1 Release 3.22.0 (?? Oct 2023)
2 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
5 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
6 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
7 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
8 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
9 AMD64/macOS 10.13 and nanoMIPS/Linux.
11 * ==================== CORE CHANGES ===================
13 * A new configure option --with-gdbscripts-dir lets you install
14 the gdb valgrind python monitor scripts in a specific location.
15 For example a distro could use it to install the scripts in a
16 safe load location --with-gdbscripts-dir=%{_datadir}/gdb/auto-load
17 It is also possible to configure --without-gdb-scripts-dir so no
18 .debug_gdb_scripts section is added to the vgpreload library and
19 no valgrind-monitor python scripts are installed at all.
21 * ================== PLATFORM CHANGES =================
23 * Support has been added for FreeBSD 14 and FreeBSD 15.
24 * Add support for the folllowing FreeBSD system calls:
25 close_range, kqueuex, membarrier, timerfd_create,
26 timerfd_settime and timerfd_gettime (all added in FreeBSD 15).
28 * ==================== TOOL CHANGES ===================
30 * Memcheck now tests and warns about the values used for
31 alignment and size. These apply to various functions: memalign,
32 posix_memalign and aligned_alloc in C and various overloads
33 of operators new and delete in C++. The kinds of error that can
35 - invalid alignment, for instance the alignment is usually required
37 - mismatched alignment between aligned allocation and aligned
39 - mismatched size when sized delete is used
40 - bad size for functions that have implementation defined behaviour
41 when the requested size is zero
44 - You can now profile part of a program's execution using the new
45 `CACHEGRIND_START_INSTRUMENTATION` and `CACHEGRIND_STOP_INSTRUMENTATION`
46 client requests, along with the new `--instr-at-start` option. The
47 behaviour is the same as Callgrind's equivalent functionality.
49 * ==================== FIXED BUGS ====================
51 The following bugs have been fixed or resolved. Note that "n-i-bz"
52 stands for "not in bugzilla" -- that is, a bug that was reported to us
53 but never got a bugzilla entry. We encourage you to file bugs in
54 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
55 than mailing the developers (or mailing lists) directly -- bugs that
56 are not entered into bugzilla tend to get forgotten about or ignored.
58 426751 Valgrind reports "still reachable" memory using musl (alpine running inside docker)
59 432801 Valgrind 3.16.1 reports a jump based on uninitialized memory somehow related to clang and signals
60 433857 Add validation to C++17 aligned new/delete alignment size
61 433859 Add mismatched detection to C++ 17 aligned new/delete
62 460192 Add epoll_pwait2
63 466105 aligned_alloc problems, part 2
64 467441 Add mismatched detection to C++ 14 sized delete
65 469049 link failure on ppc64 (big endian) valgrind 3.20
66 469146 massif --ignore-fn does not ignore inlined functions
67 469768 Make it possible to install gdb scripts in a different location
68 470121 Can't run callgrind_control with valgrind 3.21.0 because of perl errors
69 470132 s390x: Assertion failure on VGM instruction
70 470520 Multiple realloc zero errors crash in MC_(eq_Error)
71 470713 Failure on the Yosys project: valgrind: m_libcfile.c:1802
72 (Bool vgPlain_realpath(const HChar *, HChar *)):
73 Assertion 'resolved' failed
74 470830 Don't print actions vgdb me ... continue for vgdb --multi mode
75 470978 s390x: Valgrind cannot start qemu-kvm when "sysctl vm.allocate_pgste=0"
76 471311 gdb --multi mode stdout redirecting to stderr
77 471807 Add support for lazy reading and downloading of DWARF debuginfo
78 472219 Syscall param ppoll(ufds.events) points to uninitialised byte(s)
79 472875 none/tests/s390x/dfp-1 failure
80 472963 Broken regular expression in configure.ac
81 473604 Fix bug472219.c compile failure with Clang 16
82 473677 make check compile failure with Clang 16 based on GCC 13.x
83 473745 must-be-redirected function - strlen
84 473870 FreeBSD 14 applications fail early at startup
85 473944 Handle mold linker split RW PT_LOAD segments correctly
86 474332 aligned_alloc under Valgrind returns nullptr when alignment is not a multiple of sizeof(void *)
87 n-i-bz Allow arguments with spaces in .valgrindrc files
88 n-i-bz FreeBSD fixed reading of Valgrind tools own debuginfo
90 To see details of a given bug, visit
91 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
92 where XXXXXX is the bug number as listed above.
94 Release 3.21.0 (28 Apr 2023)
95 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
97 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
98 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
99 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
100 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
101 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
102 AMD64/macOS 10.13 and nanoMIPS/Linux.
104 * ==================== CORE CHANGES ===================
106 * When GDB is used to debug a program running under valgrind using
107 the valgrind gdbserver, GDB will automatically load some
108 python code provided in valgrind defining GDB front end commands
109 corresponding to the valgrind monitor commands.
110 These GDB front end commands accept the same format as
111 the monitor commands directly sent to the Valgrind gdbserver.
112 These GDB front end commands provide a better integration
113 in the GDB command line interface, so as to use for example
114 GDB auto-completion, command specific help, searching for
115 a command or command help matching a regexp, ...
116 For relevant monitor commands, GDB will evaluate arguments
117 to make the use of monitor commands easier.
118 For example, instead of having to print the address of a variable
119 to pass it to a subsequent monitor command, the GDB front end
120 command will evaluate the address argument. It is for example
122 (gdb) memcheck who_points_at &some_struct sizeof(some_struct)
125 $2 = (some_struct_type *) 0x1130a0 <some_struct>
126 (gdb) p sizeof(some_struct)
128 (gdb) monitor who_point_at 0x1130a0 40
130 * The vgdb utility now supports extended-remote protocol when
131 invoked with --multi. In this mode the GDB run command is
132 supported. Which means you don't need to run gdb and valgrind
133 from different terminals. So for example to start your program
134 in gdb and run it under valgrind you can do:
136 (gdb) set remote exec-file prog
138 (gdb) target extended-remote | vgdb --multi
141 * The behaviour of realloc with a size of zero can now
142 be changed for tools that intercept malloc. Those
143 tools are memcheck, helgrind, drd, massif and dhat.
144 Realloc implementations generally do one of two things
145 - free the memory like free() and return NULL
146 (GNU libc and ptmalloc).
147 - either free the memory and then allocate a
148 minimum sized block or just return the
149 original pointer. Return NULL if the
150 allocation of the minimum sized block fails
151 (jemalloc, musl, snmalloc, Solaris, macOS).
152 When Valgrind is configured and built it will
153 try to match the OS and libc behaviour. However
154 if you are using a non-default library to replace
155 malloc and family (e.g., musl on a glibc Linux or
156 tcmalloc on FreeBSD) then you can use a command line
157 option to change the behaviour of Valgrind:
158 --realloc-zero-bytes-frees=yes|no [yes on Linux glibc, no otherwise]
160 * ================== PLATFORM CHANGES =================
162 * Make the address space limit on FreeBSD amd64 128Gbytes
163 (the same as Linux and Solaris, it was 32Gbytes)
165 * ==================== TOOL CHANGES ===================
168 - When doing a delta leak_search, it is now possible to only
169 output the new loss records compared to the previous leak search.
170 This is available in the memcheck monitor command 'leak_search'
171 by specifying the "new" keyword or in your program by using
172 the client request VALGRIND_DO_NEW_LEAK_CHECK.
173 Whenever a "delta" leak search is done (i.e. when specifying
174 "new" or "increased" or "changed" in the monitor command),
175 the new loss records have a "new" marker.
176 - Valgrind now contains python code that defines GDB memcheck
177 front end monitor commands. See CORE CHANGES.
178 - Performs checks for the use of realloc with a size of zero.
179 This is non-portable and a source of errors. If memcheck
180 detects such a usage it will generate an error
181 realloc() with size 0
182 followed by the usual callstacks.
183 A switch has been added to allow this to be turned off:
184 --show-realloc-size-zero=yes|no [yes]
187 - The option ---history-backtrace-size=<number> allows to configure
188 the number of entries to record in the stack traces of "old"
189 accesses. Previously, this number was hardcoded to 8.
190 - Valgrind now contains python code that defines GDB helgrind
191 front end monitor commands. See CORE CHANGES.
194 - `--cache-sim=no` is now the default. The cache simulation is old and
195 unlikely to match any real modern machine. This means only the `Ir`
196 event are gathered by default, but that is by far the most useful
198 - `cg_annotate`, `cg_diff`, and `cg_merge` have been rewritten in
199 Python. As a result, they all have more flexible command line
200 argument handling, e.g. supporting `--show-percs` and
201 `--no-show-percs` forms as well as the existing `--show-percs=yes`
202 and `--show-percs=no`.
203 - `cg_annotate` has some functional changes.
204 - It's much faster, e.g. 3-4x on common cases.
205 - It now supports diffing (with `--diff`, `--mod-filename`, and
206 `--mod-funcname`) and merging (by passing multiple data files).
207 - It now provides more information at the file and function level.
208 There are now "File:function" and "Function:file" sections. These
209 are very useful for programs that use inlining a lot.
210 - Support for user-annotated files and the `-I`/`--include` option
211 has been removed, because it was of little use and blocked other
213 - The `--auto` option is renamed `--annotate`, though the old
214 `--auto=yes`/`--auto=no` forms are still supported.
215 - `cg_diff` and `cg_merge` are now deprecated, because `cg_annotate`
216 now does a better job of diffing and merging.
217 - The Cachegrind output file format has changed very slightly, but in
218 ways nobody is likely to notice.
221 - Valgrind now contains python code that defines GDB callgrind
222 front end monitor commands. See CORE CHANGES.
225 - Valgrind now contains python code that defines GDB massif
226 front end monitor commands. See CORE CHANGES.
229 - A new kind of user request has been added which allows you to
230 override the 1024 byte limit on access count histograms for blocks
231 of memory. The client request is DHAT_HISTOGRAM_MEMORY.
233 * ==================== FIXED BUGS ====================
235 The following bugs have been fixed or resolved. Note that "n-i-bz"
236 stands for "not in bugzilla" -- that is, a bug that was reported to us
237 but never got a bugzilla entry. We encourage you to file bugs in
238 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
239 than mailing the developers (or mailing lists) directly -- bugs that
240 are not entered into bugzilla tend to get forgotten about or ignored.
242 170510 Don't warn about ioctl of size 0 without direction hint
243 241072 List tools in --help output
244 327548 false positive while destroying mutex
245 382034 Testcases build fixes for musl
246 351857 confusing error message about valid command line option
247 374596 inconsistent RDTSCP support on x86_64
248 392331 Spurious lock not held error from inside pthread_cond_timedwait
249 397083 Likely false positive "uninitialised value(s)" for __wmemchr_avx2 and __wmemcmp_avx2_movbe
250 400793 pthread_rwlock_timedwrlock false positive
251 419054 Unhandled syscall getcpu on arm32
252 433873 openat2 syscall unimplemented on Linux
253 434057 Add stdio mode to valgrind's gdbserver
254 435441 valgrind fails to interpose malloc on musl 1.2.2 due to weak symbol name and no libc soname
255 436413 Warn about realloc of size zero
256 439685 compiler warning in callgrind/main.c
257 444110 priv/guest_ppc_toIR.c:36198:31: warning: duplicated 'if' condition.
258 444487 hginfo test detects an extra lock inside data symbol "_rtld_local"
259 444488 Use glibc.pthread.stack_cache_size tunable
260 444568 drd/tests/pth_barrier_thr_cr fails on Fedora 38
261 445743 "The impossible happened: mutex is locked simultaneously by two threads"
262 while using mutexes with priority inheritance and signals
263 449309 Missing loopback device ioctl(s)
264 459476 vgdb: allow address reuse to avoid "address already in use" errorsuse" errors
265 460356 s390: Sqrt32Fx4 -- cannot reduce tree
266 462830 WARNING: unhandled amd64-freebsd syscall: 474
267 463027 broken check for MPX instruction support in assembler
268 464103 Enhancement: add a client request to DHAT to mark memory to be histogrammed
269 464476 Firefox fails to start under Valgrind
270 464609 Valgrind memcheck should support Linux pidfd_open
271 464680 Show issues caused by memory policies like selinux deny_execmem
272 464859 Build failures with GCC-13 (drd tsan_unittest)
273 464969 D language demangling
274 465435 m_libcfile.c:66 (vgPlain_safe_fd): Assertion 'newfd >= VG_(fd_hard_limit)' failed.
275 466104 aligned_alloc problems, part 1
276 467036 Add time cost statistics for Regtest
277 467482 Build failure on aarch64 Alpine
278 467714 fdleak_* and rlimit tests fail when parent process has more than
279 64 descriptors opened
280 467839 Gdbserver: Improve compatibility of library directory name
281 468401 [PATCH] Add a style file for clang-format
282 468556 Build failure for vgdb
283 468606 build: remove "Valgrind relies on GCC" check/output
284 469097 ppc64(be) doesn't support SCV syscall instruction
285 n-i-bz FreeBSD rfork syscall fail with EINVAL or ENOSYS rather than VG_(unimplemented)
287 To see details of a given bug, visit
288 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
289 where XXXXXX is the bug number as listed above.
291 * ==================== KNOWN ISSUES ===================
293 * configure --enable-lto=yes is know to not work in all setups.
294 See bug 469049. Workaround: Build without LTO.
296 (3.21.0.RC1: 14 Apr 2023)
297 (3.21.0.RC2: 21 Apr 2023)
299 Release 3.20.0 (24 Oct 2022)
300 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
302 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
303 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
304 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
305 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
306 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
307 AMD64/macOS 10.13 and nanoMIPS/Linux.
309 * ==================== CORE CHANGES ===================
311 * The option "--vgdb-stop-at=event1,event2,..." accepts the new value abexit.
312 This indicates to invoke gdbserver when your program exits abnormally
313 (i.e. with a non zero exit code).
314 * Fix Rust v0 name demangling.
