 system call support
2943 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2944 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2945 319494 VEX Makefile-gcc standalone build update after r2702
2946 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2947 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2948 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2949 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2950 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2951 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2952 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2953 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2954 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2955 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2956 320895 add fanotify support (patch included)
2957 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2958 321065 Valgrind updates for Xen 4.3
2959 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2960 321363 Unhandled instruction: SSAX (ARM + Thumb)
2961 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2962 321466 Unhandled instruction: SHASX (ARM + Thumb)
2963 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2964 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2965 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2966 321620 Unhandled instruction: UASX (ARM + Thumb)
2967 321621 Unhandled instruction: USAX (ARM + Thumb)
2968 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2969 321693 Unhandled instruction: LDRSBT (Thumb)
2970 321694 Unhandled instruction: UQASX (ARM + Thumb)
2971 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2972 321697 Unhandled instruction: UHASX (ARM + Thumb)
2973 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2974 321704 Unhandled instruction: REVSH (ARM + Thumb)
2975 321730 Add cg_diff and cg_merge man pages
2976 321738 Add vgdb and valgrind-listener man pages
2978 321891 Unhandled instruction: LDRHT (Thumb)
2979 321960 pthread_create() then alloca() causing invalid stack write errors
2980 321969 ppc32 and ppc64 don't support [lf]setxattr
2981 322254 Show threadname together with tid if set by application
2982 322294 Add initial support for IBM Power ISA 2.07
2983 322368 Assertion failure in wqthread_hijack under OS X 10.8
2984 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2985 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2986 322851 0bXXX binary literal syntax is not standard
2987 323035 Unhandled instruction: LDRSHT(Thumb)
2988 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2989 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2990 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2991 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2992 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2993 323437 Phase 2 support for IBM Power ISA 2.07
2994 323713 Support mmxext (integer sse) subset on i386 (athlon)
2995 323803 Transactional memory instructions are not supported for Power
2996 323893 SSE3 not available on amd cpus in valgrind
2997 323905 Probable false positive from Valgrind/drd on close()
2998 323912 valgrind.h header isn't compatible for mingw64
2999 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
3000 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
3001 324181 mmap does not handle MAP_32BIT
3002 324227 memcheck false positive leak when a thread calls exit+block [..]
3003 324421 Support for fanotify API on ARM architecture
3004 324514 gdbserver monitor cmd output behaviour consistency [..]
3005 324518 ppc64: Emulation of dcbt instructions does not handle [..]
3006 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
3007 324582 When access is made to freed memory, report both allocation [..]
3008 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
3009 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
3010 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
3011 324834 Unhandled instructions in Microsoft C run-time for x86_64
3012 324894 Phase 3 support for IBM Power ISA 2.07
3013 326091 drd: Avoid false race reports from optimized strlen() impls
3014 326113 valgrind libvex hwcaps error on AMD64
3015 n-i-bz Some wrong command line options could be ignored
3016 n-i-bz patch to allow fair-sched on android
3017 n-i-bz report error for vgdb snapshot requested before execution
3018 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
3020 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
3024 Release 3.8.1 (19 September 2012)
3025 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3026 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
3027 that occur moderately frequently in real use cases, adds support for
3028 some missing instructions on ARM, and fixes a deadlock condition on
3029 MacOSX. If you package or deliver 3.8.0 for others to use, you might
3030 want to consider upgrading to 3.8.1 instead.
3032 The following bugs have been fixed or resolved. Note that "n-i-bz"
3033 stands for "not in bugzilla" -- that is, a bug that was reported to us
3034 but never got a bugzilla entry. We encourage you to file bugs in
3035 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3036 than mailing the developers (or mailing lists) directly -- bugs that
3037 are not entered into bugzilla tend to get forgotten about or ignored.
3039 To see details of a given bug, visit
3040 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3041 where XXXXXX is the bug number as listed below.
3044 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
3045 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
3046 298281 wcslen causes false(?) uninitialised value warnings
3047 301281 valgrind hangs on OS X when the process calls system()
3048 304035 disInstr(arm): unhandled instruction 0xE1023053
3049 304867 implement MOVBE instruction in x86 mode
3050 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
3051 305042 amd64: implement 0F 7F encoding of movq between two registers
3052 305199 ARM: implement QDADD and QDSUB
3053 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
3054 305513 killed by fatal signal: SIGSEGV
3055 305690 DRD reporting invalid semaphore when sem_trywait fails
3056 305926 Invalid alignment checks for some AVX instructions
3057 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
3058 306310 3.8.0 release tarball missing some files
3059 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
3060 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
3061 n-i-bz shmat of a segment > 4Gb does not work
3062 n-i-bz simulate_control_c script wrong USR1 signal number on mips
3063 n-i-bz vgdb ptrace calls wrong on mips [...]
3064 n-i-bz Fixes for more MPI false positives
3065 n-i-bz exp-sgcheck's memcpy causes programs to segfault
3066 n-i-bz OSX build w/ clang: asserts at startup
3067 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
3068 n-i-bz fix a couple of union tag-vs-field mixups
3069 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
3071 The following bugs were fixed in 3.8.0 but not listed in this NEWS
3074 254088 Valgrind should know about UD2 instruction
3077 304754 NEWS blows TeX's little mind
3079 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
3083 Release 3.8.0 (10 August 2012)
3084 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3085 3.8.0 is a feature release with many improvements and the usual
3086 collection of bug fixes.
3088 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3089 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
3090 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
3091 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
3092 There is initial support for MacOSX 10.8, but it is not usable for
3093 serious work at present.
3095 * ================== PLATFORM CHANGES =================
3097 * Support for MIPS32 platforms running Linux. Valgrind has been
3098 tested on MIPS32 and MIPS32r2 platforms running different Debian
3099 Squeeze and MeeGo distributions. Both little-endian and big-endian
3100 cores are supported. The tools Memcheck, Massif and Lackey have
3101 been tested and are known to work. See README.mips for more details.
3103 * Preliminary support for Android running on x86.
3105 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
3107 * Support for Intel AVX instructions and for AES instructions. This
3108 support is available only for 64 bit code.
3110 * Support for POWER Decimal Floating Point instructions.
3112 * ==================== TOOL CHANGES ====================
3114 * Non-libc malloc implementations are now supported. This is useful
3115 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
3116 Using the new option --soname-synonyms, such tools can be informed
3117 that the malloc implementation is either linked statically into the
3118 executable, or is present in some other shared library different
3119 from libc.so. This makes it possible to process statically linked
3120 programs, and programs using other malloc libraries, for example
3121 TCMalloc or JEMalloc.
3123 * For tools that provide their own replacement for malloc et al, the
3124 option --redzone-size=<number> allows users to specify the size of
3125 the padding blocks (redzones) added before and after each client
3126 allocated block. Smaller redzones decrease the memory needed by
3127 Valgrind. Bigger redzones increase the chance to detect blocks
3128 overrun or underrun. Prior to this change, the redzone size was
3129 hardwired to 16 bytes in Memcheck.
3133 - The leak_check GDB server monitor command now can
3134 control the maximum nr of loss records to output.
3136 - Reduction of memory use for applications allocating
3137 many blocks and/or having many partially defined bytes.
3139 - Addition of GDB server monitor command 'block_list' that lists
3140 the addresses/sizes of the blocks of a leak search loss record.
3142 - Addition of GDB server monitor command 'who_points_at' that lists
3143 the locations pointing at a block.
3145 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
3146 detect an invalid access of these redzones, by marking them
3147 noaccess. Similarly, if a redzone size is given for a memory
3148 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
3149 This still allows to find some bugs if the user has forgotten to
3150 mark the pool superblock noaccess.
3152 - Performance of memory leak check has been improved, especially in
3153 cases where there are many leaked blocks and/or many suppression
3154 rules used to suppress leak reports.
3156 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
3157 more precise analysis, which is important for LLVM/Clang
3158 generated code. This is at the cost of somewhat reduced
3159 performance. Note there is no change to analysis precision or
3160 costs on Linux targets.
3164 - Added even more facilities that can help finding the cause of a data
3165 race, namely the command-line option --ptrace-addr and the macro
3166 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
3168 - Fixed a subtle bug that could cause false positive data race reports.
3170 * ==================== OTHER CHANGES ====================
3172 * The C++ demangler has been updated so as to work well with C++
3173 compiled by up to at least g++ 4.6.
3175 * Tool developers can make replacement/wrapping more flexible thanks
3176 to the new option --soname-synonyms. This was reported above, but
3177 in fact is very general and applies to all function
3178 replacement/wrapping, not just to malloc-family functions.
3180 * Round-robin scheduling of threads can be selected, using the new
3181 option --fair-sched= yes. Prior to this change, the pipe-based
3182 thread serialisation mechanism (which is still the default) could
3183 give very unfair scheduling. --fair-sched=yes improves
3184 responsiveness of interactive multithreaded applications, and
3185 improves repeatability of results from the thread checkers Helgrind
3188 * For tool developers: support to run Valgrind on Valgrind has been
3189 improved. We can now routinely Valgrind on Helgrind or Memcheck.
3191 * gdbserver now shows the float shadow registers as integer
3192 rather than float values, as the shadow values are mostly
3193 used as bit patterns.
3195 * Increased limit for the --num-callers command line flag to 500.
3197 * Performance improvements for error matching when there are many
3198 suppression records in use.
3200 * Improved support for DWARF4 debugging information (bug 284184).
3202 * Initial support for DWZ compressed Dwarf debug info.
3204 * Improved control over the IR optimiser's handling of the tradeoff
3205 between performance and precision of exceptions. Specifically,
3206 --vex-iropt-precise-memory-exns has been removed and replaced by
3207 --vex-iropt-register-updates, with extended functionality. This
3208 allows the Valgrind gdbserver to always show up to date register
3211 * Modest performance gains through the use of translation chaining for
3214 * ==================== FIXED BUGS ====================
3216 The following bugs have been fixed or resolved. Note that "n-i-bz"
3217 stands for "not in bugzilla" -- that is, a bug that was reported to us
3218 but never got a bugzilla entry. We encourage you to file bugs in
3219 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3220 than mailing the developers (or mailing lists) directly -- bugs that
3221 are not entered into bugzilla tend to get forgotten about or ignored.
3223 To see details of a given bug, visit
3224 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3225 where XXXXXX is the bug number as listed below.
3227 197914 Building valgrind from svn now requires automake-1.10
3228 203877 increase to 16Mb maximum allowed alignment for memalign et al
3229 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
3230 247386 make perf does not run all performance tests
3231 270006 Valgrind scheduler unfair
3232 270777 Adding MIPS/Linux port to Valgrind
3233 270796 s390x: Removed broken support for the TS insn
3234 271438 Fix configure for proper SSE4.2 detection
3235 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
3236 273475 Add support for AVX instructions
3237 274078 improved configure logic for mpicc
3238 276993 fix mremap 'no thrash checks'
3239 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
3240 281482 memcheck incorrect byte allocation count in realloc() for silly argument
3241 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
3242 283413 Fix wrong sanity check
3243 283671 Robustize alignment computation in LibVEX_Alloc
3244 283961 Adding support for some HCI IOCTLs
3245 284124 parse_type_DIE: confused by: DWARF 4
3246 284864 == 273475 (Add support for AVX instructions)
3247 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
3248 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
3249 285725 == 273475 (Add support for AVX instructions)
3250 286261 add wrapper for linux I2C_RDWR ioctl
3251 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
3252 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
3253 286384 configure fails "checking for a supported version of gcc"
3254 286497 == 273475 (Add support for AVX instructions)
3255 286596 == 273475 (Add support for AVX instructions)
3256 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
3257 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
3258 287260 Incorrect conditional jump or move depends on uninitialised value(s)
3259 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
3260 287307 == 273475 (Add support for AVX instructions)
3261 287858 VG_(strerror): unknown error
3262 288298 (MacOSX) unhandled syscall shm_unlink
3263 288995 == 273475 (Add support for AVX instructions)
3264 289470 Loading of large Mach-O thin binaries fails.
3265 289656 == 273475 (Add support for AVX instructions)
3266 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
3267 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
3268 289839 s390x: Provide support for unicode conversion instructions
3269 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
3270 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
3271 290655 Add support for AESKEYGENASSIST instruction
3272 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
3273 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
3274 291253 ES register not initialised in valgrind simulation
3275 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
3276 291865 s390x: Support the "Compare Double and Swap" family of instructions
3277 292300 == 273475 (Add support for AVX instructions)
3278 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
3279 292493 == 273475 (Add support for AVX instructions)
3280 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
3281 292627 Missing support for some SCSI ioctls
3282 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
3283 292841 == 273475 (Add support for AVX instructions)
3284 292993 implement the getcpu syscall on amd64-linux
3285 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
3286 293088 Add some VEX sanity checks for ppc64 unhandled instructions
3287 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
3288 293754 PCMPxSTRx not implemented for 16-bit characters
3289 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
3290 293808 CLFLUSH not supported by latest VEX for amd64
3291 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
3292 294048 MPSADBW instruction not implemented
3293 294055 regtest none/tests/shell fails when locale is not set to C
3294 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
3295 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
3296 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
3297 294260 disInstr_AMD64: disInstr miscalculated next %rip
3298 294523 --partial-loads-ok=yes causes false negatives
3299 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
3300 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
3301 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
3302 295089 can not annotate source for both helgrind and drd
3303 295221 POWER Processor decimal floating point instruction support missing
3304 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
3305 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
3306 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
3307 295617 ARM - Add some missing syscalls
3308 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
3309 296229 Linux user input device ioctls missing wrappers
3310 296318 ELF Debug info improvements (more than one rx/rw mapping)
3311 296422 Add translation chaining support
3312 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
3313 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
3314 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
3315 297078 gdbserver signal handling problems [..]
3316 297147 drd false positives on newly allocated memory
3317 297329 disallow decoding of IBM Power DFP insns on some machines
3318 297497 POWER Processor decimal floating point instruction support missing
3319 297701 Another alias for strncasecmp_l in libc-2.13.so
3320 297911 'invalid write' not reported when using APIs for custom mem allocators.
3321 297976 s390x: revisit EX implementation
3322 297991 Valgrind interferes with mmap()+ftell()
3323 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
3324 297993 Fix compilation of valgrind with gcc -g3.
3325 298080 POWER Processor DFP support missing, part 3
3326 298227 == 273475 (Add support for AVX instructions)
3327 298335 == 273475 (Add support for AVX instructions)
3328 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
3329 298394 s390x: Don't bail out on an unknown machine model. [..]
3330 298421 accept4() syscall (366) support is missing for ARM
3331 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
3332 298732 valgrind installation problem in ubuntu with kernel version 3.x
3333 298862 POWER Processor DFP instruction support missing, part 4
3334 298864 DWARF reader mis-parses DW_FORM_ref_addr
3335 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
3336 299053 Support DWARF4 DW_AT_high_pc constant form
3337 299104 == 273475 (Add support for AVX instructions)
3338 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
3339 299629 dup3() syscall (358) support is missing for ARM
3340 299694 POWER Processor DFP instruction support missing, part 5
3341 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
3342 299803 == 273475 (Add support for AVX instructions)
3343 299804 == 273475 (Add support for AVX instructions)
3344 299805 == 273475 (Add support for AVX instructions)
3345 300140 ARM - Missing (T1) SMMUL
3346 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
3347 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
3348 300414 FCOM and FCOMP unimplemented for amd64 guest
3349 301204 infinite loop in canonicaliseSymtab with ifunc symbol
3350 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
3351 301265 add x86 support to Android build
3352 301984 configure script doesn't detect certain versions of clang
3353 302205 Fix compiler warnings for POWER VEX code and POWER test cases
3354 302287 Unhandled movbe instruction on Atom processors
3355 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
3356 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
3357 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
3358 302656 == 273475 (Add support for AVX instructions)
3359 302709 valgrind for ARM needs extra tls support for android emulator [..]
3360 302827 add wrapper for CDROM_GET_CAPABILITY
3361 302901 Valgrind crashes with dwz optimized debuginfo
3362 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
3363 303116 Add support for the POWER instruction popcntb
3364 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
3365 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
3366 303466 == 273475 (Add support for AVX instructions)
3367 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
3368 303963 strstr() function produces wrong results under valgrind callgrind
3369 304054 CALL_FN_xx macros need to enforce stack alignment
3370 304561 tee system call not supported
3371 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
3372 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
3373 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
3374 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
3375 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
3376 n-i-bz s390x: Shadow registers can now be examined using vgdb
3378 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
3379 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
3383 Release 3.7.0 (5 November 2011)
3384 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3385 3.7.0 is a feature release with many significant improvements and the
3386 usual collection of bug fixes.
3388 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3389 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
3390 Support for recent distros and toolchain components (glibc 2.14, gcc
3391 4.6, MacOSX 10.7) has been added.