315 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
316 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
317 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with
319 * The option --enable-debuginfod=<no|yes> [default: yes] has been added on
321 * More DWARF5 support as generated by clang14.
323 * ==================== FIXED BUGS ====================
325 The following bugs have been fixed or resolved. Note that "n-i-bz"
326 stands for "not in bugzilla" -- that is, a bug that was reported to us
327 but never got a bugzilla entry. We encourage you to file bugs in
328 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
329 than mailing the developers (or mailing lists) directly -- bugs that
330 are not entered into bugzilla tend to get forgotten about or ignored.
332 131186 writev reports error in (vector[...])
333 434764 iconv_open causes ld.so v2.28+ to use optimised strncmp
334 446754 Improve error codes from alloc functions under memcheck
335 452274 memcheck crashes with Assertion 'sci->status.what == SsIdle' failed
336 452779 Valgrind fails to build on FreeBSD 13.0 with llvm-devel (15.0.0)
337 453055 shared_timed_mutex drd test fails with "Lock shared failed" message
338 453602 Missing command line option to enable/disable debuginfod
339 452802 Handle lld 9+ split RW PT_LOAD segments correctly
340 454040 s390x: False-positive memcheck:cond in memmem on arch13 systems
341 456171 [PATCH] FreeBSD: Don't record address errors when accessing the 'kern.ps_strings' sysctl struct
342 n-i-bz Implement vgdb invoker on FreeBSD
343 458845 PowerPC: The L field for the dcbf and sync instruction should be
345 458915 Remove register cache to fix 458915 gdbserver causes wrong syscall return
346 459031 Documentation on --error-exitcode incomplete
347 459477 XERROR messages lacks ending '\n' in vgdb
348 462007 Implicit int in none/tests/faultstatus.c
350 To see details of a given bug, visit
351 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
352 where XXXXXX is the bug number as listed above.
354 (3.20.0.RC1: 20 Oct 2022)
357 Release 3.19.0 (11 Apr 2022)
358 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
360 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
361 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
362 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
363 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
364 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
365 AMD64/macOS 10.13 and nanoMIPS/Linux.
367 * ==================== CORE CHANGES ===================
369 * Fix Rust v0 name demangling.
370 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
371 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
372 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with HardenedBSD
374 * ================== PLATFORM CHANGES =================
377 - ignore the "v8.x" architecture levels, only look at actual CPU features
378 present. Fixes mismatch detected between RDMA and atomics features
379 preventing startup on some QEMU configurations.
380 - Implement LD{,A}XP and ST{,L}XP
381 - Fix incorrect code emitted for doubleword CAS.
384 - Fix sys_ipc semtimedop syscall
385 - Fix VFLRX and WFLRX instructions
386 - Fix EXRL instruction with negative offset
389 - Reimplement the vbpermq instruction support to generate less Iops and
390 avoid overflowing internal buffers.
391 - Fix checking for scv support to avoid "Facility 'SCV' unavailable (12),
392 exception" messages in dmsg.
393 - Fix setting condition code for Vector Compare quad word instructions.
394 - Fix fix lxsibzx, lxsihzx and lxsihzx instructions so they only load
395 their respective sized data.
396 - Fix the prefixed stq instruction in PC relative mode.
398 * ==================== TOOL CHANGES ===================
401 - Speed up --track-origins=yes for large (in the range of hundreds to
402 thousands of megabytes) mmap/munmaps.
404 - Several fixes for new versions of libstd++ using new posix try_lock
407 * ==================== FIXED BUGS ====================
409 The following bugs have been fixed or resolved. Note that "n-i-bz"
410 stands for "not in bugzilla" -- that is, a bug that was reported to us
411 but never got a bugzilla entry. We encourage you to file bugs in
412 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
413 than mailing the developers (or mailing lists) directly -- bugs that
414 are not entered into bugzilla tend to get forgotten about or ignored.
416 403802 leak_cpp_interior fails with some reachable blocks different than expected
417 435732 memcheck/tests/leak_cpp_interior fails with gcc11
418 444242 s390x: Valgrind crashes on EXRL with negative offset
419 444399 arm64: unhandled instruction 0xC87F2D89 (LD{,A}XP and ST{,L}XP).
421 444481 gdb_server test failures on s390x
422 444495 dhat/tests/copy fails on s390x
423 444552 memcheck/tests/sem fails on s390x with glibc 2.34
424 444571 PPC, fix the lxsibzx and lxsihzx so they only load their respective
426 444836 PPC, pstq instruction for R=1 is not storing to the correct address.
427 444925 fexecve syscall wrapper not properly implemented
428 445032 valgrind/memcheck crash with SIGSEGV when SIGVTALRM timer used and
430 445211 Fix out of tree builds
431 445300 [PATCH] Fix building tests with Musl
432 445011 SIGCHLD is sent when valgrind uses debuginfod-find
433 445354 arm64 backend: incorrect code emitted for doubleword CAS
434 445415 arm64 front end: alignment checks missing for atomic instructions
435 445504 Using C++ condition_variable results in bogus "mutex is locked simultaneously by two threads" warning
436 445607 Unhandled amd64-freebsd syscall: 247
437 445668 Inline stack frame generation is broken for Rust binaries
438 445916 Demangle Rust v0 symbols with .llvm suffix
439 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false positives
440 446138 DRD/Helgrind with std::timed_mutex::try_lock_until false positives
441 446281 Add a DRD suppression for fwrite
442 446103 Memcheck: `--track-origins=yes` causes extreme slowdowns for large mmap/munmap
443 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false
444 446251 TARGET_SIGNAL_THR added to enum target_signal
445 446823 FreeBSD - missing syscalls when using libzm4
446 447991 s390x: Valgrind indicates illegal instruction on wflrx
447 447995 Valgrind segfault on power10 due to hwcap checking code
448 449483 Powerpc: vcmpgtsq., vcmpgtuq,, vcmpequq. instructions not setting the
449 condition code correctly.
450 449672 ppc64 --track-origins=yes failures because of bad cmov addHRegUse
451 449838 sigsegv liburing the 'impossible' happened for io_uring_setup
452 450025 Powerc: ACC file not implemented as a logical overlay of the VSR
454 450437 Warn for execve syscall with argv or argv[0] being NULL
455 450536 Powerpc: valgrind throws 'facility scv unavailable exception'
456 451626 Syscall param bpf(attr->raw_tracepoint.name) points to unaddressable byte(s)
457 451827 [ppc64le] VEX temporary storage exhausted with several vbpermq instructions
458 451843 valgrind fails to start on a FreeBSD system which enforces W^X
460 To see details of a given bug, visit
461 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
462 where XXXXXX is the bug number as listed above.
464 (3.19.0.RC1: 02 Apr 2022)
465 (3.19.0.RC2: 08 Apr 2022)
468 Release 3.18.0 (15 Oct 2021)
469 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
471 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
472 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
473 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
474 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
475 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
476 AMD64/macOS 10.13 and nanoMIPS/Linux.
478 * ==================== CORE CHANGES ===================
480 * The libiberty demangler has been updated, which brings support for
481 Rust v0 name demangling. [Update: alas, due to a bug, this support
482 isn't working in 3.18.0.]
484 * __libc_freeres isn't called anymore after the program recieves a
485 fatal signal. Causing some internal glibc resources to hang around,
486 but preventing any crashes after the program has ended.
488 * The DWARF reader is now very much faster at startup when just
489 --read-inline-info=yes (the default in most cases) is given.
491 * glibc 2.34, which moved various functions from libpthread.so into
492 libc.so, is now supported.
494 * ================== PLATFORM CHANGES =================
498 - v8.2 scalar and vector FABD, FACGE, FACGT and FADD.
499 - v8.2 FP compare & conditional compare instructions.
500 - Zero variants of v8.2 FP compare instructions.
504 - Support the miscellaneous-instruction-extensions facility 3 and
505 the vector-enhancements facility 2. This enables programs
506 compiled with "-march=arch13" or "-march=z15" to be executed
511 - ISA 3.1 support is now complete
512 - ISA 3.0 support for the darn instruction added.
513 - ISA 3.0 support for the vector system call instruction scv added.
514 - ISA 3.0 support for the copy, paste and cpabort instructions added.
516 * Support for X86/FreeBSD and AMD64/FreeBSD has been added.
518 * ==================== OTHER CHANGES ====================
520 * Memcheck on amd64: minor fixes to remove some false positive
523 * ==================== FIXED BUGS ====================
525 The following bugs have been fixed or resolved. Note that "n-i-bz"
526 stands for "not in bugzilla" -- that is, a bug that was reported to us
527 but never got a bugzilla entry. We encourage you to file bugs in
528 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
529 than mailing the developers (or mailing lists) directly -- bugs that
530 are not entered into bugzilla tend to get forgotten about or ignored.
532 208531 [PATCH]: FreeBSD support for valgrind
533 368960 WARNING: unhandled amd64-linux syscall: 163 (acct)
534 407589 [Linux] Add support for C11 aligned_alloc() and GNU reallocarray()
535 423963 Error in child thread when CLONE_PIDFD is used
536 426148 crash with "impossible happened" when running BPF CO-RE programs
537 429375 PPC ISA 3.1 support is missing, part 9
538 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
539 431306 Update demangler to support Rust v0 name mangling
540 432387 s390x: z15 instructions support
541 433437 FreeBSD support, part 1
542 433438 FreeBSD support, part 2
543 433439 FreeBSD support, part 3
544 433469 FreeBSD support, part 4
545 433473 FreeBSD support, part 5
546 433477 FreeBSD support, part 6
547 433479 FreeBSD support, part 7
548 433504 FreeBSD support, part 8
549 433506 FreeBSD support, part 9
550 433507 FreeBSD support, part 10
551 433508 FreeBSD support, part 11
552 433510 FreeBSD support, part 12
553 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
554 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
555 434296 s390x: False-positive memcheck diagnostics from vector string
557 434840 PPC64 darn instruction not supported
558 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
559 435908 valgrind tries to fetch from deubginfod for files which already
560 have debug information
561 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2
562 439046 valgrind is unusably large when linked with lld
563 439090 Implement close_range(2)
564 439326 Valgrind 3.17.0 won't compile with Intel 2021 oneAPI compilers
565 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
566 440670 unhandled ppc64le-linux syscall: 252 statfs64 and 253 fstatfs64
567 440906 Fix impossible constraint issue in P10 testcase.
568 441512 Remove a unneeded / unnecessary prefix check.
569 441534 Update the expected output for test_isa_3_1_VRT.
570 442061 very slow execution under Fedora 34 (readdwarf3)
571 443031 Gcc -many change requires explicit .machine directives
572 443033 Add support for the ISA 3.0 mcrxrx instruction
573 443034 Sraw, srawi, srad, sradi, mfs
574 443178 Powerpc, test jm-mfspr expected output needs to be updated.
575 443179 Need new test for the lxvx and stxvx instructions on ISA 2.07 and
577 443180 The subnormal test and the ISA 3.0 test generate compiler warnings
578 443314 In the latest GIT version, Valgrind with "--trace-flags" crashes
580 443605 Don't call final_tidyup (__libc_freeres) on FatalSignal
582 To see details of a given bug, visit
583 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
584 where XXXXXX is the bug number as listed below.
586 (3.18.0.RC1: 12 Oct 2021)
587 (3.18.0: 15 Oct 2021)
591 Release 3.17.0 (19 Mar 2021)
592 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
594 3.17.0 fixes a number of bugs and adds some functional changes: support for
595 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
596 some new instructions for Arm64, S390 and POWER. There are also some tool
599 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
600 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
601 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
602 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
603 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
605 * ==================== CORE CHANGES ===================
607 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
610 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
611 debugging information. When a debuginfo file cannot be found locally,
612 Valgrind is able to query debuginfod servers for the file using its
613 build-id. See the user manual for more information about debuginfod support.
615 * ================== PLATFORM CHANGES =================
619 - Inaccuracies resulting from double-rounding in the simulation of
620 floating-point multiply-add/subtract instructions have been fixed. These
621 should now behave exactly as the hardware does.
623 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
624 ongoing. Support for the half-word variants of at least the following
625 instructions has been added:
627 FABS <Vd>.<T>, <Vn>.<T>
629 FNEG <Vd>.<T>, <Vn>.<T>
631 FSQRT <Vd>.<T>, <Vn>.<T>
636 - Implement the new instructions/features that were added to z/Architecture
637 with the vector-enhancements facility 1. Also cover the instructions from
638 the vector-packed-decimal facility that are defined outside the chapter
639 "Vector Decimal Instructions", but not the ones from that chapter itself.
641 For a detailed list of newly supported instructions see the updates to
642 `docs/internals/s390-opcodes.csv'.
644 Since the miscellaneous instruction extensions facility 2 was already
645 added in Valgrind 3.16.0, this completes the support necessary to run
646 general programs built with `--march=z14' under Valgrind. The
647 vector-packed-decimal facility is currently not exploited by the standard
648 toolchain and libraries.
652 - Various bug fixes. Fix for the sync field to limit setting just two of
653 the two bits in the L-field. Fix the write size for the stxsibx and
654 stxsihx instructions. Fix the modsw and modsd instructions.
656 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
657 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
658 conversion instructions are still missing.
660 * ==================== TOOL CHANGES ====================
662 * General tool changes
664 - All the tools and their vgpreload libraries are now installed under
665 libexec because they cannot be executed directly and should be run through
666 the valgrind executable. This should be an internal, not user visible,
667 change, but might impact valgrind packagers.
669 - The --track-fds option now respects -q, --quiet and won't output anything
670 if no file descriptors are leaked. It also won't report the standard stdin
671 (0), stdout (1) or stderr (2) descriptors as being leaked with
672 --trace-fds=yes anymore. To track whether the standard file descriptors
673 are still open at the end of the program run use --trace-fds=all.
677 - DHAT has been extended, with two new modes of operation. The new
678 --mode=copy flag triggers copy profiling, which records calls to memcpy,
679 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
680 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
681 the new dhat/dhat.h file. This is useful for learning more about hot code
682 paths. See the user manual for more information about the new modes.