3393 * ================== PLATFORM CHANGES =================
3395 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
3396 analyse 64-bit programs running on z/Architecture. Most user space
3397 instructions up to and including z10 are supported. Valgrind has
3398 been tested extensively on z9, z10, and z196 machines running SLES
3399 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
3400 known to work well. Callgrind, Helgrind, and DRD work reasonably
3401 well on z9 and later models. See README.s390 for more details.
3403 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
3404 64-bit processes are supported. Some complex threaded applications
3405 (Firefox) are observed to hang when run as 32 bit applications,
3406 whereas 64-bit versions run OK. The cause is unknown. Memcheck
3407 will likely report some false errors. In general, expect some rough
3408 spots. This release also supports MacOSX 10.6, but drops support
3411 * Preliminary support for Android (on ARM). Valgrind can now run
3412 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
3413 README.android for more details, plus instructions on how to get
3416 * Support for the IBM Power ISA 2.06 (Power7 instructions)
3418 * General correctness and performance improvements for ARM/Linux, and,
3419 by extension, ARM/Android.
3421 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
3422 instruction set support is under development but is not available in
3425 * Support for AIX5 has been removed.
3427 * ==================== TOOL CHANGES ====================
3429 * Memcheck: some incremental changes:
3431 - reduction of memory use in some circumstances
3433 - improved handling of freed memory, which in some circumstances
3434 can cause detection of use-after-free that would previously have
3437 - fix of a longstanding bug that could cause false negatives (missed
3438 errors) in programs doing vector saturated narrowing instructions.
3440 * Helgrind: performance improvements and major memory use reductions,
3441 particularly for large, long running applications which perform many
3442 synchronisation (lock, unlock, etc) events. Plus many smaller
3445 - display of locksets for both threads involved in a race
3447 - general improvements in formatting/clarity of error messages
3449 - addition of facilities and documentation regarding annotation
3450 of thread safe reference counted C++ classes
3452 - new flag --check-stack-refs=no|yes [yes], to disable race checking
3453 on thread stacks (a performance hack)
3455 - new flag --free-is-write=no|yes [no], to enable detection of races
3456 where one thread accesses heap memory but another one frees it,
3457 without any coordinating synchronisation event
3459 * DRD: enabled XML output; added support for delayed thread deletion
3460 in order to detect races that occur close to the end of a thread
3461 (--join-list-vol); fixed a memory leak triggered by repeated client
3462 memory allocatation and deallocation; improved Darwin support.
3464 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
3466 * exp-sgcheck: this tool has been reduced in scope so as to improve
3467 performance and remove checking that Memcheck does better.
3468 Specifically, the ability to check for overruns for stack and global
3469 arrays is unchanged, but the ability to check for overruns of heap
3470 blocks has been removed. The tool has accordingly been renamed to
3471 exp-sgcheck ("Stack and Global Array Checking").
3473 * ==================== OTHER CHANGES ====================
3475 * GDB server: Valgrind now has an embedded GDB server. That means it
3476 is possible to control a Valgrind run from GDB, doing all the usual
3477 things that GDB can do (single stepping, breakpoints, examining
3478 data, etc). Tool-specific functionality is also available. For
3479 example, it is possible to query the definedness state of variables
3480 or memory from within GDB when running Memcheck; arbitrarily large
3481 memory watchpoints are supported, etc. To use the GDB server, start
3482 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3485 * Improved support for unfriendly self-modifying code: a new option
3486 --smc-check=all-non-file is available. This adds the relevant
3487 consistency checks only to code that originates in non-file-backed
3488 mappings. In effect this confines the consistency checking only to
3489 code that is or might be JIT generated, and avoids checks on code
3490 that must have been compiled ahead of time. This significantly
3491 improves performance on applications that generate code at run time.
3493 * It is now possible to build a working Valgrind using Clang-2.9 on
3496 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3497 These enable and disable error reporting on a per-thread, and
3498 nestable, basis. This is useful for hiding errors in particularly
3499 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3500 now uses this facility.
3502 * Added the --mod-funcname option to cg_diff.
3504 * ==================== FIXED BUGS ====================
3506 The following bugs have been fixed or resolved. Note that "n-i-bz"
3507 stands for "not in bugzilla" -- that is, a bug that was reported to us
3508 but never got a bugzilla entry. We encourage you to file bugs in
3509 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3510 mailing the developers (or mailing lists) directly -- bugs that are
3511 not entered into bugzilla tend to get forgotten about or ignored.
3513 To see details of a given bug, visit
3514 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3515 where XXXXXX is the bug number as listed below.
3517 79311 malloc silly arg warning does not give stack trace
3518 210935 port valgrind.h (not valgrind) to win32 to support client requests
3519 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3520 243404 Port to zSeries
3521 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3522 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3523 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3524 253206 Some fixes for the faultstatus testcase
3525 255223 capget testcase fails when running as root
3526 256703 xlc_dbl_u32.c testcase broken
3527 256726 Helgrind tests have broken inline asm
3528 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3529 264800 testcase compile failure on zseries
3530 265762 make public VEX headers compilable by G++ 3.x
3531 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3532 266753 configure script does not give the user the option to not use QtCore
3533 266931 gen_insn_test.pl is broken
3534 266961 ld-linux.so.2 i?86-linux strlen issues
3535 266990 setns instruction causes false positive
3536 267020 Make directory for temporary files configurable at run-time.
3537 267342 == 267997 (segmentation fault on Mac OS 10.6)
3538 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3539 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3540 267488 regtest: darwin support for 64-bit build
3541 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3542 267630 Add support for IBM Power ISA 2.06 -- stage 1
3543 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3544 267819 Add client request for informing the core about reallocation
3545 267925 laog data structure quadratic for a single sequence of lock
3546 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3547 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3548 268513 missed optimizations in fold_Expr
3549 268619 s390x: fpr - gpr transfer facility
3550 268620 s390x: reconsider "long displacement" requirement
3551 268621 s390x: improve IR generation for XC
3552 268715 s390x: FLOGR is not universally available
3553 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3554 268930 s390x: MHY is not universally available
3555 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3556 269079 Support ptrace system call on ARM
3557 269144 missing "Bad option" error message
3558 269209 conditional load and store facility (z196)
3559 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3560 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3561 269736 s390x: minor code generation tweaks
3562 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3563 269863 s390x: remove unused function parameters
3564 269864 s390x: tweak s390_emit_load_cc
3565 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3566 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3567 270115 s390x: rewrite some testcases
3568 270309 == 267997 (valgrind crash on startup)
3569 270320 add support for Linux FIOQSIZE ioctl() call
3570 270326 segfault while trying to sanitize the environment passed to execle
3571 270794 IBM POWER7 support patch causes regression in none/tests
3572 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3573 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3574 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3575 270959 s390x: invalid use of R0 as base register
3576 271042 VSX configure check fails when it should not
3577 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3578 271259 s390x: fix code confusion
3579 271337 == 267997 (Valgrind segfaults on MacOS X)
3580 271385 s390x: Implement Ist_MBE
3581 271501 s390x: misc cleanups
3582 271504 s390x: promote likely and unlikely
3583 271579 ppc: using wrong enum type
3584 271615 unhandled instruction "popcnt" (arch=amd10h)
3585 271730 Fix bug when checking ioctls: duplicate check
3586 271776 s390x: provide STFLE instruction support
3587 271779 s390x: provide clock instructions like STCK
3588 271799 Darwin: ioctls without an arg report a memory error
3589 271820 arm: fix type confusion
3590 271917 pthread_cond_timedwait failure leads to not-locked false positive
3591 272067 s390x: fix DISP20 macro
3592 272615 A typo in debug output in mc_leakcheck.c
3593 272661 callgrind_annotate chokes when run from paths containing regex chars
3594 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3595 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3596 272967 make documentation build-system more robust
3597 272986 Fix gcc-4.6 warnings with valgrind.h
3598 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3599 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3600 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3601 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3602 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3603 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3604 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3605 273778 exp-ptrcheck: unhandled sysno == 259
3606 274089 exp-ptrcheck: unhandled sysno == 208
3607 274378 s390x: Various dispatcher tweaks
3608 274447 WARNING: unhandled syscall: 340
3609 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3610 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3611 274926 valgrind does not build against linux-3
3612 275148 configure FAIL with glibc-2.14
3613 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3614 275168 Make Valgrind work for MacOSX 10.7 Lion
3615 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3616 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3617 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3618 275308 Fix implementation for ppc64 fres instruc
3619 275339 s390x: fix testcase compile warnings
3620 275517 s390x: Provide support for CKSM instruction
3621 275710 s390x: get rid of redundant address mode calculation
3622 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3623 275852 == 250101 (valgrind uses all swap space and is killed)
3624 276784 Add support for IBM Power ISA 2.06 -- stage 3
3625 276987 gdbsrv: fix tests following recent commits
3626 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3627 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3628 277471 Unhandled syscall: 340
3629 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3630 277653 ARM: support Thumb2 PLD instruction
3631 277663 ARM: NEON float VMUL by scalar incorrect
3632 277689 ARM: tests for VSTn with register post-index are broken
3633 277694 ARM: BLX LR instruction broken in ARM mode
3634 277780 ARM: VMOV.F32 (immediate) instruction is broken
3635 278057 fuse filesystem syscall deadlocks
3636 278078 Unimplemented syscall 280 on ppc32
3637 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3638 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3639 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3640 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3641 279027 Support for MVCL and CLCL instruction
3642 279027 s390x: Provide support for CLCL and MVCL instructions
3643 279062 Remove a redundant check in the insn selector for ppc.