684 - Because of these changes, DHAT's file format has changed. DHAT output
685 files produced with earlier versions of DHAT will not work with this
686 version of DHAT's viewer, and DHAT output files produced with this version
687 of DHAT will not work with earlier versions of DHAT's viewer.
689 * ==================== FIXED BUGS ====================
691 The following bugs have been fixed or resolved. Note that "n-i-bz"
692 stands for "not in bugzilla" -- that is, a bug that was reported to us
693 but never got a bugzilla entry. We encourage you to file bugs in
694 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
695 than mailing the developers (or mailing lists) directly -- bugs that
696 are not entered into bugzilla tend to get forgotten about or ignored.
698 To see details of a given bug, visit
699 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
700 where XXXXXX is the bug number as listed below.
702 140178 open("/proc/self/exe", ...); doesn't quite work
703 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
704 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
705 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
706 345077 linux syscall execveat support (linux 3.19)
707 361770 Missing F_ADD_SEALS
708 369029 handle linux syscalls sched_getattr and sched_setattr
709 384729 __libc_freeres inhibits cross-platform valgrind
710 388787 Support for C++17 new/delete
711 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
712 396656 Warnings while reading debug info
713 397605 ioctl FICLONE mishandled
714 401416 Compile failure with openmpi 4.0
715 408663 Suppression file for musl libc
716 404076 s390x: z14 vector instructions not implemented
717 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
718 (actually affected all x86 and nanomips regardless of host bitness)
719 413547 regression test does not check for Arm 64 features.
720 414268 Enable AArch64 feature detection and decoding for v8.x instructions
721 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
722 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
723 422261 platform selection fails for unqualified client name
724 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
725 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
726 423195 PPC ISA 3.1 support is missing, part 1
727 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
728 424012 crash with readv/writev having invalid but not NULL arg2 iovec
729 424298 amd64: Implement RDSEED
730 425232 PPC ISA 3.1 support is missing, part 2
731 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
732 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
733 426123 PPC ISA 3.1 support is missing, part 3
734 426144 Fix "condition variable has not been initialized" on Fedora 33.
735 427400 PPC ISA 3.1 support is missing, part 4
736 427401 PPC ISA 3.1 support is missing, part 5
737 427404 PPC ISA 3.1 support is missing, part 6
738 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
739 427787 Support new faccessat2 linux syscall (439)
740 427969 debuginfo section duplicates a section in the main ELF file
741 428035 drd: Unbreak the musl build
742 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
743 428716 cppcheck detects potential leak in VEX/useful/smchash.c
744 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
745 429352 PPC ISA 3.1 support is missing, part 7
746 429354 PPC ISA 3.1 support is missing, part 8
747 429692 unhandled ppc64le-linux syscall: 147 (getsid)
748 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
750 429952 Errors when building regtest with clang
751 430354 ppc stxsibx and stxsihx instructions write too much data
752 430429 valgrind.h doesn't compile on s390x with clang
753 430485 expr_is_guardable doesn't handle Iex_Qop
754 431556 Complete arm64 FADDP v8.2 instruction support
755 432102 Add support for DWARF5 as produced by GCC11
756 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
757 432381 drd: Process STACK_REGISTER client requests
758 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
759 432672 vg_regtest: test-specific environment variables not reset between tests
760 432809 VEX should support REX.W + POPF
761 432861 PPC modsw and modsd give incorrect results for 1 mod 12
762 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
763 432215 Add debuginfod functionality
764 433323 Use pkglibexecdir as vglibdir
765 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
766 433629 valgrind/README has type "abd" instead of "and"
767 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
768 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
769 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
770 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
771 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
773 (3.17.0.RC1: 13 Mar 2021)
774 (3.17.0.RC2: 17 Mar 2021)
775 (3.17.0: 19 Mar 2021)
779 Release 3.16.1 (22 June 2020)
780 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
782 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
783 fixes character encoding problems in the documentation HTML.
785 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
786 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
788 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
792 Release 3.16.0 (27 May 2020)
793 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
795 3.16.0 is a feature release with many improvements and the usual collection of
798 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
799 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
800 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
801 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
802 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
804 * ==================== CORE CHANGES ===================
806 * It is now possible to dynamically change the value of many command line
807 options while your program (or its children) are running under Valgrind.
809 To see the list of dynamically changeable options, run
810 "valgrind --help-dyn-options".
812 You can change the options from the shell by using vgdb to launch
813 the monitor command "v.clo <clo option>...".
814 The same monitor command can be used from a gdb connected
815 to the valgrind gdbserver.
816 Your program can also change the dynamically changeable options using
817 the client request VALGRIND_CLO_CHANGE(option).
819 * ================== PLATFORM CHANGES =================
821 * MIPS: preliminary support for nanoMIPS instruction set has been added.
823 * ==================== TOOL CHANGES ====================
827 - The implicit memcpy done by each call to realloc now counts towards the
828 read and write counts of resized heap blocks, making those counts higher
833 - cg_annotate's --auto and --show-percs options now default to 'yes', because
834 they are usually wanted.
838 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
839 because they are usually wanted.
841 - The command option --collect-systime has been enhanced to specify
842 the unit used to record the elapsed time spent during system calls.
843 The command option now accepts the values no|yes|msec|usec|nsec,
844 where yes is a synonym of msec. When giving the value nsec, the
845 system cpu time of system calls is also recorded.
849 - Several memcheck options are now dynamically changeable.
850 Use valgrind --help-dyn-options to list them.
852 - The release 3.15 introduced a backward incompatible change for
853 some suppression entries related to preadv and pwritev syscalls.
854 When reading a suppression entry using the unsupported 3.14 format,
855 valgrind will now produce a warning to say the suppression entry will not
856 work, and suggest the needed change.
858 - Significantly fewer false positive errors on optimised code generated by
859 Clang and GCC. In particular, Memcheck now deals better with the
860 situation where the compiler will transform C-level "A && B" into "B && A"
861 under certain circumstances (in which the transformation is valid).
862 Handling of integer equality/non-equality checks on partially defined
863 values is also improved on some architectures.
867 - The exprimental Stack and Global Array Checking tool has been removed.
868 It only ever worked on x86 and amd64, and even on those it had a
869 high false positive rate and was slow. An alternative for detecting
870 stack and global array overruns is using the AddressSanitizer (ASAN)
871 facility of the GCC and Clang compilers, which require you to rebuild
872 your code with -fsanitize=address.
874 * ==================== OTHER CHANGES ====================
876 * New and modified GDB server monitor features:
878 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
880 - The gdbserver monitor commands that require an address and an optional
881 length argument now accepts the alternate 'C like' syntax "address[length]".
882 For example, the memcheck command "monitor who_points_at 0x12345678 120"
883 can now also be given as "monitor who_points_at 0x12345678[120]".
885 * ==================== FIXED BUGS ====================
887 The following bugs have been fixed or resolved. Note that "n-i-bz"
888 stands for "not in bugzilla" -- that is, a bug that was reported to us
889 but never got a bugzilla entry. We encourage you to file bugs in
890 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
891 than mailing the developers (or mailing lists) directly -- bugs that
892 are not entered into bugzilla tend to get forgotten about or ignored.
894 To see details of a given bug, visit
895 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
896 where XXXXXX is the bug number as listed below.
898 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
899 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
900 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
901 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
902 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
903 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
904 400872 Add nanoMIPS support to Valgrind
905 403212 drd/tests/trylock hangs on FreeBSD
906 404406 s390x: z14 miscellaneous instructions not implemented
907 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
908 406561 mcinfcallWSRU gdbserver_test fails on ppc64
909 406824 Unsupported baseline
910 407218 Add support for the copy_file_range syscall
911 407307 Intercept stpcpy also in ld.so for arm64
912 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
914 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
915 408009 Expose rdrand and f16c even on avx if host cpu supports them
916 408091 Missing pkey syscalls
917 408414 Add support for missing for preadv2 and pwritev2 syscalls
918 409141 Valgrind hangs when SIGKILLed
919 409206 Support for Linux PPS and PTP ioctls
920 409367 exit_group() after signal to thread waiting in futex() causes hangs
921 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
922 409780 References to non-existent configure.in
923 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
924 410599 Non-deterministic behaviour of pth_self_kill_15_other test
925 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
926 411134 Allow the user to change a set of command line options during execution
927 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
928 412344 Problem setting mips flags with specific paths
929 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
930 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
931 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
932 413603 callgrind_annotate/cg_annotate truncate function names at '#'
933 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
934 415136 ARMv8.1 Compare-and-Swap instructions are not supported
935 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
936 416239 valgrind crashes when handling clock_adjtime
937 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
938 416286 DRD reports "conflicting load" error on std::mutex::lock()
939 416301 s390x: "compare and signal" not supported
940 416387 finit_module and bpf syscalls are unhandled on arm64
941 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
942 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
943 416753 new 32bit time syscalls for 2038+
944 417075 pwritev(vector[...]) suppression ignored
945 417075 is not fixed, but incompatible supp entries are detected
946 and a warning is produced for these.
947 417187 [MIPS] Conditional branch problem since 'grail' changes
948 417238 Test memcheck/tests/vbit-test fails on mips64 BE
949 417266 Make memcheck/tests/linux/sigqueue usable with musl
950 417281 s390x: /bin/true segfaults with "grail" enabled
951 417427 commit to fix vki_siginfo_t definition created numerous regression
953 417452 s390_insn_store_emit: dst->tag for HRcVec128
954 417578 Add suppressions for glibc DTV leaks
955 417906 clone with CLONE_VFORK and no CLONE_VM fails
956 418004 Grail code additions break ppc64.
957 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
958 418997 s390x: Support Iex_ITE for float and vector types
959 419503 s390x: Avoid modifying registers returned from isel functions
960 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
961 421570 std_mutex fails on Arm v8.1 h/w
962 434035 vgdb might crash if valgrind is killed
963 n-i-bz Fix minor one time leaks in dhat.
964 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
965 n-i-bz Add support for the Linux io_uring system calls
966 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
967 n-i-bz Fix non-glibc build of test suite with s390x_features
968 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
969 423195 PPC ISA 3.1 support is missing, part 1
971 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
972 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
973 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
977 Release 3.15.0 (12 April 2019)
978 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
980 3.15.0 is a feature release with many improvements and the usual collection of
983 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
984 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
985 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
986 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
987 support for X86/macOS 10.13 and AMD64/macOS 10.13.
989 * ==================== CORE CHANGES ===================
991 * The XTree Massif output format now makes use of the information obtained
992 when specifying --read-inline-info=yes.
994 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
996 * ==================== TOOL CHANGES ====================
1000 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
1001 it has been promoted from an experimental tool to a regular tool. Run it
1002 with --tool=dhat instead of --tool=exp-dhat.
1004 - DHAT now prints only minimal data when the program ends, instead writing
1005 the bulk of the profiling data to a file. As a result, the --show-top-n
1006 and --sort-by options have been removed.
1008 - Profile results can be viewed with the new viewer, dh_view.html. When
1009 a run ends, a short message is printed, explaining how to view the result.
1011 - See the documentation for more details.
1015 - cg_annotate has a new option, --show-percs, which prints percentages next
1016 to all event counts.
1020 - callgrind_annotate has a new option, --show-percs, which prints percentages
1021 next to all event counts.
1023 - callgrind_annotate now inserts commas in call counts, and
1024 sort the caller/callee lists in the call tree.
1028 - The default value for --read-inline-info is now "yes" on
1029 Linux/Android/Solaris. It is still "no" on other OS.
1033 - The option --xtree-leak=yes (to output leak result in xtree format)
1034 automatically activates the option --show-leak-kinds=all, as xtree
1035 visualisation tools such as kcachegrind can in any case select what kind
1036 of leak to visualise.
1038 - There has been further work to avoid false positives. In particular,
1039 integer equality on partially defined inputs (C == and !=) is now handled
1042 * ==================== OTHER CHANGES ====================
1044 * The new option --show-error-list=no|yes displays, at the end of the run, the
1045 list of detected errors and the used suppressions. Prior to this change,
1046 showing this information could only be done by specifying "-v -v", but that
1047 also produced a lot of other possibly-non-useful messages. The option -s is
1048 equivalent to --show-error-list=yes.
1050 * ==================== FIXED BUGS ====================
1052 The following bugs have been fixed or resolved. Note that "n-i-bz"
1053 stands for "not in bugzilla" -- that is, a bug that was reported to us
1054 but never got a bugzilla entry. We encourage you to file bugs in
1055 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1056 than mailing the developers (or mailing lists) directly -- bugs that
1057 are not entered into bugzilla tend to get forgotten about or ignored.
1059 To see details of a given bug, visit
1060 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1061 where XXXXXX is the bug number as listed below.
1063 385411 s390x: z13 vector floating-point instructions not implemented
1064 397187 z13 vector register support for vgdb gdbserver
1065 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
1066 398870 Please add support for instruction vcvtps2ph
1067 399287 amd64 front end: Illegal Instruction vcmptrueps
1068 399301 Use inlined frames in Massif XTree output.
1069 399322 Improve callgrind_annotate output
1070 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
1071 400164 helgrind test encounters mips x-compiler warnings and assembler error
1072 400490 s390x: VRs allocated as if separate from FPRs
1073 400491 s390x: Operand of LOCH treated as unsigned integer
1074 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
1075 options, which specify a mips64 processor
1076 401112 LLVM 5.0 generates comparison against partially initialized data
1077 401277 More bugs in z13 support
1078 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
1079 401578 drd: crashes sometimes on fork()
1080 401627 memcheck errors with glibc avx2 optimized wcsncmp
1081 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
1082 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
1083 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
1085 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
1086 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
1087 402123 invalid assembler opcodes for mips32r2
1088 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
1089 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
1090 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
1091 402351 mips64 libvexmultiarch_test fails on s390x
1092 402369 Overhaul DHAT
1093 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
1094 402480 Do not use %rsp in clobber list
1095 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
1096 402515 Implement new option --show-error-list=no|yes / -s
1097 402519 POWER 3.0 addex instruction incorrectly implemented
1098 402781 Redo the cache used to process indirect branch targets
1099 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
1100 403552 s390x: wrong facility bit checked for vector facility
1101 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
1102 404638 Add VG_(replaceIndexXA)
1103 404843 s390x: backtrace sometimes ends prematurely
1104 404888 autotools cleanup series
1105 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
1106 405182 Valgrind fails to build with Clang
1107 405205 filter_libc: remove the line holding the futex syscall error entirely
1108 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
1109 the upper and lower 32-bits of the 64-bit result
1110 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
1111 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
1112 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
1114 405403 s390x disassembler cannot be used on x86
1115 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
1116 405458 MIPS mkFormVEC arguments swapped?
1117 405716 drd: Fix an integer overflow in the stack margin calculation
1118 405722 Support arm64 core dump
1119 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
1120 of the 64-bit destination field.
1121 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
1122 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
1123 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
1125 406256 PPC64, vector floating point instructions don't handle subnormal
1126 according to VSCR[NJ] bit setting.
1127 406352 cachegrind/callgrind fails ann tests because of missing a.c
1128 406354 dhat is broken on x86 (32bit)
1129 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
1130 406357 gdbserver_tests fails because of gdb output change
1131 406360 memcheck/tests/libstdc++.supp needs more supression variants
1132 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
1133 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
1134 407340 PPC64, does not support the vlogefp, vexptefp instructions.
1135 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
1136 n-i-bz Fix callgrind_annotate non deterministic order for equal total
1137 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
1138 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
1139 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
1141 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
1142 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
1143 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
1147 Release 3.14.0 (9 October 2018)
1148 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1150 3.14.0 is a feature release with many improvements and the usual collection of
1153 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1154 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1155 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1156 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
1157 support for X86/macOS 10.13, AMD64/macOS 10.13.
1159 * ==================== CORE CHANGES ===================
1161 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
1162 debug info for unloaded code. This allows saved stack traces (e.g. for
1163 memory leaks) to include file/line info for code that has been dlclose'd (or
1164 similar). See the user manual for more information and known limitations.
1166 * Ability to specify suppressions based on source file name and line number.
1168 * Majorly overhauled register allocator. No end-user changes, but the JIT
1169 generates code a bit more quickly now.
1171 * ================== PLATFORM CHANGES =================
1173 * Preliminary support for macOS 10.13 has been added.
1175 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
1177 * mips: support for MIPS SIMD architecture (MSA) has been added.
1179 * mips: support for MIPS N32 ABI has been added.
1181 * s390: partial support for vector instructions (integer and string) has been
1184 * ==================== TOOL CHANGES ====================
1186 * Helgrind: Addition of a flag
1187 --delta-stacktrace=no|yes [yes on linux amd64/x86]
1188 which specifies how full history stack traces should be computed.
1189 Setting this to =yes can speed up Helgrind by 25% when using
1190 --history-level=full.
1192 * Memcheck: reduced false positive rate for optimised code created by Clang 6
1193 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
1194 blocks more carefully to determine where it can avoid expensive definedness
1195 checks without loss of precision. This is controlled by the flag
1196 --expensive-definedness-checks=no|auto|yes [auto].
1198 * ==================== OTHER CHANGES ====================
1200 * Valgrind is now buildable with link-time optimisation (LTO). A new
1201 configure option --enable-lto=yes allows building Valgrind with LTO. If the
1202 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
1203 Note that if you are doing Valgrind development, --enable-lto=yes massively
1204 slows down the build process.
1206 * ==================== FIXED BUGS ====================
1208 The following bugs have been fixed or resolved. Note that "n-i-bz"
1209 stands for "not in bugzilla" -- that is, a bug that was reported to us
1210 but never got a bugzilla entry. We encourage you to file bugs in
1211 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1212 than mailing the developers (or mailing lists) directly -- bugs that
1213 are not entered into bugzilla tend to get forgotten about or ignored.
1215 To see details of a given bug, visit
1216 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1217 where XXXXXX is the bug number as listed below.
1219 79362 Debug info is lost for .so files when they are dlclose'd
1220 208052 strlcpy error when n = 0
1221 255603 exp-sgcheck Assertion '!already_present' failed
1222 338252 building valgrind with -flto (link time optimisation) fails
1223 345763 MIPS N32 ABI support
1224 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
1225 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
1226 372347 Replacement problem of the additional c++14/c++17 new/delete operators
1227 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1228 376257 helgrind history full speed up using a cached stack
1229 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
1231 379748 Fix missing pselect syscall (OS X 10.11)
1232 379754 Fix missing syscall ulock_wait (OS X 10.12)
1233 380397 s390x: __GI_strcspn() replacemenet needed
1234 381162 possible array overrun in VEX register allocator
1235 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
1236 381274 powerpc too chatty even with --sigill-diagnostics=no
1237 381289 epoll_pwait can have a NULL sigmask
1238 381553 VEX register allocator v3
1239 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
1240 381769 Use ucontext_t instead of struct ucontext
1241 381805 arm32 needs ld.so index hardwire for new glibc security fixes
1242 382256 gz compiler flag test doesn't work for gold
1243 382407 vg_perf needs "--terse" command line option
1244 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
1245 382563 MIPS MSA ASE support
1246 382998 xml-socket doesn't work
1247 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
1248 383723 Fix missing kevent_qos syscall (macOS 10.11)
1249 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
1250 384096 Mention AddrCheck at Memcheck's command line option [..]
1251 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
1252 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
1253 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
1254 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
1255 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
1256 384337 performance improvements to VEX register allocator v2 and v3
1257 384526 reduce number of spill insns generated by VEX register allocator v3
1258 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
1259 384631 Sanitise client args as printed with -v
1260 384633 Add a simple progress-reporting facility
1261 384987 VEX regalloc: allocate caller-save registers for short lived vregs
1262 385055 PPC VEX temporary storage exhausted
1263 385182 PPC64 is missing support for the DSCR
1264 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
1265 385207 PPC64, generate_store_FPRF() generates too many Iops
1266 385208 PPC64, xxperm instruction exhausts temporary memory
1267 385210 PPC64, vpermr instruction could exhaust temporary memory
1268 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
1269 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
1270 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
1271 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
1272 385408 s390x: z13 vector "support" instructions not implemented
1273 385409 s390x: z13 vector integer instructions not implemented
1274 385410 s390x: z13 vector string instructions not implemented
1275 385412 s390x: new non-vector z13 instructions not implemented
1276 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
1277 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
1278 385939 Optionally exit on the first error
1279 386318 valgrind.org/info/tools.html is missing SGCheck
1280 386425 running valgrind + wine on armv7l gives illegal opcode
1281 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
1282 387410 MIPSr6 support
1283 387664 Memcheck: make expensive-definedness-checks be the default
1284 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
1285 387766 asm shifts cause false positive "Conditional jump or move depends
1286 on uninitialised value"
1287 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
1288 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
1289 388786 Support bpf syscall in amd64 Linux
1290 388862 Add replacements for wmemchr and wcsnlen on Linux
1291 389065 valgrind meets gcc flag -Wlogical-op
1292 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
1293 390471 suppression by specification of source-file line number
1294 390723 make xtree dump files world wide readable, similar to log files
1295 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
1296 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
1297 392118 unhandled amd64-linux syscall: 332 (statx)
1298 392449 callgrind not clearing the number of calls properly
1299 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
1300 lxssp, stxssp and stxvl instructions.
1301 393023 callgrind_control risks using the wrong vgdb
1302 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
1303 393099 posix_memalign() invalid write if alignment == 0
1304 393146 failing assert "is_DebugInfo_active(di)"
1305 395709 PPC64 is missing support for the xvnegsp instruction
1306 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
1308 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
1309 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
1310 396839 s390x: Trap instructions not implemented
1311 396887 arch_prctl should return EINVAL on unknown option
1312 == 397286 crash before launching binary (Unsupported arch_prctl option)
1313 == 397393 valgrind: the 'impossible' happened: (Archlinux)
1314 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
1315 396906 compile tests failure on mips32-linux: broken inline asm in tests on
1317 397012 glibc ld.so uses arch_prctl on i386
1318 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
1319 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
1320 397424 glibc 2.27 and gdb_server tests
1321 398028 Assertion `cfsi_fits` failing in simple C program
1322 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
1324 n-i-bz Fix missing workq_ops operations (macOS)
1325 n-i-bz fix bug in strspn replacement
1326 n-i-bz Add support for the Linux BLKFLSBUF ioctl
1327 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
1328 n-i-bz Fix possible stack trashing by semctl syscall wrapping
1329 n-i-bz Add support for the Linux membarrier() system call
1330 n-i-bz x86 front end: recognise and handle UD2 correctly
1331 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
1332 correctly aligned before entering the handler.
1334 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
1335 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
1336 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
1340 Release 3.13.0 (15 June 2017)
1341 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1343 3.13.0 is a feature release with many improvements and the usual collection of
1346 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1347 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1348 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1349 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
1351 * ==================== CORE CHANGES ===================
1353 * The translation cache size has been increased to keep up with the demands of
1354 large applications. The maximum number of sectors has increased from 24 to
1355 48. The default number of sectors has increased from 16 to 32 on all
1356 targets except Android, where the increase is from 6 to 12.
1358 * The amount of memory that Valgrind can use has been increased from 64GB to
1359 128GB. In particular this means your application can allocate up to about
1360 60GB when running on Memcheck.
1362 * Valgrind's default load address has been changed from 0x3800'0000 to
1363 0x5800'0000, so as to make it possible to load larger executables. This
1364 should make it possible to load executables of size at least 1200MB.
1366 * A massive spaceleak caused by reading compressed debuginfo files has been
1367 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
1370 * The C++ demangler has been updated.
1372 * Support for demangling Rust symbols has been added.
1374 * A new representation of stack traces, the "XTree", has been added. An XTree
1375 is a tree of stacktraces with data associated with the stacktraces. This is
1376 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
1377 consumption of your program. Reporting is controlled by the new options
1378 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
1380 A report can also be produced on demand using the gdbserver monitor command
1381 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
1382 format' and 'massif format. The existing visualisers for these formats (e.g.
1383 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
1384 analyse these reports.
1386 Memcheck can also produce XTree leak reports using the Callgrind file
1387 format. For more details, see the user manual.
1389 * ================== PLATFORM CHANGES =================
1391 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
1393 * amd64: fixes for JIT failure problems on long AVX2 code blocks
1395 * amd64 and x86: support for CET prefixes has been added
1397 * arm32: a few missing ARMv8 instructions have been implemented
1399 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
1400 Store-Conditional instructions has been added. This is to deal with
1401 processor implementations that implement the LL/SC specifications strictly
1402 and as a result cause Valgrind to hang in certain situations. The
1403 alternative implementation is automatically enabled at startup, as required.
1404 You can use the option --sim-hints=fallback-llsc to force-enable it if you
1407 * Support for OSX 10.12 has been improved.
1409 * On Linux, clone handling has been improved to honour CLONE_VFORK that
1410 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
1411 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
1412 CLONE_VM exact semantics will (still) not work.
1414 * The TileGX/Linux port has been removed because it appears to be both unused
1417 * ==================== TOOL CHANGES ====================
1421 - Memcheck should give fewer false positives when running optimised
1422 Clang/LLVM generated code.
1424 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1426 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
1427 to produce the end of execution leak report in a xtree callgrind format
1430 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
1431 the leak report in an xtree file.
1435 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1437 - For some workloads (typically, for big applications), Massif memory
1438 consumption and CPU consumption has decreased significantly.
1442 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1444 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
1445 for Ada gnat compiled applications.
1447 * ==================== OTHER CHANGES ====================
1449 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
1450 append the inner guest stacktrace to the inner host stacktrace. This helps
1451 to investigate the errors reported by the outer, when they are caused by the
1452 inner guest program (such as an inner regtest). See README_DEVELOPERS for
1455 * To allow fast detection of callgrind files by desktop environments and file
1456 managers, the format was extended to have an optional first line that
1457 uniquely identifies the format ("# callgrind format"). Callgrind creates
1458 this line now, as does the new xtree functionality.
1460 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
1461 have a new %n format letter that is replaced by a sequence number.
1463 * "--version -v" now shows the SVN revision numbers from which Valgrind was
1466 * ==================== FIXED BUGS ====================
1468 The following bugs have been fixed or resolved. Note that "n-i-bz"
1469 stands for "not in bugzilla" -- that is, a bug that was reported to us
1470 but never got a bugzilla entry. We encourage you to file bugs in
1471 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1472 than mailing the developers (or mailing lists) directly -- bugs that
1473 are not entered into bugzilla tend to get forgotten about or ignored.
1475 To see details of a given bug, visit
1476 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1477 where XXXXXX is the bug number as listed below.
1479 162848 --log-file output isn't split when a program forks
1480 340777 Illegal instruction on mips (ar71xx)
1481 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
1482 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
1483 to a different stack.
1484 344139 x86 stack-seg overrides, needed by the Wine people
1485 344524 store conditional of guest applications always fail - observed on
1487 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
1488 352395 Please provide SVN revision info in --version -v
1489 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
1490 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
1491 INVALID_POSIX_THREADID' failed
1492 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
1493 barrier implementation
1494 358697 valgrind.h: Some code remains even when defining NVALGRIND
1495 359202 Add musl libc configure/compile
1496 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
1497 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
1498 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
1499 362223 assertion failed when .valgrindrc is a directory instead of a file
1500 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
1501 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
1502 368507 can't malloc chunks larger than about 34GB
1503 368529 Android arm target link error, missing atexit and pthread_atfork
1504 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
1505 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
1506 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
1507 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
1508 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
1509 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
1510 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
1511 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
1512 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
1513 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
1514 370028 Reduce the number of compiler warnings on MIPS platforms
1515 370635 arm64 missing syscall getcpu
1516 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
1517 371227 Clean AArch64 syscall table
1518 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
1519 371471 Valgrind complains about non legit memory leaks on placement new (C++)
1520 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
1521 371503 disInstr(arm64): unhandled instruction 0xF89F0000
1522 371869 support '%' in symbol Z-encoding
1523 371916 execution tree xtree concept
1524 372120 c++ demangler demangles symbols which are not c++
1525 372185 Support of valgrind on ARMv8 with 32 bit executable
1526 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1527 372195 Power PC, xxsel instruction is not always recognized.