3644 279071 JDK creates PTEST with redundant REX.W prefix
3645 279212 gdbsrv: add monitor cmd v.info scheduler.
3646 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3647 279698 memcheck discards valid-bits for packuswb
3648 279795 memcheck reports uninitialised values for mincore on amd64
3649 279994 Add support for IBM Power ISA 2.06 -- stage 3
3650 280083 mempolicy syscall check errors
3651 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3652 280710 s390x: config files for nightly builds
3653 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3654 280965 Valgrind breaks fcntl locks when program does mmap
3655 281138 WARNING: unhandled syscall: 340
3656 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3657 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3658 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3659 281468 s390x: handle do_clone and gcc clones in call traces
3660 281488 ARM: VFP register corruption
3661 281828 == 275284 (false memmove warning: "Source and destination overlap")
3662 281883 s390x: Fix system call wrapper for "clone".
3663 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3664 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3665 282238 SLES10: make check fails
3666 282979 strcasestr needs replacement with recent(>=2.12) glibc
3667 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3668 283243 Regression in ppc64 memcheck tests
3669 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3670 283427 re-connect epoll_pwait syscall on ARM linux
3671 283600 gdbsrv: android: port vgdb.c
3672 283709 none/tests/faultstatus needs to account for page size
3673 284305 filter_gdb needs enhancement to work on ppc64
3674 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3675 284472 Thumb2 ROR.W encoding T2 not implemented
3676 284621 XML-escape process command line in XML output
3677 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3678 that have non-power-of-2 sizes (also AMDs)
3679 n-i-bz don't be spooked by libraries mashed by elfhack
3680 n-i-bz don't be spooked by libxul.so linked with gold
3681 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3683 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3684 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3685 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3689 Release 3.6.1 (16 February 2011)
3690 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3691 3.6.1 is a bug fix release. It adds support for some SSE4
3692 instructions that were omitted in 3.6.0 due to lack of time. Initial
3693 support for glibc-2.13 has been added. A number of bugs causing
3694 crashing or assertion failures have been fixed.
3696 The following bugs have been fixed or resolved. Note that "n-i-bz"
3697 stands for "not in bugzilla" -- that is, a bug that was reported to us
3698 but never got a bugzilla entry. We encourage you to file bugs in
3699 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3700 mailing the developers (or mailing lists) directly -- bugs that are
3701 not entered into bugzilla tend to get forgotten about or ignored.
3703 To see details of a given bug, visit
3704 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3705 where XXXXXX is the bug number as listed below.
3707 188572 Valgrind on Mac should suppress setenv() mem leak
3708 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3709 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3710 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3711 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3712 254420 memory pool tracking broken
3713 254957 Test code failing to compile due to changes in memcheck.h
3714 255009 helgrind/drd: crash on chmod with invalid parameter
3715 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3716 255355 helgrind/drd: crash on threaded programs doing fork
3718 255418 (SSE4.x) rint call compiled with ICC
3719 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3720 255888 closing valgrindoutput tag outputted to log-stream on error
3721 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3722 255966 Slowness when using mempool annotations
3723 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3724 256600 super-optimized strcasecmp() false positive
3725 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3726 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3727 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3728 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3729 257276 Missing case in memcheck --track-origins=yes
3730 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3731 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3732 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3733 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3734 263099 callgrind_annotate counts Ir improperly [...]
3735 263877 undefined coprocessor instruction on ARMv7
3736 265964 configure FAIL with glibc-2.13
3737 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3738 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3739 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3740 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3741 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3743 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3747 Release 3.6.0 (21 October 2010)
3748 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3749 3.6.0 is a feature release with many significant improvements and the
3750 usual collection of bug fixes.
3752 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3753 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3754 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3756 -------------------------
3758 Here are some highlights. Details are shown further down:
3760 * Support for ARM/Linux.
3762 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3764 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3766 * Support for the SSE4.2 instruction set.
3768 * Enhancements to the Callgrind profiler, including the ability to
3769 handle CPUs with three levels of cache.
3771 * A new experimental heap profiler, DHAT.
3773 * A huge number of bug fixes and small enhancements.
3775 -------------------------
3777 Here are details of the above changes, together with descriptions of
3778 many other changes, and a list of fixed bugs.
3780 * ================== PLATFORM CHANGES =================
3782 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3783 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3784 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3786 This requires a CPU capable of running the ARMv7-A instruction set
3787 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3788 of the user space instruction set, including ARM and Thumb integer
3789 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3790 Cachegrind and Massif tools work properly; other tools work to
3793 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3794 with support for recent releases of the underlying toolchain
3795 components, notably gcc-4.5 and glibc-2.12.
3797 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3798 support also works much better on OS X 10.5, and is as solid as
3801 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3802 64-bit mode. In 32-bit mode, support is only available up to and
3803 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3804 supported in 64-bit mode, and 32-bit mode does in fact support the
3805 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3806 10.6 on 32-bit targets.
3808 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3809 and including version 2.05 is supported.
3811 * ==================== TOOL CHANGES ====================
3813 * Cachegrind has a new processing script, cg_diff, which finds the
3814 difference between two profiles. It's very useful for evaluating
3815 the performance effects of a change in a program.
3817 Related to this change, the meaning of cg_annotate's (rarely-used)
3818 --threshold option has changed; this is unlikely to affect many
3819 people, if you do use it please see the user manual for details.
3821 * Callgrind now can do branch prediction simulation, similar to
3822 Cachegrind. In addition, it optionally can count the number of
3823 executed global bus events. Both can be used for a better
3824 approximation of a "Cycle Estimation" as derived event (you need to
3825 update the event formula in KCachegrind yourself).
3827 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3828 rather than the L2 cache. This is to accommodate machines with
3829 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3830 cache configuration of such a machine it will run the simulation as
3831 if the L2 cache isn't present. This means the results are less
3832 likely to match the true result for the machine, but
3833 Cachegrind/Callgrind's results are already only approximate, and
3834 should not be considered authoritative. The results are still
3835 useful for giving a general idea about a program's locality.
3837 * Massif has a new option, --pages-as-heap, which is disabled by
3838 default. When enabled, instead of tracking allocations at the level
3839 of heap blocks (as allocated with malloc/new/new[]), it instead
3840 tracks memory allocations at the level of memory pages (as mapped by
3841 mmap, brk, etc). Each mapped page is treated as its own block.
3842 Interpreting the page-level output is harder than the heap-level
3843 output, but this option is useful if you want to account for every
3844 byte of memory used by a program.
3846 * DRD has two new command-line options: --free-is-write and
3847 --trace-alloc. The former allows to detect reading from already freed
3848 memory, and the latter allows tracing of all memory allocations and
3851 * DRD has several new annotations. Custom barrier implementations can
3852 now be annotated, as well as benign races on static variables.
3854 * DRD's happens before / happens after annotations have been made more
3855 powerful, so that they can now also be used to annotate e.g. a smart
3856 pointer implementation.
3858 * Helgrind's annotation set has also been drastically improved, so as
3859 to provide to users a general set of annotations to describe locks,
3860 semaphores, barriers and condition variables. Annotations to
3861 describe thread-safe reference counted heap objects have also been
3864 * Memcheck has a new command-line option, --show-possibly-lost, which
3865 is enabled by default. When disabled, the leak detector will not
3866 show possibly-lost blocks.
3868 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3869 has been added. DHAT keeps track of allocated heap blocks, and also
3870 inspects every memory reference to see which block (if any) is being
3871 accessed. This gives a lot of insight into block lifetimes,
3872 utilisation, turnover, liveness, and the location of hot and cold
3873 fields. You can use DHAT to do hot-field profiling.