1528 372504 Hanging on exit_group
1529 372600 process loops forever when fatal signals are arriving quickly
1530 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1531 373046 Stacks registered by core are never deregistered
1532 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1533 373086 Implement additional Xen hypercalls
1534 373192 Calling posix_spawn in glibc 2.24 completely broken
1535 373488 Support for fanotify API on ARM64 architecture
1536 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1537 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1538 373938 const IRExpr arguments for matchIRExpr()
1539 374719 some spelling fixes
1540 374963 increase valgrind's load address to prevent mmap failure
1541 375514 valgrind_get_tls_addr() does not work in case of static TLS
1542 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1543 for ML_(find_rx_mapping)()
1544 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1545 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1546 == 377159 "vex: the `impossible' happened" still present
1547 == 375150 Assertion 'tres.status == VexTransOK' failed
1548 == 378068 valgrind crashes on AVX2 function in FFmpeg
1549 376142 Segfaults on MIPS Cavium Octeon boards
1550 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1551 376455 Solaris: unhandled syscall lgrpsys(180)
1552 376518 Solaris: unhandled fast trap getlgrp(6)
1553 376611 ppc64 and arm64 don't know about prlimit64 syscall
1554 376729 PPC64, remove R2 from the clobber list
1556 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1557 to be wrongly marked as addressable
1558 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1559 PIE enabled by default
1560 377376 memcheck/tests/linux/getregset fails with glibc2.24
1561 377427 PPC64, lxv instruction failing on odd destination register
1562 377478 PPC64: ISA 3.0 setup fixes
1563 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1564 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1565 and 2 args for FUTEX_TRYLOCK_PI
1566 377717 Fix massive space leak when reading compressed debuginfo sections
1567 377891 Update Xen 4.6 domctl wrappers
1568 377930 fcntl syscall wrapper is missing flock structure check
1569 378524 libvexmultiarch_test regression on s390x and ppc64
1570 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1571 378673 Update libiberty demangler
1572 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1573 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1574 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1575 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1576 (task_register_dyld_image_infos)
1577 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1578 (task_register_dyld_shared_cache_image_info)
1579 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1580 379473 MIPS: add support for rdhwr cycle counter register
1581 379504 remove TileGX/Linux port
1582 379525 Support more x86 nop opcodes
1583 379838 disAMode(x86): not an addr!
1584 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1585 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1586 379895 clock_gettime does not execute POST syscall wrapper
1587 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1588 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1589 380200 xtree generated callgrind files refer to files without directory name
1590 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1591 380397 s390x: __GI_strcspn() replacement needed
1592 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1594 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1595 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1596 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1600 Release 3.12.0 (20 October 2016)
1601 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1603 3.12.0 is a feature release with many improvements and the usual
1604 collection of bug fixes.
1606 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1607 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1608 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1609 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1610 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1611 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1613 * ================== PLATFORM CHANGES =================
1615 * POWER: Support for ISA 3.0 has been added
1617 * mips: support for O32 FPXX ABI has been added.
1618 * mips: improved recognition of different processors
1619 * mips: determination of page size now done at run time
1621 * amd64: Partial support for AMD FMA4 instructions.
1623 * arm, arm64: Support for v8 crypto and CRC instructions.
1625 * Improvements and robustification of the Solaris port.
1627 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1629 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1630 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1631 where possible. Valgrind's support for 32-bit x86 has stagnated in
1632 recent years and has fallen far behind that for 64-bit x86
1633 instructions. By contrast 64-bit x86 is well supported, up to and
1636 * ==================== TOOL CHANGES ====================
1640 - Added meta mempool support for describing a custom allocator which:
1641 - Auto-frees all chunks assuming that destroying a pool destroys all
1643 - Uses itself to allocate other memory blocks
1645 - New flag --ignore-range-below-sp to ignore memory accesses below
1646 the stack pointer, if you really have to. The related flag
1647 --workaround-gcc296-bugs=yes is now deprecated. Use
1648 --ignore-range-below-sp=1024-1 as a replacement.
1652 - Improved thread startup time significantly on non-Linux platforms.
1656 - Added collection of the metric "tot-blocks-allocd"
1658 * ==================== OTHER CHANGES ====================
1660 * Replacement/wrapping of malloc/new related functions is now done not just
1661 for system libraries by default, but for any globally defined malloc/new
1662 related function (both in shared libraries and statically linked alternative
1663 malloc implementations). The dynamic (runtime) linker is excluded, though.
1664 To only intercept malloc/new related functions in
1665 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1666 "nouserintercepts" can be any non-existing library name).
1667 This new functionality is not implemented for MacOS X.
1669 * The maximum number of callers in a suppression entry is now equal to
1670 the maximum size for --num-callers (500).
1671 Note that --gen-suppressions=yes|all similarly generates suppressions
1672 containing up to --num-callers frames.
1674 * New and modified GDB server monitor features:
1676 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1677 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1680 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1681 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1684 * Valgrind is able to read compressed debuginfo sections in two formats:
1685 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1686 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1688 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1689 for the most common use case (x86_64-linux, Memcheck) has been
1692 * Improved performance for programs that do a lot of discarding of
1693 instruction address ranges of 8KB or less.
1695 * The C++ symbol demangler has been updated.
1697 * More robustness against invalid syscall parameters on Linux.
1699 * ==================== FIXED BUGS ====================
1701 The following bugs have been fixed or resolved. Note that "n-i-bz"
1702 stands for "not in bugzilla" -- that is, a bug that was reported to us
1703 but never got a bugzilla entry. We encourage you to file bugs in
1704 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1705 than mailing the developers (or mailing lists) directly -- bugs that
1706 are not entered into bugzilla tend to get forgotten about or ignored.
1708 To see details of a given bug, visit
1709 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1710 where XXXXXX is the bug number as listed below.
1712 191069 Exiting due to signal not reported in XML output
1713 199468 Suppressions: stack size limited to 25
1714 while --num-callers allows more frames
1715 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1716 278744 cvtps2pd with redundant RexW
1717 303877 valgrind doesn't support compressed debuginfo sections.
1718 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1719 348345 Assertion fails for negative lineno
1720 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1721 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1722 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1723 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1724 352197 mips: mmap2() not wrapped correctly for page size > 4096
1725 353083 arm64 doesn't implement various xattr system calls
1726 353084 arm64 doesn't support sigpending system call
1727 353137 www: update info for Supported Platforms
1728 353138 www: update "The Valgrind Developers" page
1729 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1732 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1733 353398 WARNING: unhandled amd64-solaris syscall: 207
1734 353660 XML in auxwhat tag not escaping reserved symbols properly
1735 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1736 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1737 353802 ELF debug info reader confused with multiple .rodata sections
1738 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1739 353917 unhandled amd64-solaris syscall fchdir(120)
1740 353920 unhandled amd64-solaris syscall: 170
1741 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1742 354392 unhandled amd64-solaris syscall: 171
1743 354797 Vbit test does not include Iops for Power 8 instruction support
1744 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1748 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1749 355188 valgrind should intercept all malloc related global functions
1750 355454 do not intercept malloc related symbols from the runtime linker
1751 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1752 356044 Dwarf line info reader misinterprets is_stmt register
1753 356112 mips: replace addi with addiu
1754 356393 valgrind (vex) crashes because isZeroU happened
1757 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1758 356678 arm64-linux: unhandled syscall 232 (mincore)
1759 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1760 356823 Unsupported ARM instruction: stlex
1761 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1762 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1763 357673 crash if I try to run valgrind with a binary link with libcurl
1764 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1765 357871 pthread_spin_destroy not properly wrapped
1766 357887 Calls to VG_(fclose) do not close the file descriptor
1767 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1768 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1769 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1770 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1771 359181 Buffer Overflow during Demangling
1772 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1773 359289 s390x: popcnt (B9E1) not implemented
1774 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1775 359503 Add missing syscalls for aarch64 (arm64)
1776 359645 "You need libc6-dbg" help message could be more helpful
1777 359703 s390: wire up separate socketcalls system calls
1778 359724 getsockname might crash - deref_UInt should call safe_to_deref
1779 359733 amd64 implement ld.so strchr/index override like x86
1780 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1781 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1783 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1784 359871 Incorrect mask handling in ppoll
1785 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1786 360008 Contents of Power vr registers contents is not printed correctly when
1787 the --vgdb-shadow-registers=yes option is used
1788 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1789 non-zero shadow bits
1790 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1791 360425 arm64 unsupported instruction ldpsw
1793 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1794 360571 Error about the Android Runtime reading below the stack pointer on ARM
1795 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1796 360749 kludge for multiple .rodata sections on Solaris no longer needed
1797 360752 raise the number of reserved fds in m_main.c from 10 to 12
1798 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1799 361226 s390x: risbgn (EC59) not implemented
1800 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1801 361354 ppc64[le]: wire up separate socketcalls system calls
1802 361615 Inconsistent termination for multithreaded process terminated by signal
1803 361926 Unhandled Solaris syscall: sysfs(84)
1804 362009 V dumps core on unimplemented functionality before threads are created
1805 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1806 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1807 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1808 362953 Request for an update to the Valgrind Developers page
1809 363680 add renameat2() support
1810 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1811 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1812 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1813 364058 clarify in manual limitations of array overruns detections
1814 364413 pselect sycallwrapper mishandles NULL sigmask
1815 364728 Power PC, missing support for several HW registers in
1816 get_otrack_shadow_offset_wrk()
1817 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1818 365273 Invalid write to stack location reported after signal handler runs
1819 365912 ppc64BE segfault during jm-insns test (RELRO)
1820 366079 FPXX Support for MIPS32 Valgrind
1821 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1822 366344 Multiple unhandled instruction for Aarch64
1823 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1824 367995 Integration of memcheck with custom memory allocator
1825 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1826 368412 False positive result for altivec capability check
1827 368416 Add tc06_two_races_xml.exp output for ppc64
1828 368419 Perf Events ioctls not implemented
1829 368461 mmapunmap test fails on ppc64
1830 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1831 369000 AMD64 fma4 instructions unsupported.
1832 369169 ppc64 fails jm_int_isa_2_07 test
1833 369175 jm_vec_isa_2_07 test crashes on ppc64
1834 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1835 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1836 369359 msghdr_foreachfield can crash when handling bad iovec
1837 369360 Bad sigprocmask old or new sets can crash valgrind
1838 369361 vmsplice syscall wrapper crashes on bad iovec
1839 369362 Bad sigaction arguments crash valgrind
1840 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1841 369402 Bad set/get_thread_area pointer crashes valgrind
1842 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1843 369446 valgrind crashes on unknown fcntl command
1844 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1845 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1846 370265 ISA 3.0 HW cap stuff needs updating
1847 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1848 372195 Power PC, xxsel instruction is not always recognized
1850 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1851 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1852 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1853 n-i-bz Fix ppoll_alarm exclusion on OS X
1854 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1855 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1856 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1857 n-i-bz amd64: memcheck false positive with shr %edx
1858 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1859 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1860 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1861 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1863 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1864 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1865 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1869 Release 3.11.0 (22 September 2015)
1870 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1872 3.11.0 is a feature release with many improvements and the usual
1873 collection of bug fixes.
1875 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1876 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1877 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1878 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1879 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1880 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1882 * ================== PLATFORM CHANGES =================
1884 * Support for Solaris/x86 and Solaris/amd64 has been added.
1886 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1888 * Preliminary support for the Tilera TileGX architecture has been added.
1890 * s390x: It is now required for the host to have the "long displacement"
1891 facility. The oldest supported machine model is z990.
1893 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1894 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1896 * The JIT's register allocator is significantly faster, making the JIT
1897 as a whole somewhat faster, so JIT-intensive activities, for example
1898 program startup, are modestly faster, around 5%.
1900 * There have been changes to the default settings of several command
1901 line flags, as detailed below.
1903 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1904 capable hosts, the simulated CPUID will now indicate AVX2 support.
1906 * ==================== TOOL CHANGES ====================
1910 - The default value for --leak-check-heuristics has been changed from
1911 "none" to "all". This helps to reduce the number of possibly
1912 lost blocks, in particular for C++ applications.
1914 - The default value for --keep-stacktraces has been changed from
1915 "malloc-then-free" to "malloc-and-free". This has a small cost in
1916 memory (one word per malloc-ed block) but allows Memcheck to show the
1917 3 stacktraces of a dangling reference: where the block was allocated,
1918 where it was freed, and where it is acccessed after being freed.
1920 - The default value for --partial-loads-ok has been changed from "no" to
1921 "yes", so as to avoid false positive errors resulting from some kinds
1922 of vectorised loops.
1924 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1925 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1926 than get_vbits when you need to associate byte data value with
1927 their corresponding validity bits.
1929 - The 'block_list' monitor command has been enhanced:
1930 o it can print a range of loss records
1931 o it now accepts an optional argument 'limited <max_blocks>'
1932 to control the number of blocks printed.
1933 o if a block has been found using a heuristic, then
1934 'block_list' now shows the heuristic after the block size.
1935 o the loss records/blocks to print can be limited to the blocks
1936 found via specified heuristics.
1938 - The C helper functions used to instrument loads on
1939 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1940 replaced by handwritten assembly sequences. This gives speedups
1941 in the region of 0% to 7% for those targets only.
1943 - A new command line option, --expensive-definedness-checks=yes|no,
1944 has been added. This is useful for avoiding occasional invalid
1945 uninitialised-value errors in optimised code. Watch out for
1946 runtime degradation, as this can be up to 25%. As always, though,
1947 the slowdown is highly application specific. The default setting
1952 - A new monitor command 'all_snapshots <filename>' dumps all
1953 snapshots taken so far.
1957 - Significant memory reduction and moderate speedups for
1958 --history-level=full for applications accessing a lot of memory
1959 with many different stacktraces.
1961 - The default value for --conflict-cache-size=N has been doubled to
1962 2000000. Users that were not using the default value should
1963 preferably also double the value they give.
1965 The default was changed due to the changes in the "full history"
1966 implementation. Doubling the value gives on average a slightly more
1967 complete history and uses similar memory (or significantly less memory
1968 in the worst case) than the previous implementation.
1970 - The Helgrind monitor command 'info locks' now accepts an optional
1971 argument 'lock_addr', which shows information about the lock at the
1974 - When using --history-level=full, the new Helgrind monitor command
1975 'accesshistory <addr> [<len>]' will show the recorded accesses for
1976 <len> (or 1) bytes at <addr>.
1978 * ==================== OTHER CHANGES ====================
1980 * The default value for the --smc-check option has been changed from
1981 "stack" to "all-non-file" on targets that provide automatic D-I
1982 cache coherence (x86, amd64 and s390x). The result is to provide,
1983 by default, transparent support for JIT generated and self-modifying
1984 code on all targets.
1986 * Mac OS X only: the default value for the --dsymutil option has been
1987 changed from "no" to "yes", since any serious usage on Mac OS X
1988 always required it to be "yes".
1990 * The command line options --db-attach and --db-command have been removed.
1991 They were deprecated in 3.10.0.
1993 * When a process dies due to a signal, Valgrind now shows the signal
1994 and the stacktrace at default verbosity (i.e. verbosity 1).
1996 * The address description logic used by Memcheck and Helgrind now
1997 describes addresses in anonymous segments, file mmap-ed segments,
1998 shared memory segments and the brk data segment.
2000 * The new option --error-markers=<begin>,<end> can be used to mark the
2001 begin/end of errors in textual output mode, to facilitate
2002 searching/extracting errors in output files that mix valgrind errors
2003 with program output.
2005 * The new option --max-threads=<number> can be used to change the number
2006 of threads valgrind can handle. The default is 500 threads which
2007 should be more than enough for most applications.
2009 * The new option --valgrind-stacksize=<number> can be used to change the
2010 size of the private thread stacks used by Valgrind. This is useful
2011 for reducing memory use or increasing the stack size if Valgrind
2012 segfaults due to stack overflow.
2014 * The new option --avg-transtab-entry-size=<number> can be used to specify
2015 the expected instrumented block size, either to reduce memory use or
2016 to avoid excessive retranslation.
2018 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
2020 * New and modified GDB server monitor features:
2022 - When a signal is reported in GDB, you can now use the GDB convenience
2023 variable $_siginfo to examine detailed signal information.
2025 - Valgrind's gdbserver now allows the user to change the signal
2026 to deliver to the process. So, use 'signal SIGNAL' to continue execution
2027 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
2028 continue without passing the signal to the process.
2030 - With GDB >= 7.10, the command 'target remote'
2031 will automatically load the executable file of the process running
2032 under Valgrind. This means you do not need to specify the executable
2033 file yourself, GDB will discover it itself. See GDB documentation about
2034 'qXfer:exec-file:read' packet for more info.
2036 * ==================== FIXED BUGS ====================
2038 The following bugs have been fixed or resolved. Note that "n-i-bz"
2039 stands for "not in bugzilla" -- that is, a bug that was reported to us
2040 but never got a bugzilla entry. We encourage you to file bugs in
2041 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2042 than mailing the developers (or mailing lists) directly -- bugs that
2043 are not entered into bugzilla tend to get forgotten about or ignored.
2045 To see details of a given bug, visit
2046 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2047 where XXXXXX is the bug number as listed below.
2049 116002 VG_(printf): Problems with justification of strings and integers
2050 155125 avoid cutting away file:lineno after long function name
2051 197259 Unsupported arch_prtctl PR_SET_GS option
2052 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
2053 201216 Fix Valgrind does not support pthread_sigmask() on OS X
2054 201435 Fix Darwin: -v does not show kernel version
2055 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
2056 211256 Fixed an outdated comment regarding the default platform.
2057 211529 Incomplete call stacks for code compiled by newer versions of MSVC
2058 211926 Avoid compilation warnings in valgrind.h with -pedantic
2059 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
2061 226609 Crediting upstream authors in man page
2062 231257 Valgrind omits path when executing script from shebang line
2063 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
2064 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
2065 269360 s390x: Fix addressing mode selection for compare-and-swap
2066 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
2068 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
2069 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
2070 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
2071 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
2072 330147 libmpiwrap PMPI_Get_count returns undefined value
2073 333051 mmap of huge pages fails due to incorrect alignment
2075 334802 valgrind does not always explain why a given option is bad
2076 335618 mov.w rN, pc/sp (ARM32)
2077 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
2082 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
2083 338602 AVX2 bit in CPUID missing
2084 338606 Strange message for scripts with invalid interpreter
2085 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
2086 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
2087 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
2089 339156 gdbsrv not called for fatal signal
2090 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
2091 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
2092 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
2093 339442 Fix testsuite build failure on OS X 10.9
2094 339542 Enable compilation with Intel's ICC compiler
2095 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
2096 339688 Mac-specific ASM does not support .version directive (cpuid,
2097 tronical and pushfpopf tests)
2098 339745 Valgrind crash when check Marmalade app (partial fix)
2099 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
2100 339778 Linux/TileGx platform support for Valgrind
2101 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
2102 339789 Fix none/tests/execve test on Mac OS X 10.9
2103 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
2104 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
2105 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
2106 340392 Allow user to select more accurate definedness checking in memcheck
2107 to avoid invalid complaints on optimised code
2108 340430 Fix some grammatical weirdness in the manual.
2109 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
2110 341419 Signal handler ucontext_t not filled out correctly on OS X
2111 341539 VG_(describe_addr) should not describe address as belonging to client
2112 segment if it is past the heap end
2113 341613 Enable building of manythreads and thread-exits tests on Mac OS X
2114 341615 Fix none/tests/darwin/access_extended test on Mac OS X
2115 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
2116 341789 aarch64: shmat fails with valgrind on ARMv8
2117 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
2118 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
2119 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
2120 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
2121 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
2122 342221 socket connect false positive uninit memory for unknown af family
2123 342353 Allow dumping full massif output while valgrind is still running
2124 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
2128 342603 Add I2C_SMBUS ioctl support
2129 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
2130 342683 Mark memory past the initial brk limit as unaddressable
2131 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
2132 342795 Internal glibc __GI_mempcpy call should be intercepted
2133 342841 s390x: Support instructions fiebr(a) and fidbr(a)
2134 343012 Unhandled syscall 319 (memfd_create)
2135 343069 Patch updating v4l2 API support
2136 343173 helgrind crash during stack unwind
2137 343219 fix GET_STARTREGS for arm
2138 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
2139 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
2140 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
2141 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
2142 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
2143 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
2144 343597 ppc64le: incorrect use of offseof macro
2145 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
2146 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
2147 343732 Unhandled syscall 144 (setgid) on aarch64
2148 343733 Unhandled syscall 187 (msgctl and related) on aarch64
2149 343802 s390x: False positive "conditional jump or move depends on [..]
2150 343902 --vgdb=yes doesn't break when --xml=yes is used
2151 343967 Don't warn about setuid/setgid/setcap executable for directories
2152 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
2153 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
2154 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
2155 344054 www - update info for Solaris/illumos
2156 344416 'make regtest' does not work cleanly on OS X
2157 344235 Remove duplicate include of pub_core_aspacemgr.h
2158 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
2159 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
2160 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
2161 344314 callgrind_annotate ... warnings about commands containing newlines
2162 344318 socketcall should wrap recvmmsg and sendmmsg
2163 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
2164 344416 Fix 'make regtest' does not work cleanly on OS X
2165 344499 Fix compilation for Linux kernel >= 4.0.0
2166 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
2167 unix:349 (__pthread_fchdir)
2168 344559 Garbage collection of unused segment names in address space manager
2169 344560 Fix stack traces missing penultimate frame on OS X
2170 344621 Fix memcheck/tests/err_disable4 test on OS X
2171 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
2172 344702 Fix missing libobjc suppressions on OS X 10.10
2174 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
2175 344939 Fix memcheck/tests/xml1 on OS X 10.10
2176 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
2177 345079 Fix build problems in VEX/useful/test_main.c
2178 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
2179 345177 arm64: prfm (reg) not implemented
2180 345215 Performance improvements for the register allocator
2181 345248 add support for Solaris OS in valgrind
2182 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
2183 345394 Fix memcheck/tests/strchr on OS X
2184 345637 Fix memcheck/tests/sendmsg on OS X
2185 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
2186 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
2187 345887 Fix an assertion in the address space manager
2188 345928 amd64: callstack only contains current function for small stacks
2189 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
2190 345987 MIPS64: Implement cavium LHX instruction
2191 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
2192 346185 Fix typo saving altivec register v24
2193 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
2194 and LibVEX_GuestPPC64_get_CR()
2195 346270 Regression tests none/tests/jm_vec/isa_2_07 and
2196 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
2197 346307 fuse filesystem syscall deadlocks
2198 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
2199 346411 MIPS: SysRes::_valEx handling is incorrect
2200 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
2201 346474 PPC64 Power 8, spr TEXASRU register not supported
2202 346487 Compiler generates "note" about a future ABI change for PPC64
2203 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
2204 and causing spurious "invalid read of size 8" warnings
2205 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
2206 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
2207 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
2208 347322 Power PC regression test cleanup
2209 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
2211 347389 unhandled syscall: 373 (Linux ARM syncfs)
2212 347686 Patch set to cleanup PPC64 regtests
2213 347978 Remove bash dependencies where not needed
2214 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
2215 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
2217 348102 Patch updating v4l2 API support
2218 348247 amd64 front end: jno jumps wrongly when overflow is not set
2219 348269 Improve mmap MAP_HUGETLB support.
2220 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
2221 348345 Assertion fails for negative lineno
2222 348377 Unsupported ARM instruction: yield
2223 348565 Fix detection of command line option availability for clang
2224 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
2225 348728 Fix broken check for VIDIOC_G_ENC_INDEX
2226 348748 Fix redundant condition
2227 348890 Fix clang warning about unsupported --param inline-unit-growth=900
2228 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
2229 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
2230 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
2231 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
2232 349626 Implemented additional Xen hypercalls
2233 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
2234 349790 Clean up of the hardware capability checking utilities.
2235 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
2236 349874 Fix typos in source code
2237 349879 memcheck: add handwritten assembly for helperc_LOADV*
2238 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
2239 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
2240 350202 Add limited param to 'monitor block_list'
2241 350290 s390x: Support instructions fixbr(a)
2242 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
2243 350809 Fix none/tests/async-sigs for Solaris
2244 350811 Remove reference to --db-attach which has been removed.
2245 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
2246 350854 hard-to-understand code in VG_(load_ELF)()
2247 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2248 351386 Solaris: Cannot run ld.so.1 under Valgrind
2249 351474 Fix VG_(iseqsigset) as obvious
2250 351531 Typo in /include/vki/vki-xen-physdev.h header guard
2251 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
2252 351858 ldsoexec support on Solaris
2253 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
2254 352130 helgrind reports false races for printfs using mempcpy on FILE* state
2255 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
2256 352320 arm64 crash on none/tests/nestedfs
2257 352765 Vbit test fails on Power 6
2258 352768 The mbar instruction is missing from the Power PC support
2259 352769 Power PC program priority register (PPR) is not supported
2260 n-i-bz Provide implementations of certain compiler builtins to support
2261 compilers that may not provide those
2262 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
2263 n-i-bz Fix compilation on distros with glibc < 2.5
2264 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
2265 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
2266 n-i-bz mremap did not work properly on shared memory
2267 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
2268 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
2270 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
2271 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
2272 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
2276 Release 3.10.1 (25 November 2014)
2277 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2278 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
2279 and backports fixes for all reported missing AArch64 ARMv8 instructions
2280 and syscalls from the trunk. If you package or deliver 3.10.0 for others
2281 to use, you might want to consider upgrading to 3.10.1 instead.
2283 The following bugs have been fixed or resolved. Note that "n-i-bz"
2284 stands for "not in bugzilla" -- that is, a bug that was reported to us
2285 but never got a bugzilla entry. We encourage you to file bugs in
2286 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2287 than mailing the developers (or mailing lists) directly -- bugs that
2288 are not entered into bugzilla tend to get forgotten about or ignored.
2290 To see details of a given bug, visit
2291 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2292 where XXXXXX is the bug number as listed below.
2294 335440 arm64: ld1 (single structure) is not implemented
2295 335713 arm64: unhanded instruction: prfm (immediate)
2296 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
2297 339182 ppc64: AvSplat ought to load destination vector register with [..]
2298 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
2299 339433 ppc64 lxvw4x instruction uses four 32-byte loads
2300 339645 Use correct tag names in sys_getdents/64 wrappers
2301 339706 Fix false positive for ioctl(TIOCSIG) on linux
2302 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
2303 339853 arm64 times syscall unknown
2304 339855 arm64 unhandled getsid/setsid syscalls
2305 339858 arm64 dmb sy not implemented
2306 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
2307 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
2308 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
2310 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
2311 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
2312 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
2313 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
2314 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
2315 340509 arm64: unhandled instruction fcvtas
2316 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
2317 340632 arm64: unhandled instruction fcvtas
2318 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
2319 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
2320 340788 warning: unhandled syscall: 318 (getrandom)
2321 340807 disInstr(arm): unhandled instruction: 0xEE989B20
2322 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
2323 340922 arm64: unhandled getgroups/setgroups syscalls
2324 350251 Fix typo in VEX utility program (test_main.c).
2325 350407 arm64: unhandled instruction ucvtf (vector, integer)
2326 350809 none/tests/async-sigs breaks when run under cron on Solaris
2327 350811 update README.solaris after r15445
2328 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
2329 350854 strange code in VG_(load_ELF)()
2330 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2331 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
2332 n-i-bz Add missing ]] to terminate CDATA.
2333 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
2334 n-i-bz Enable sys_fadvise64_64 on arm32.
2335 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
2336 n-i-bz Add test cases for all known arm64 load/store instructions.