3875 * ==================== OTHER CHANGES ====================
3877 * Improved support for unfriendly self-modifying code: the extra
3878 overhead incurred by --smc-check=all has been reduced by
3879 approximately a factor of 5 as compared with 3.5.0.
3881 * Ability to show directory names for source files in error messages.
3882 This is combined with a flexible mechanism for specifying which
3883 parts of the paths should be shown. This is enabled by the new flag
3886 * A new flag, --require-text-symbol, which will stop the run if a
3887 specified symbol is not found it a given shared object when it is
3888 loaded into the process. This makes advanced working with function
3889 intercepting and wrapping safer and more reliable.
3891 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3892 and control of Valgrind is now available for the tools Memcheck and
3893 Helgrind. XML output from Valgrind is available for Memcheck,
3894 Helgrind and exp-Ptrcheck.
3896 * More reliable stack unwinding on amd64-linux, particularly in the
3897 presence of function wrappers, and with gcc-4.5 compiled code.
3899 * Modest scalability (performance improvements) for massive
3900 long-running applications, particularly for those with huge amounts
3903 * Support for analyzing programs running under Wine with has been
3904 improved. The header files <valgrind/valgrind.h>,
3905 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3906 Windows-programs compiled with MinGW or one of the Microsoft Visual
3909 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3910 The 32-bit simulator was not affected. This did not occur often,
3911 but when it did would usually crash the program under test.
3914 * A large number of bugs were fixed. These are shown below.
3916 * A number of bugs were investigated, and were candidates for fixing,
3917 but are not fixed in 3.6.0, due to lack of developer time. They may
3918 get fixed in later releases. They are:
3920 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3921 212419 false positive "lock order violated" (A+B vs A)
3922 213685 Undefined value propagates past dependency breaking instruction
3923 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3924 237920 valgrind segfault on fork failure
3925 242137 support for code compiled by LLVM-2.8
3926 242423 Another unknown Intel cache config value
3927 243232 Inconsistent Lock Orderings report with trylock
3928 243483 ppc: callgrind triggers VEX assertion failure
3929 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3930 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3932 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3933 249435 Analyzing wine programs with callgrind triggers a crash
3934 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3935 250065 Handling large allocations
3936 250101 huge "free" memory usage due to m_mallocfree.c
3937 "superblocks fragmentation"
3938 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3939 252091 Callgrind on ARM does not detect function returns correctly
3940 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3941 254420 memory pool tracking broken
3942 n-i-bz support for adding symbols for JIT generated code
3945 The following bugs have been fixed or resolved. Note that "n-i-bz"
3946 stands for "not in bugzilla" -- that is, a bug that was reported to us
3947 but never got a bugzilla entry. We encourage you to file bugs in
3948 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3949 mailing the developers (or mailing lists) directly -- bugs that are
3950 not entered into bugzilla tend to get forgotten about or ignored.
3952 To see details of a given bug, visit
3953 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3954 where XXXXXX is the bug number as listed below.
3956 135264 dcbzl instruction missing
3958 153699 Valgrind should report unaligned reads with movdqa
3960 190429 Valgrind reports lost of errors in ld.so
3961 with x86_64 2.9.90 glibc
3962 197266 valgrind appears to choke on the xmms instruction
3964 197988 Crash when demangling very large symbol names
3965 202315 unhandled syscall: 332 (inotify_init1)
3966 203256 Add page-level profiling to Massif
3967 205093 dsymutil=yes needs quotes, locking (partial fix)
3968 205241 Snow Leopard 10.6 support (partial fix)
3969 206600 Leak checker fails to upgrade indirect blocks when their
3970 parent becomes reachable
3971 210935 port valgrind.h (not valgrind) to win32 so apps run under
3972 wine can make client requests
3973 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3974 within Linux ip-stack checksum functions
3975 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3977 213685 Undefined value propagates past dependency breaking instruction
3979 215914 Valgrind inserts bogus empty environment variable
3981 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3982 222545 shmat fails under valgind on some arm targets
3983 222560 ARM NEON support
3986 232509 Docs build fails with formatting inside <title></title> elements
3988 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3989 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3990 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3991 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3992 237485 symlink (syscall 57) is not supported on Mac OS
3993 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3995 238208 is_just_below_ESP doesn't take into account red-zone
3996 238345 valgrind passes wrong $0 when executing a shell script
3997 238679 mq_timedreceive syscall doesn't flag the reception buffer
3999 238696 fcntl command F_DUPFD_CLOEXEC not supported
4000 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4001 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4002 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
4004 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
4010 242606 unhandled syscall: setegid (in Ptrcheck)
4011 242814 Helgrind "Impossible has happened" during
4012 QApplication::initInstance();
4013 243064 Valgrind attempting to read debug information from iso
4014 243270 Make stack unwinding in Valgrind wrappers more reliable
4015 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
4016 sysno = 277 (mq_open)
4017 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
4018 244493 ARM VFP d16-d31 registers support
4019 244670 add support for audit_session_self syscall on Mac OS 10.6
4020 244921 The xml report of helgrind tool is not well format
4021 244923 In the xml report file, the <preamble> not escape the
4022 xml char, eg '<','&','>'
4023 245535 print full path names in plain text reports
4024 245925 x86-64 red zone handling problem
4025 246258 Valgrind not catching integer underruns + new [] s
4026 246311 reg/reg cmpxchg doesn't work on amd64
4027 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
4028 246888 Improve Makefile.vex.am
4029 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
4030 to [f]chmod_extended
4031 247526 IBM POWER6 (ISA 2.05) support is incomplete
4032 247561 Some leak testcases fails due to reachable addresses in
4034 247875 sizeofIRType to handle Ity_I128
4035 247894 [PATCH] unhandled syscall sys_readahead
4036 247980 Doesn't honor CFLAGS passed to configure
4037 248373 darwin10.supp is empty in the trunk
4038 248822 Linux FIBMAP ioctl has int parameter instead of long
4039 248893 [PATCH] make readdwarf.c big endianess safe to enable
4040 unwinding on big endian systems
4041 249224 Syscall 336 not supported (SYS_proc_info)
4043 249775 Incorrect scheme for detecting NEON capabilities of host CPU
4044 249943 jni JVM init fails when using valgrind
4045 249991 Valgrind incorrectly declares AESKEYGENASSIST support
4047 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
4048 250799 frexp$fenv_access_off function generates SIGILL
4049 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
4050 251251 support pclmulqdq insn
4051 251362 valgrind: ARM: attach to debugger either fails or provokes
4053 251674 Unhandled syscall 294
4056 254257 Add support for debugfiles found by build-id
4057 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
4058 254646 Wrapped functions cause stack misalignment on OS X
4059 (and possibly Linux)
4060 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
4062 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
4066 Release 3.5.0 (19 August 2009)
4067 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4068 3.5.0 is a feature release with many significant improvements and the
4069 usual collection of bug fixes. The main improvement is that Valgrind
4070 now works on Mac OS X.
4072 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
4073 and X86/Darwin. Support for recent distros and toolchain components
4074 (glibc 2.10, gcc 4.5) has been added.
4076 -------------------------
4078 Here is a short summary of the changes. Details are shown further
4081 * Support for Mac OS X (10.5.x).
4083 * Improvements and simplifications to Memcheck's leak checker.
4085 * Clarification and simplifications in various aspects of Valgrind's
4088 * XML output for Helgrind and Ptrcheck.
4090 * Performance and stability improvements for Helgrind and DRD.
4092 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
4094 * A new experimental tool, BBV, useful for computer architecture
4097 * Improved Wine support, including ability to read Windows PDB
4100 -------------------------
4102 Here are details of the above changes, followed by descriptions of
4103 many other minor changes, and a list of fixed bugs.
4106 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
4107 called "Darwin" because that is the name of the OS core, which is the
4108 level that Valgrind works at.)
4112 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
4113 because it would require work and 10.4 is only becoming less common.
4115 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
4116 fairly well. For 10.5.x, 32-bit programs are the default even on
4117 64-bit machines, so it handles most current programs.
4119 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
4120 officially supported, but simple programs at least will probably work.
4121 However, start-up is slow.
4123 - PowerPC machines are not supported.
4125 Things that don't work:
4127 - The Ptrcheck tool.
4129 - Objective-C garbage collection.
4133 - If you have Rogue Amoeba's "Instant Hijack" program installed,
4134 Valgrind will fail with a SIGTRAP at start-up. See
4135 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
4140 - You will likely find --dsymutil=yes a useful option, as error
4141 messages may be imprecise without it.
4143 - Mac OS X support is new and therefore will be less robust than the
4144 Linux support. Please report any bugs you find.
4146 - Threaded programs may run more slowly than on Linux.