2337 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
2338 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
2339 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
2340 n-i-bz arm64: Implement "fcvtpu w, s".
2341 n-i-bz arm64: implement ADDP and various others
2342 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
2343 n-i-bz arm64: enable FCVT{A,N}S X,S.
2345 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
2349 Release 3.10.0 (10 September 2014)
2350 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2352 3.10.0 is a feature release with many improvements and the usual
2353 collection of bug fixes.
2355 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
2356 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
2357 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
2358 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
2359 significantly improved relative to the 3.9.0 release.
2361 * ================== PLATFORM CHANGES =================
2363 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
2364 is mostly complete, and is usable, but some SIMD instructions are as
2367 * Support for little-endian variant of the 64-bit POWER architecture.
2369 * Support for Android on MIPS32.
2371 * Support for 64bit FPU on MIPS32 platforms.
2373 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
2375 * Configuration for and running on Android targets has changed.
2376 See README.android in the source tree for details.
2378 * ================== DEPRECATED FEATURES =================
2380 * --db-attach is now deprecated and will be removed in the next
2381 valgrind feature release. The built-in GDB server capabilities are
2382 superior and should be used instead. Learn more here:
2383 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
2385 * ==================== TOOL CHANGES ====================
2389 - Client code can now selectively disable and re-enable reporting of
2390 invalid address errors in specific ranges using the new client
2391 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
2392 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
2394 - Leak checker: there is a new leak check heuristic called
2395 "length64". This is used to detect interior pointers pointing 8
2396 bytes inside a block, on the assumption that the first 8 bytes
2397 holds the value "block size - 8". This is used by
2398 sqlite3MemMalloc, for example.
2400 - Checking of system call parameters: if a syscall parameter
2401 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
2402 several fields not initialised, an error is now reported for each
2403 field. Previously, an error was reported only for the first
2404 uninitialised field.
2406 - Mismatched alloc/free checking: a new flag
2407 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
2408 such checks if necessary.
2412 - Improvements to error messages:
2414 o Race condition error message involving heap allocated blocks also
2415 show the thread number that allocated the raced-on block.
2417 o All locks referenced by an error message are now announced.
2418 Previously, some error messages only showed the lock addresses.
2420 o The message indicating where a lock was first observed now also
2421 describes the address/location of the lock.
2423 - Helgrind now understands the Ada task termination rules and
2424 creates a happens-before relationship between a terminated task
2425 and its master. This avoids some false positives and avoids a big
2426 memory leak when a lot of Ada tasks are created and terminated.
2427 The interceptions are only activated with forthcoming releases of
2428 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
2430 - A new GDB server monitor command "info locks" giving the list of
2431 locks, their location, and their status.
2435 - callgrind_control now supports the --vgdb-prefix argument,
2436 which is needed if valgrind was started with this same argument.
2438 * ==================== OTHER CHANGES ====================
2440 * Unwinding through inlined function calls. Stack unwinding can now
2441 make use of Dwarf3 inlined-unwind information if it is available.
2442 The practical effect is that inlined calls become visible in stack
2443 traces. The suppression matching machinery has been adjusted
2444 accordingly. This is controlled by the new option
2445 --read-inline-info=yes|no. Currently this is enabled by default
2446 only on Linux and Android targets and only for the tools Memcheck,
2449 * Valgrind can now read EXIDX unwind information on 32-bit ARM
2450 targets. If an object contains both CFI and EXIDX unwind
2451 information, Valgrind will prefer the CFI over the EXIDX. This
2452 facilitates unwinding through system libraries on arm-android
2455 * Address description logic has been improved and is now common
2456 between Memcheck and Helgrind, resulting in better address
2457 descriptions for some kinds of error messages.
2459 * Error messages about dubious arguments (eg, to malloc or calloc) are
2460 output like other errors. This means that they can be suppressed
2461 and they have a stack trace.
2463 * The C++ demangler has been updated for better C++11 support.
2465 * New and modified GDB server monitor features:
2467 - Thread local variables/storage (__thread) can now be displayed.
2469 - The GDB server monitor command "v.info location <address>"
2470 displays information about an address. The information produced
2471 depends on the tool and on the options given to valgrind.
2472 Possibly, the following are described: global variables, local
2473 (stack) variables, allocated or freed blocks, ...
2475 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
2476 ask the GDB server to stop at the start of program execution, at
2477 the end of the program execution and on Valgrind internal errors.
2479 - A new monitor command "v.info stats" shows various Valgrind core
2480 and tool statistics.
2482 - A new monitor command "v.set hostvisibility" allows the GDB server
2483 to provide access to Valgrind internal host status/memory.
2485 * A new option "--aspace-minaddr=<address>" can in some situations
2486 allow the use of more memory by decreasing the address above which
2487 Valgrind maps memory. It can also be used to solve address
2488 conflicts with system libraries by increasing the default value.
2489 See user manual for details.
2491 * The amount of memory used by Valgrind to store debug info (unwind
2492 info, line number information and symbol data) has been
2493 significantly reduced, even though Valgrind now reads more
2494 information in order to support unwinding of inlined function calls.
2496 * Dwarf3 handling with --read-var-info=yes has been improved:
2498 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
2500 - Code compiled with
2501 -ffunction-sections -fdata-sections -Wl,--gc-sections
2502 no longer causes assertion failures.
2504 * Improved checking for the --sim-hints= and --kernel-variant=
2505 options. Unknown strings are now detected and reported to the user
2508 * The semantics of stack start/end boundaries in the valgrind.h
2509 VALGRIND_STACK_REGISTER client request has been clarified and
2510 documented. The convention is that start and end are respectively
2511 the lowest and highest addressable bytes of the stack.
2513 * ==================== FIXED BUGS ====================
2515 The following bugs have been fixed or resolved. Note that "n-i-bz"
2516 stands for "not in bugzilla" -- that is, a bug that was reported to us
2517 but never got a bugzilla entry. We encourage you to file bugs in
2518 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2519 than mailing the developers (or mailing lists) directly -- bugs that
2520 are not entered into bugzilla tend to get forgotten about or ignored.
2522 To see details of a given bug, visit
2523 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2524 where XXXXXX is the bug number as listed below.
2526 175819 Support for ipv6 socket reporting with --track-fds
2527 232510 make distcheck fails
2528 249435 Analyzing wine programs with callgrind triggers a crash
2529 278972 support for inlined function calls in stacktraces and suppression
2531 291310 FXSAVE instruction marks memory as undefined on amd64
2532 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2533 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2534 315199 vgcore file for threaded app does not show which thread crashed
2535 315952 tun/tap ioctls are not supported
2536 323178 Unhandled instruction: PLDW register (ARM)
2537 323179 Unhandled instruction: PLDW immediate (ARM)
2538 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2539 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2540 325124 [MIPSEL] Compilation error
2541 325477 Phase 4 support for IBM Power ISA 2.07
2542 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2543 325628 Phase 5 support for IBM Power ISA 2.07
2544 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2545 325751 Missing the two privileged Power PC Transactional Memory Instructions
2546 325816 Phase 6 support for IBM Power ISA 2.07
2547 325856 Make SGCheck fail gracefully on unsupported platforms
2548 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2549 326436 DRD: False positive in libstdc++ std::list::push_back
2550 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2551 326462 Refactor vgdb to isolate invoker stuff into separate module
2552 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2553 326623 DRD: false positive conflict report in a field assignment
2554 326724 Valgrind does not compile on OSX 1.9 Mavericks
2555 326816 Intercept for __strncpy_sse2_unaligned missing?
2556 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2557 326983 Clear direction flag after tests on amd64.
2558 327212 Do not prepend the current directory to absolute path names.
2559 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2560 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2561 327284 s390x: Fix translation of the risbg instruction
2562 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2563 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2564 327916 DW_TAG_typedef may have no name
2565 327943 s390x: add a redirection for the 'index' function
2566 328100 XABORT not implemented
2567 328205 Implement additional Xen hypercalls
2568 328454 add support Backtraces with ARM unwind tables (EXIDX)
2569 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2570 328711 valgrind.1 manpage "memcheck options" section is badly generated
2571 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2572 329612 Incorrect handling of AT_BASE for image execution
2573 329694 clang warns about using uninitialized variable
2574 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2575 330228 mmap must align to VKI_SHMLBA on mips32
2576 330257 LLVM does not support `-mno-dynamic-no-pic` option
2577 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2578 330459 --track-fds=yes doesn't track eventfds
2579 330469 Add clock_adjtime syscall support
2580 330594 Missing sysalls on PowerPC / uClibc
2581 330622 Add test to regression suite for POWER instruction: dcbzl
2582 330939 Support for AMD's syscall instruction on x86
2584 330941 Typo in PRE(poll) syscall wrapper
2585 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2586 331254 Fix expected output for memcheck/tests/dw4
2587 331255 Fix race condition in test none/tests/coolo_sigaction
2588 331257 Fix type of jump buffer in test none/tests/faultstatus
2589 331305 configure uses bash specific syntax
2590 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2591 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2592 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2593 331829 Unexpected ioctl opcode sign extension
2594 331830 ppc64: WARNING: unhandled syscall: 96/97
2595 331839 drd/tests/sem_open specifies invalid semaphore name
2596 331847 outcome of drd/tests/thread_name is nondeterministic
2597 332037 Valgrind cannot handle Thumb "add pc, reg"
2598 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2599 consistency checks enabled
2600 332263 intercepts for pthread_rwlock_timedrdlock and
2601 pthread_rwlock_timedwrlock are incorrect
2602 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2604 332276 Implement additional Xen hypercalls
2605 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2606 332765 Fix ms_print to create temporary files in a proper directory
2607 333072 drd: Add semaphore annotations
2608 333145 Tests for missaligned PC+#imm access for arm
2609 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2610 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2611 333248 WARNING: unhandled syscall: unix:443
2612 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2613 333501 cachegrind: assertion: Cache set count is not a power of two.
2616 333666 Recognize MPX instructions and bnd prefix.
2617 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2618 333817 Valgrind reports the memory areas written to by the SG_IO
2620 334049 lzcnt fails silently (x86_32)
2621 334384 Valgrind does not have support Little Endian support for
2623 334585 recvmmsg unhandled (+patch) (arm)
2624 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2625 334727 Build fails with -Werror=format-security
2626 334788 clarify doc about --log-file initial program directory
2627 334834 PPC64 Little Endian support, patch 2
2628 334836 PPC64 Little Endian support, patch 3 testcase fixes
2629 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2630 335034 Unhandled ioctl: HCIGETDEVLIST
2631 335155 vgdb, fix error print statement.
2632 335262 arm64: movi 8bit version is not supported
2633 335263 arm64: dmb instruction is not implemented
2634 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2635 335496 arm64: sbc/abc instructions are not implemented
2636 335554 arm64: unhandled instruction: abs
2637 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2638 335735 arm64: unhandled instruction: cnt
2639 335736 arm64: unhandled instruction: uaddlv
2640 335848 arm64: unhandled instruction: {s,u}cvtf
2641 335902 arm64: unhandled instruction: sli
2642 335903 arm64: unhandled instruction: umull (vector)
2643 336055 arm64: unhandled instruction: mov (element)
2644 336062 arm64: unhandled instruction: shrn{,2}
2645 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2646 336189 arm64: unhandled Instruction: mvn
2647 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2648 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2649 336772 Make moans about unknown ioctls more informative
2650 336957 Add a section about the Solaris/illumos port on the webpage
2651 337094 ifunc wrapper is broken on ppc64
2652 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2653 337528 leak check heuristic for block prefixed by length as 64bit number
2654 337740 Implement additional Xen hypercalls
2655 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2656 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2657 337871 deprecate --db-attach
2658 338023 Add support for all V4L2/media ioctls
2659 338024 inlined functions are not shown if DW_AT_ranges is used
2660 338106 Add support for 'kcmp' syscall
2661 338115 DRD: computed conflict set differs from actual after fork
2662 338160 implement display of thread local storage in gdbsrv
2663 338205 configure.ac and check for -Wno-tautological-compare
2664 338300 coredumps are missing one byte of every segment
2665 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2666 338499 --sim-hints parsing broken due to wrong order in tokens
2667 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2668 338681 Unable to unwind through clone thread created on i386-linux
2669 338698 race condition between gdbsrv and vgdb on startup
2670 338703 helgrind on arm-linux gets false positives in dynamic loader
2671 338791 alt dwz files can be relative of debug/main file
2672 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2673 338932 build V-trunk with gcc-trunk
2674 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2675 345079 Fix build problems in VEX/useful/test_main.c
2676 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2677 n-i-bz s390x: Fix memory corruption for multithreaded applications
2678 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2679 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2680 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2681 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2682 n-i-bz Add command-line option --default-suppressions
2683 n-i-bz Add support for BLKDISCARDZEROES ioctl
2684 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2685 n-i-bz Add support for sys_pivot_root and sys_unshare
2687 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2688 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2689 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2693 Release 3.9.0 (31 October 2013)
2694 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2695 3.9.0 is a feature release with many improvements and the usual
2696 collection of bug fixes.
2698 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2699 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2700 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2701 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2703 * ================== PLATFORM CHANGES =================
2705 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2706 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2708 * Support for MIPS DSP ASE on MIPS32 platforms.
2710 * Support for s390x Decimal Floating Point instructions on hosts that
2711 have the DFP facility installed.
2713 * Support for POWER8 (Power ISA 2.07) instructions
2715 * Support for Intel AVX2 instructions. This is available only on 64
2718 * Initial support for Intel Transactional Synchronization Extensions,
2721 * Initial support for Hardware Transactional Memory on POWER.
2723 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2724 run large GUI apps tolerably well.
2726 * ==================== TOOL CHANGES ====================
2730 - Improvements in handling of vectorised code, leading to
2731 significantly fewer false error reports. You need to use the flag
2732 --partial-loads-ok=yes to get the benefits of these changes.