4148 Many thanks to Greg Parker for developing this port over several years.
4151 * Memcheck's leak checker has been improved.
4153 - The results for --leak-check=summary now match the summary results
4154 for --leak-check=full. Previously they could differ because
4155 --leak-check=summary counted "indirectly lost" blocks and
4156 "suppressed" blocks as "definitely lost".
4158 - Blocks that are only reachable via at least one interior-pointer,
4159 but are directly pointed to by a start-pointer, were previously
4160 marked as "still reachable". They are now correctly marked as
4163 - The default value for the --leak-resolution option has been
4164 changed from "low" to "high". In general, this means that more
4165 leak reports will be produced, but each leak report will describe
4166 fewer leaked blocks.
4168 - With --leak-check=full, "definitely lost" and "possibly lost"
4169 leaks are now considered as proper errors, ie. they are counted
4170 for the "ERROR SUMMARY" and affect the behaviour of
4171 --error-exitcode. These leaks are not counted as errors if
4172 --leak-check=summary is specified, however.
4174 - Documentation for the leak checker has been improved.
4177 * Various aspects of Valgrind's text output have changed.
4179 - Valgrind's start-up message has changed. It is shorter but also
4180 includes the command being run, which makes it easier to use
4181 --trace-children=yes. An example:
4183 - Valgrind's shut-down messages have also changed. This is most
4184 noticeable with Memcheck, where the leak summary now occurs before
4185 the error summary. This change was necessary to allow leaks to be
4186 counted as proper errors (see the description of the leak checker
4187 changes above for more details). This was also necessary to fix a
4188 longstanding bug in which uses of suppressions against leaks were
4189 not "counted", leading to difficulties in maintaining suppression
4190 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
4192 - Behavior of -v has changed. In previous versions, -v printed out
4193 a mixture of marginally-user-useful information, and tool/core
4194 statistics. The statistics printing has now been moved to its own
4195 flag, --stats=yes. This means -v is less verbose and more likely
4196 to convey useful end-user information.
4198 - The format of some (non-XML) stack trace entries has changed a
4199 little. Previously there were six possible forms:
4201 0x80483BF: really (a.c:20)
4202 0x80483BF: really (in /foo/a.out)
4204 0x80483BF: (within /foo/a.out)
4205 0x80483BF: ??? (a.c:20)
4208 The third and fourth of these forms have been made more consistent
4209 with the others. The six possible forms are now:
4211 0x80483BF: really (a.c:20)
4212 0x80483BF: really (in /foo/a.out)
4213 0x80483BF: really (in ???)
4214 0x80483BF: ??? (in /foo/a.out)
4215 0x80483BF: ??? (a.c:20)
4218 Stack traces produced when --xml=yes is specified are different
4222 * Helgrind and Ptrcheck now support XML output, so they can be used
4223 from GUI tools. Also, the XML output mechanism has been
4226 - The XML format has been overhauled and generalised, so it is more
4227 suitable for error reporting tools in general. The Memcheck
4228 specific aspects of it have been removed. The new format, which
4229 is an evolution of the old format, is described in
4230 docs/internals/xml-output-protocol4.txt.
4232 - Memcheck has been updated to use the new format.
4234 - Helgrind and Ptrcheck are now able to emit output in this format.
4236 - The XML output mechanism has been overhauled. XML is now output
4237 to its own file descriptor, which means that:
4239 * Valgrind can output text and XML independently.
4241 * The longstanding problem of XML output being corrupted by
4242 unexpected un-tagged text messages is solved.
4244 As before, the destination for text output is specified using
4245 --log-file=, --log-fd= or --log-socket=.
4247 As before, XML output for a tool is enabled using --xml=yes.
4249 Because there's a new XML output channel, the XML output
4250 destination is now specified by --xml-file=, --xml-fd= or
4253 Initial feedback has shown this causes some confusion. To
4254 clarify, the two envisaged usage scenarios are:
4256 (1) Normal text output. In this case, do not specify --xml=yes
4257 nor any of --xml-file=, --xml-fd= or --xml-socket=.
4259 (2) XML output. In this case, specify --xml=yes, and one of
4260 --xml-file=, --xml-fd= or --xml-socket= to select the XML
4261 destination, one of --log-file=, --log-fd= or --log-socket=
4262 to select the destination for any remaining text messages,
4263 and, importantly, -q.
4265 -q makes Valgrind completely silent on the text channel,
4266 except in the case of critical failures, such as Valgrind
4267 itself segfaulting, or failing to read debugging information.
4268 Hence, in this scenario, it suffices to check whether or not
4269 any output appeared on the text channel. If yes, then it is
4270 likely to be a critical error which should be brought to the
4271 attention of the user. If no (the text channel produced no
4272 output) then it can be assumed that the run was successful.
4274 This allows GUIs to make the critical distinction they need to
4275 make (did the run fail or not?) without having to search or
4276 filter the text output channel in any way.
4278 It is also recommended to use --child-silent-after-fork=yes in
4282 * Improvements and changes in Helgrind:
4284 - XML output, as described above
4286 - Checks for consistent association between pthread condition
4287 variables and their associated mutexes are now performed.
4289 - pthread_spinlock functions are supported.
4291 - Modest performance improvements.
4293 - Initial (skeletal) support for describing the behaviour of
4294 non-POSIX synchronisation objects through ThreadSanitizer
4295 compatible ANNOTATE_* macros.
4297 - More controllable tradeoffs between performance and the level of
4298 detail of "previous" accesses in a race. There are now three
4301 * --history-level=full. This is the default, and was also the
4302 default in 3.4.x. It shows both stacks involved in a race, but
4303 requires a lot of memory and can be very slow in programs that
4304 do many inter-thread synchronisation events.
4306 * --history-level=none. This only shows the later stack involved
4307 in a race. This can be much faster than --history-level=full,
4308 but makes it much more difficult to find the other access
4309 involved in the race.
4311 The new intermediate setting is
4313 * --history-level=approx
4315 For the earlier (other) access, two stacks are presented. The
4316 earlier access is guaranteed to be somewhere in between the two
4317 program points denoted by those stacks. This is not as useful
4318 as showing the exact stack for the previous access (as per
4319 --history-level=full), but it is better than nothing, and it's
4320 almost as fast as --history-level=none.
4323 * New features and improvements in DRD:
4325 - The error messages printed by DRD are now easier to interpret.
4326 Instead of using two different numbers to identify each thread
4327 (Valgrind thread ID and DRD thread ID), DRD does now identify
4328 threads via a single number (the DRD thread ID). Furthermore
4329 "first observed at" information is now printed for all error
4330 messages related to synchronization objects.
4332 - Added support for named semaphores (sem_open() and sem_close()).
4334 - Race conditions between pthread_barrier_wait() and
4335 pthread_barrier_destroy() calls are now reported.
4337 - Added support for custom allocators through the macros
4338 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
4339 in <valgrind/valgrind.h>). An alternative for these two macros is
4340 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
4343 - Added support for annotating non-POSIX synchronization objects
4344 through several new ANNOTATE_*() macros.
4346 - OpenMP: added support for the OpenMP runtime (libgomp) included
4347 with gcc versions 4.3.0 and 4.4.0.
4351 - Added two new command-line options (--first-race-only and
4352 --segment-merging-interval).
4355 * Genuinely atomic support for x86/amd64/ppc atomic instructions
4357 Valgrind will now preserve (memory-access) atomicity of LOCK-
4358 prefixed x86/amd64 instructions, and any others implying a global
4359 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
4361 This means that Valgrinded processes will "play nicely" in
4362 situations where communication with other processes, or the kernel,
4363 is done through shared memory and coordinated with such atomic
4364 instructions. Prior to this change, such arrangements usually
4365 resulted in hangs, races or other synchronisation failures, because
4366 Valgrind did not honour atomicity of such instructions.
4369 * A new experimental tool, BBV, has been added. BBV generates basic
4370 block vectors for use with the SimPoint analysis tool, which allows
4371 a program's overall behaviour to be approximated by running only a
4372 fraction of it. This is useful for computer architecture
4373 researchers. You can run BBV by specifying --tool=exp-bbv (the
4374 "exp-" prefix is short for "experimental"). BBV was written by
4378 * Modestly improved support for running Windows applications under
4379 Wine. In particular, initial support for reading Windows .PDB debug
4380 information has been added.
4383 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
4384 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
4388 * The Valgrind client requests VALGRIND_PRINTF and
4389 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
4390 the string was always printed immediately on its own line. Now, the
4391 string will be added to a buffer but not printed until a newline is
4392 encountered, or other Valgrind output is printed (note that for
4393 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
4394 "other Valgrind output"). This allows you to use multiple
4395 VALGRIND_PRINTF calls to build up a single output line, and also to
4396 print multiple output lines with a single request (by embedding
4397 multiple newlines in the string).
4400 * The graphs drawn by Massif's ms_print program have changed slightly:
4402 - The half-height chars '.' and ',' are no longer drawn, because
4403 they are confusing. The --y option can be used if the default
4404 y-resolution is not high enough.
4406 - Horizontal lines are now drawn after the top of a snapshot if
4407 there is a gap until the next snapshot. This makes it clear that
4408 the memory usage has not dropped to zero between snapshots.
4411 * Something that happened in 3.4.0, but wasn't clearly announced: the
4412 option --read-var-info=yes can be used by some tools (Memcheck,
4413 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
4414 variable type and location information. This makes those tools
4415 start up more slowly and increases memory consumption, but
4416 descriptions of data addresses in error messages become more
4420 * exp-Omega, an experimental instantaneous leak-detecting tool, was
4421 disabled in 3.4.0 due to a lack of interest and maintenance,
4422 although the source code was still in the distribution. The source
4423 code has now been removed from the distribution. For anyone
4424 interested, the removal occurred in SVN revision r10247.
4427 * Some changes have been made to the build system.
4429 - VEX/ is now integrated properly into the build system. This means
4430 that dependency tracking within VEX/ now works properly, "make
4431 install" will work without requiring "make" before it, and
4432 parallel builds (ie. 'make -j') now work (previously a
4433 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
4434 was effectively ignored).
4436 - The --with-vex configure option has been removed. It was of
4437 little use and removing it simplified the build system.
4439 - The location of some install files has changed. This should not
4440 affect most users. Those who might be affected:
4442 * For people who use Valgrind with MPI programs, the installed
4443 libmpiwrap.so library has moved from
4444 $(INSTALL)/<platform>/libmpiwrap.so to
4445 $(INSTALL)/libmpiwrap-<platform>.so.
4447 * For people who distribute standalone Valgrind tools, the
4448 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
4449 have moved to $(INSTALL)/libcoregrind-<platform>.a.
4451 These changes simplify the build system.
4453 - Previously, all the distributed suppression (*.supp) files were
4454 installed. Now, only default.supp is installed. This should not
4455 affect users as the other installed suppression files were not
4456 read; the fact that they were installed was a mistake.
4459 * KNOWN LIMITATIONS:
4461 - Memcheck is unusable with the Intel compiler suite version 11.1,
4462 when it generates code for SSE2-and-above capable targets. This
4463 is because of icc's use of highly optimised inlined strlen
4464 implementations. It causes Memcheck to report huge numbers of
4465 false errors even in simple programs. Helgrind and DRD may also
4468 Versions 11.0 and earlier may be OK, but this has not been
4472 The following bugs have been fixed or resolved. Note that "n-i-bz"
4473 stands for "not in bugzilla" -- that is, a bug that was reported to us
4474 but never got a bugzilla entry. We encourage you to file bugs in
4475 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4476 mailing the developers (or mailing lists) directly -- bugs that are
4477 not entered into bugzilla tend to get forgotten about or ignored.
4479 To see details of a given bug, visit
4480 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4481 where XXXXXX is the bug number as listed below.
4483 84303 How about a LockCheck tool?
4484 91633 dereference of null ptr in vgPlain_st_basetype
4485 97452 Valgrind doesn't report any pthreads problems
4486 100628 leak-check gets assertion failure when using
4487 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4488 108528 NPTL pthread cleanup handlers not called
4489 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4490 110128 mallinfo is not implemented...
4491 110770 VEX: Generated files not always updated when making valgrind
4492 111102 Memcheck: problems with large (memory footprint) applications
4493 115673 Vex's decoder should never assert
4494 117564 False positive: Syscall param clone(child_tidptr) contains
4495 uninitialised byte(s)
4496 119404 executing ssh from inside valgrind fails
4497 133679 Callgrind does not write path names to sources with dwarf debug
4499 135847 configure.in problem with non gnu compilers (and possible fix)
4500 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4501 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4502 136230 memcheck reports "possibly lost", should be "still reachable"
4503 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4504 137904 Valgrind reports a memory leak when using POSIX threads,
4506 139076 valgrind VT_GETSTATE error
4507 142228 complaint of elf_dynamic_do_rela in trivial usage
4508 145347 spurious warning with USBDEVFS_REAPURB
4509 148441 (wine) can't find memory leak in Wine, win32 binary
4511 148742 Leak-check fails assert on exit
4512 149878 add (proper) check for calloc integer overflow
4513 150606 Call graph is broken when using callgrind control
4514 152393 leak errors produce an exit code of 0. I need some way to
4515 cause leak errors to result in a nonzero exit code.
4516 157154 documentation (leak-resolution doc speaks about num-callers
4517 def=4) + what is a loss record
4518 159501 incorrect handling of ALSA ioctls
4519 162020 Valgrinding an empty/zero-byte file crashes valgrind
4520 162482 ppc: Valgrind crashes while reading stabs information
4521 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4522 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4523 163560 VEX/test_main.c is missing from valgrind-3.3.1
4524 164353 malloc_usable_size() doesn't return a usable size
4525 165468 Inconsistent formatting in memcheck manual -- please fix
4526 169505 main.c:286 (endOfInstr):
4527 Assertion 'ii->cost_offset == *cost_offset' failed
4528 177206 Generate default.supp during compile instead of configure
4529 177209 Configure valt_load_address based on arch+os
4530 177305 eventfd / syscall 323 patch lost
4531 179731 Tests fail to build because of inlining of non-local asm labels
4532 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4533 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4534 181594 Bogus warning for empty text segment
4535 181707 dwarf doesn't require enumerations to have name
4536 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4537 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4538 Assertion '!already_present' failed.
4539 185359 exp-ptrcheck: unhandled syscall getresuid()
4540 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4541 185816 Valgrind is unable to handle debug info for files with split
4542 debug info that are prelinked afterwards
4543 185980 [darwin] unhandled syscall: sem_open
4544 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4545 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4546 186790 Suppression pattern used for leaks are not reported
4547 186796 Symbols with length>200 in suppression files are ignored
4548 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4549 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4550 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4551 188046 bashisms in the configure script
4552 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4553 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4554 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4555 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4556 assertion fail "!lock->heldBy"
4557 188427 Add support for epoll_create1 (with patch)
4558 188530 Support for SIOCGSTAMPNS
4559 188560 Include valgrind.spec in the tarball
4560 188572 Valgrind on Mac should suppress setenv() mem leak
4561 189054 Valgrind fails to build because of duplicate non-local asm labels
4562 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4563 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4564 189763 drd assertion failure: s_threadinfo[tid].is_recording
4565 190219 unhandled syscall: 328 (x86-linux)
4566 190391 dup of 181394; see above
4567 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4568 190820 No debug information on powerpc-linux
4569 191095 PATCH: Improve usbdevfs ioctl handling
4570 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4572 191189 --xml=yes should obey --gen-suppressions=all
4573 191192 syslog() needs a suppression on macosx
4574 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4575 191761 getrlimit on MacOSX
4576 191992 multiple --fn-skip only works sometimes; dependent on order
4577 192634 V. reports "aspacem sync_check_mapping_callback:
4578 segment mismatch" on Darwin
4579 192954 __extension__ missing on 2 client requests
4580 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4581 194474 "INSTALL" file has different build instructions than "README"
4582 194671 Unhandled syscall (sem_wait?) from mac valgrind
4583 195069 memcheck: reports leak (memory still reachable) for
4585 195169 drd: (vgDrd_barrier_post_wait):
4586 Assertion 'r->sg[p->post_iteration]' failed.
4587 195268 valgrind --log-file doesn't accept ~/...
4588 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4589 195860 WARNING: unhandled syscall: unix:223
4590 196528 need a error suppression for pthread_rwlock_init under os x?
4591 197227 Support aio_* syscalls on Darwin
4592 197456 valgrind should reject --suppressions=(directory)
4593 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4594 197591 unhandled syscall 27 (mincore)
4595 197793 Merge DCAS branch to the trunk == 85756, 142103
4596 197794 Avoid duplicate filenames in Vex
4597 197898 make check fails on current SVN
4598 197901 make check fails also under exp-ptrcheck in current SVN
4599 197929 Make --leak-resolution=high the default
4600 197930 Reduce spacing between leak reports
4601 197933 Print command line of client at start-up, and shorten preamble
4602 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4603 198395 add BBV to the distribution as an experimental tool
4604 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4605 198649 callgrind_annotate doesn't cumulate counters
4606 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4607 199977 Valgrind complains about an unrecognized instruction in the
4608 atomic_incs test program
4609 200029 valgrind isn't able to read Fedora 12 debuginfo
4610 200760 darwin unhandled syscall: unix:284
4611 200827 DRD doesn't work on Mac OS X
4612 200990 VG_(read_millisecond_timer)() does not work correctly
4613 201016 Valgrind does not support pthread_kill() on Mac OS
4614 201169 Document --read-var-info
4615 201323 Pre-3.5.0 performance sanity checking
4616 201384 Review user manual for the 3.5.0 release
4617 201585 mfpvr not implemented on ppc
4618 201708 tests failing because x86 direction flag is left set
4619 201757 Valgrind doesn't handle any recent sys_futex additions
4620 204377 64-bit valgrind can not start a shell script
4621 (with #!/path/to/shell) if the shell is a 32-bit executable
4622 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4623 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4624 about memory allocation events with memory access tracing enabled
4625 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4627 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4631 Release 3.4.1 (28 February 2009)
4632 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4633 3.4.1 is a bug-fix release that fixes some regressions and assertion
4634 failures in debug info reading in 3.4.0, most notably incorrect stack
4635 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4636 other debug info problems are also fixed. A number of bugs in the
4637 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4639 In view of the fact that 3.4.0 contains user-visible regressions
4640 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4641 encouraged to ship 3.4.1 in preference to 3.4.0.
4643 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4644 bugzilla" -- that is, a bug that was reported to us but never got a
4645 bugzilla entry. We encourage you to file bugs in bugzilla
4646 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4647 developers (or mailing lists) directly -- bugs that are not entered
4648 into bugzilla tend to get forgotten about or ignored.
4650 n-i-bz Fix various bugs reading icc-11 generated debug info
4651 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4652 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4653 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4654 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4655 179618 exp-ptrcheck crashed / exit prematurely
4656 179624 helgrind: false positive races with pthread_create and
4657 recv/open/close/read
4658 134207 pkg-config output contains @VG_PLATFORM@
4659 176926 floating point exception at valgrind startup with PPC 440EPX
4660 181594 Bogus warning for empty text segment
4661 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4662 181707 Dwarf3 doesn't require enumerations to have name
4663 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4664 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4665 Assertion '!already_present' failed.
4666 185359 exp-ptrcheck unhandled syscall getresuid()
4668 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4669 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4673 Release 3.4.0 (2 January 2009)
4674 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4675 3.4.0 is a feature release with many significant improvements and the
4676 usual collection of bug fixes. This release supports X86/Linux,
4677 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4678 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4680 3.4.0 brings some significant tool improvements. Memcheck can now
4681 report the origin of uninitialised values, the thread checkers
4682 Helgrind and DRD are much improved, and we have a new experimental
4683 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4684 global arrays. In detail:
4686 * Memcheck is now able to track the origin of uninitialised values.
4687 When it reports an uninitialised value error, it will try to show
4688 the origin of the value, as either a heap or stack allocation.
4689 Origin tracking is expensive and so is not enabled by default. To
4690 use it, specify --track-origins=yes. Memcheck's speed will be
4691 essentially halved, and memory usage will be significantly
4692 increased. Nevertheless it can drastically reduce the effort
4693 required to identify the root cause of uninitialised value errors,
4694 and so is often a programmer productivity win, despite running more
4697 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4698 3.4.0, will be released shortly.
4700 * Helgrind's race detection algorithm has been completely redesigned
4701 and reimplemented, to address usability and scalability concerns:
4703 - The new algorithm has a lower false-error rate: it is much less
4704 likely to report races that do not really exist.
4706 - Helgrind will display full call stacks for both accesses involved
4707 in a race. This makes it easier to identify the root causes of
4710 - Limitations on the size of program that can run have been removed.
4712 - Performance has been modestly improved, although that is very
4715 - Direct support for Qt4 threading has been added.
4717 - pthread_barriers are now directly supported.
4719 - Helgrind works well on all supported Linux targets.
4721 * The DRD thread debugging tool has seen major improvements:
4723 - Greatly improved performance and significantly reduced memory
4726 - Support for several major threading libraries (Boost.Thread, Qt4,
4727 glib, OpenMP) has been added.
4729 - Support for atomic instructions, POSIX semaphores, barriers and
4730 reader-writer locks has been added.
4732 - Works now on PowerPC CPUs too.
4734 - Added support for printing thread stack usage at thread exit time.
4736 - Added support for debugging lock contention.
4738 - Added a manual for Drd.
4740 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4741 checks for misuses of pointers. In that sense it is a bit like
4742 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4743 detect overruns of stack and global arrays, it can detect
4744 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4745 detect accesses heap blocks that have been freed a very long time
4746 ago (millions of blocks in the past).
4748 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4749 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4750 of the main Valgrind documentation. As this is an experimental
4751 tool, we would be particularly interested in hearing about your
4752 experiences with it.
4754 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4755 longer built by default, although the code remains in the repository
4756 and the tarball. This is due to three factors: a perceived lack of
4757 users, a lack of maintenance, and concerns that it may not be
4758 possible to achieve reliable operation using the existing design.
4760 * As usual, support for the latest Linux distros and toolchain
4761 components has been added. It should work well on Fedora Core 10,
4762 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4763 state) is supported, as is glibc-2.9. The C++ demangler has been
4764 updated so as to work well with C++ compiled by even the most recent
4767 * You can now use frame-level wildcards in suppressions. This was a
4768 frequently-requested enhancement. A line "..." in a suppression now
4769 matches zero or more frames. This makes it easier to write
4770 suppressions which are precise yet insensitive to changes in
4773 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4775 * Very basic support for IBM Power6 has been added (64-bit processes only).
4777 * Valgrind is now cross-compilable. For example, it is possible to
4778 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4779 on a ppc32/64-linux target.
4781 * You can set the main thread's stack size at startup using the
4782 new --main-stacksize= flag (subject of course to ulimit settings).
4783 This is useful for running apps that need a lot of stack space.
4785 * The limitation that you can't use --trace-children=yes together
4786 with --db-attach=yes has been removed.
4788 * The following bugs have been fixed. Note that "n-i-bz" stands for
4789 "not in bugzilla" -- that is, a bug that was reported to us but
4790 never got a bugzilla entry. We encourage you to file bugs in
4791 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4792 mailing the developers (or mailing lists) directly.
4794 n-i-bz Make return types for some client requests 64-bit clean
4795 n-i-bz glibc 2.9 support
4796 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4797 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4798 n-i-bz Building in an env without gdb gives bogus gdb attach
4799 92456 Tracing the origin of uninitialised memory
4800 106497 Valgrind does not demangle some C++ template symbols
4802 151612 Suppression with "..." (frame-level wildcards in .supp files)
4803 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4804 159285 unhandled syscall:25 (stime, on x86-linux)
4805 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4806 160954 ppc build of valgrind crashes with illegal instruction (isel)
4807 160956 mallinfo implementation, w/ patch
4808 162092 Valgrind fails to start gnome-system-monitor
4809 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4810 163794 assertion failure with "--track-origins=yes"
4811 163933 sigcontext.err and .trapno must be set together
4812 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4813 164476 Missing kernel module loading system calls
4814 164669 SVN regression: mmap() drops posix file locks
4815 166581 Callgrind output corruption when program forks
4816 167288 Patch file for missing system calls on Cell BE
4817 168943 unsupported scas instruction pentium
4818 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4819 172417 x86->IR: 0x82 ...
4820 172563 amd64->IR: 0xD9 0xF5 - fprem1
4821 173099 .lds linker script generation error
4822 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4823 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4825 174908 --log-file value not expanded correctly for core file
4826 175044 Add lookup_dcookie for amd64
4827 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4829 Developer-visible changes:
4831 * Valgrind's debug-info reading machinery has been majorly overhauled.
4832 It can now correctly establish the addresses for ELF data symbols,
4833 which is something that has never worked properly before now.
4835 Also, Valgrind can now read DWARF3 type and location information for
4836 stack and global variables. This makes it possible to use the
4837 framework to build tools that rely on knowing the type and locations
4838 of stack and global variables, for example exp-Ptrcheck.
4840 Reading of such information is disabled by default, because most
4841 tools don't need it, and because it is expensive in space and time.
4842 However, you can force Valgrind to read it, using the
4843 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4844 make use of such information, if present, to provide source-level
4845 descriptions of data addresses in the error messages they create.
4847 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4848 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).