2734 - Better control over the leak checker. It is now possible to
2735 specify which leak kinds (definite/indirect/possible/reachable)
2736 should be displayed, which should be regarded as errors, and which
2737 should be suppressed by a given leak suppression. This is done
2738 using the options --show-leak-kinds=kind1,kind2,..,
2739 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2740 "match-leak-kinds:" line in suppression entries, respectively.
2742 Note that generated leak suppressions contain this new line and
2743 are therefore more specific than in previous releases. To get the
2744 same behaviour as previous releases, remove the "match-leak-kinds:"
2745 line from generated suppressions before using them.
2747 - Reduced "possible leak" reports from the leak checker by the use
2748 of better heuristics. The available heuristics provide detection
2749 of valid interior pointers to std::stdstring, to new[] allocated
2750 arrays with elements having destructors and to interior pointers
2751 pointing to an inner part of a C++ object using multiple
2752 inheritance. They can be selected individually using the
2753 option --leak-check-heuristics=heur1,heur2,...
2755 - Better control of stacktrace acquisition for heap-allocated
2756 blocks. Using the --keep-stacktraces option, it is possible to
2757 control independently whether a stack trace is acquired for each
2758 allocation and deallocation. This can be used to create better
2759 "use after free" errors or to decrease Valgrind's resource
2760 consumption by recording less information.
2762 - Better reporting of leak suppression usage. The list of used
2763 suppressions (shown when the -v option is given) now shows, for
2764 each leak suppressions, how many blocks and bytes it suppressed
2765 during the last leak search.
2769 - False errors resulting from the use of statically initialised
2770 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2773 - False errors resulting from the use of pthread_cond_waits that
2774 timeout, have been removed.
2776 * ==================== OTHER CHANGES ====================
2778 * Some attempt to tune Valgrind's space requirements to the expected
2779 capabilities of the target:
2781 - The default size of the translation cache has been reduced from 8
2782 sectors to 6 on Android platforms, since each sector occupies
2783 about 40MB when using Memcheck.
2785 - The default size of the translation cache has been increased to 16
2786 sectors on all other platforms, reflecting the fact that large
2787 applications require instrumentation and storage of huge amounts
2788 of code. For similar reasons, the number of memory mapped
2789 segments that can be tracked has been increased by a factor of 6.
2791 - In all cases, the maximum number of sectors in the translation
2792 cache can be controlled by the new flag --num-transtab-sectors.
2794 * Changes in how debug info (line numbers, etc) is read:
2796 - Valgrind no longer temporarily mmaps the entire object to read
2797 from it. Instead, reading is done through a small fixed sized
2798 buffer. This avoids virtual memory usage spikes when Valgrind
2799 reads debuginfo from large shared objects.
2801 - A new experimental remote debug info server. Valgrind can read
2802 debug info from a different machine (typically, a build host)
2803 where debuginfo objects are stored. This can save a lot of time
2804 and hassle when running Valgrind on resource-constrained targets
2805 (phones, tablets) when the full debuginfo objects are stored
2806 somewhere else. This is enabled by the --debuginfo-server=
2809 - Consistency checking between main and debug objects can be
2810 disabled using the --allow-mismatched-debuginfo option.
2812 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2813 scanning can recover stack traces in some cases when the normal
2814 unwind mechanisms fail. Stack scanning is best described as "a
2815 nasty, dangerous and misleading hack" and so is disabled by default.
2816 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2819 * Detection and merging of recursive stack frame cycles. When your
2820 program has recursive algorithms, this limits the memory used by
2821 Valgrind for recorded stack traces and avoids recording
2822 uninteresting repeated calls. This is controlled by the command
2823 line option --merge-recursive-frame and by the monitor command
2824 "v.set merge-recursive-frames".
2826 * File name and line numbers for used suppressions. The list of used
2827 suppressions (shown when the -v option is given) now shows, for each
2828 used suppression, the file name and line number where the suppression
2831 * New and modified GDB server monitor features:
2833 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2834 that can be used to execute gdbserver monitor commands from the
2837 - A new monitor command, "v.info open_fds", that gives the list of
2838 open file descriptors and additional details.
2840 - An optional message in the "v.info n_errs_found" monitor command,
2841 for example "v.info n_errs_found test 1234 finished", allowing a
2842 comment string to be added to the process output, perhaps for the
2843 purpose of separating errors of different tests or test phases.
2845 - A new monitor command "v.info execontext" that shows information
2846 about the stack traces recorded by Valgrind.
2848 - A new monitor command "v.do expensive_sanity_check_general" to run
2849 some internal consistency checks.
2851 * New flag --sigill-diagnostics to control whether a diagnostic
2852 message is printed when the JIT encounters an instruction it can't
2853 translate. The actual behavior -- delivery of SIGILL to the
2854 application -- is unchanged.
2856 * The maximum amount of memory that Valgrind can use on 64 bit targets
2857 has been increased from 32GB to 64GB. This should make it possible
2858 to run applications on Memcheck that natively require up to about 35GB.
2860 * ==================== FIXED BUGS ====================
2862 The following bugs have been fixed or resolved. Note that "n-i-bz"
2863 stands for "not in bugzilla" -- that is, a bug that was reported to us
2864 but never got a bugzilla entry. We encourage you to file bugs in
2865 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2866 than mailing the developers (or mailing lists) directly -- bugs that
2867 are not entered into bugzilla tend to get forgotten about or ignored.
2869 To see details of a given bug, visit
2870 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2871 where XXXXXX is the bug number as listed below.
2873 123837 system call: 4th argument is optional, depending on cmd
2874 135425 memcheck should tell you where Freed blocks were Mallocd
2875 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2876 207815 Adds some of the drm ioctls to syswrap-linux.c
2877 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2878 252955 Impossible to compile with ccache
2879 253519 Memcheck reports auxv pointer accesses as invalid reads.
2880 263034 Crash when loading some PPC64 binaries
2881 269599 Increase deepest backtrace
2882 274695 s390x: Support "compare to/from logical" instructions (z196)
2883 275800 s390x: Autodetect cache info (part 2)
2884 280271 Valgrind reports possible memory leaks on still-reachable std::string
2885 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2886 289578 Backtraces with ARM unwind tables (stack scan flags)
2887 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2888 304832 ppc32: build failure
2889 305431 Use find_buildid shdr fallback for separate .debug files
2890 305728 Add support for AVX2 instructions
2891 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2892 306035 s390x: Fix IR generation for LAAG and friends
2893 306054 s390x: Condition code computation for convert-to-int/logical
2894 306098 s390x: alternate opcode form for convert to/from fixed
2895 306587 Fix cache line detection from auxiliary vector for PPC.
2896 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2897 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2898 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2899 307101 sys_capget second argument can be NULL
2900 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2901 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2902 307113 s390x: DFP support
2903 307141 valgrind does't work in mips-linux system
2904 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2905 307285 x86_amd64 feature test for avx in test suite is wrong
2906 307290 memcheck overlap testcase needs memcpy version filter
2907 307463 Please add "&limit=0" to the "all open bugs" link
2908 307465 --show-possibly-lost=no should reduce the error count / exit code
2909 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2910 307729 pkgconfig support broken valgrind.pc
2911 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2912 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2913 308089 Unhandled syscall on ppc64: prctl
2914 308135 PPC32 MPC8xx has 16 bytes cache size
2915 308321 testsuite memcheck filter interferes with gdb_filter
2917 308341 vgdb should report process exit (or fatal signal)
2918 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2919 308495 Remove build dependency on installed Xen headers
2920 308573 Internal error on 64-bit instruction executed in 32-bit mode
2922 308627 pmovmskb validity bit propagation is imprecise
2923 308644 vgdb command for having the info for the track-fds option
2924 308711 give more info about aspacemgr and arenas in out_of_memory
2925 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2926 308718 ARM implement SMLALBB family of instructions
2927 308886 Missing support for PTRACE_SET/GETREGSET
2928 308930 syscall name_to_handle_at (303 on amd64) not handled
2929 309229 V-bit tester does not report number of tests generated
2930 309323 print unrecognized instuction on MIPS
2931 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2932 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2933 309430 Self hosting ppc64 encounters a vassert error on operand type
2934 309600 valgrind is a bit confused about 0-sized sections
2935 309823 Generate errors for still reachable blocks
2936 309921 PCMPISTRI validity bit propagation is imprecise
2937 309922 none/tests/ppc64/test_dfp5 sometimes fails
2938 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2939 310424 --read-var-info does not properly describe static variables
2940 310792 search additional path for debug symbols
2941 310931 s390x: Message-security assist (MSA) instruction extension [..]
2942 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2943 311318 ARM: "128-bit constant is not implemented" error message
2944 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2945 311690 V crashes because it redirects branches inside of a redirected function
2946 311880 x86_64: make regtest hangs at shell_valid1
2947 311922 WARNING: unhandled syscall: 170
2949 312171 ppc: insn selection for DFP
2950 312571 Rounding mode call wrong for the DFP Iops [..]
2951 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2952 312913 Dangling pointers error should also report the alloc stack trace
2953 312980 Building on Mountain Lion generates some compiler warnings
2954 313267 Adding MIPS64/Linux port to Valgrind
2957 313811 Buffer overflow in assert_fail
2958 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2959 314269 ppc: dead code in insn selection
2960 314718 ARM: implement integer divide instruction (sdiv and udiv)
2961 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2962 315441 sendmsg syscall should ignore unset msghdr msg_flags
2963 315534 msgrcv inside a thread causes valgrind to hang (block)
2964 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2965 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2966 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2967 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2968 316144 valgrind.1 manpage contains unknown ??? strings [..]
2969 316145 callgrind command line options in manpage reference (unknown) [..]
2970 316145 callgrind command line options in manpage reference [..]
2971 316181 drd: Fixed a 4x slowdown for certain applications
2972 316503 Valgrind does not support SSE4 "movntdqa" instruction
2973 316535 Use of |signed int| instead of |size_t| in valgrind messages
2974 316696 fluidanimate program of parsec 2.1 stuck
2975 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2976 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2977 317186 "Impossible happens" when occurs VCVT instruction on ARM
2978 317318 Support for Threading Building Blocks "scalable_malloc"
2979 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2980 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2981 317463 bmi testcase IR SANITY CHECK FAILURE
2982 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2983 318050 libmpiwrap fails to compile with out-of-source build
2984 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2985 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2986 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2987 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2988 318932 Add missing PPC64 and PPC32 system call support
2989 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2990 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2991 319494 VEX Makefile-gcc standalone build update after r2702
2992 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2993 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2994 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2995 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2996 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2997 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2998 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2999 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
3000 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
3001 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
3002 320895 add fanotify support (patch included)
3003 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
3004 321065 Valgrind updates for Xen 4.3
3005 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
3006 321363 Unhandled instruction: SSAX (ARM + Thumb)
3007 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
3008 321466 Unhandled instruction: SHASX (ARM + Thumb)
3009 321467 Unhandled instruction: SHSAX (ARM + Thumb)
3010 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
3011 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
3012 321620 Unhandled instruction: UASX (ARM + Thumb)
3013 321621 Unhandled instruction: USAX (ARM + Thumb)
3014 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
3015 321693 Unhandled instruction: LDRSBT (Thumb)
3016 321694 Unhandled instruction: UQASX (ARM + Thumb)
3017 321696 Unhandled instruction: UQSAX (Thumb + ARM)
3018 321697 Unhandled instruction: UHASX (ARM + Thumb)
3019 321703 Unhandled instruction: UHSAX (ARM + Thumb)
3020 321704 Unhandled instruction: REVSH (ARM + Thumb)
3021 321730 Add cg_diff and cg_merge man pages
3022 321738 Add vgdb and valgrind-listener man pages
3024 321891 Unhandled instruction: LDRHT (Thumb)
3025 321960 pthread_create() then alloca() causing invalid stack write errors
3026 321969 ppc32 and ppc64 don't support [lf]setxattr
3027 322254 Show threadname together with tid if set by application
3028 322294 Add initial support for IBM Power ISA 2.07
3029 322368 Assertion failure in wqthread_hijack under OS X 10.8
3030 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
3031 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
3032 322851 0bXXX binary literal syntax is not standard
3033 323035 Unhandled instruction: LDRSHT(Thumb)
3034 323036 Unhandled instruction: SMMLS (ARM and Thumb)
3035 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
3036 323175 Unhandled instruction: SMLALD (ARM + Thumb)
3037 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
3038 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
3039 323437 Phase 2 support for IBM Power ISA 2.07
3040 323713 Support mmxext (integer sse) subset on i386 (athlon)
3041 323803 Transactional memory instructions are not supported for Power
3042 323893 SSE3 not available on amd cpus in valgrind
3043 323905 Probable false positive from Valgrind/drd on close()
3044 323912 valgrind.h header isn't compatible for mingw64
3045 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
3046 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
3047 324181 mmap does not handle MAP_32BIT
3048 324227 memcheck false positive leak when a thread calls exit+block [..]
3049 324421 Support for fanotify API on ARM architecture
3050 324514 gdbserver monitor cmd output behaviour consistency [..]
3051 324518 ppc64: Emulation of dcbt instructions does not handle [..]
3052 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
3053 324582 When access is made to freed memory, report both allocation [..]
3054 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
3055 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
3056 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
3057 324834 Unhandled instructions in Microsoft C run-time for x86_64
3058 324894 Phase 3 support for IBM Power ISA 2.07
3059 326091 drd: Avoid false race reports from optimized strlen() impls
3060 326113 valgrind libvex hwcaps error on AMD64
3061 n-i-bz Some wrong command line options could be ignored
3062 n-i-bz patch to allow fair-sched on android
3063 n-i-bz report error for vgdb snapshot requested before execution
3064 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
3066 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
3070 Release 3.8.1 (19 September 2012)
3071 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3072 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
3073 that occur moderately frequently in real use cases, adds support for
3074 some missing instructions on ARM, and fixes a deadlock condition on
3075 MacOSX. If you package or deliver 3.8.0 for others to use, you might
3076 want to consider upgrading to 3.8.1 instead.
3078 The following bugs have been fixed or resolved. Note that "n-i-bz"
3079 stands for "not in bugzilla" -- that is, a bug that was reported to us
3080 but never got a bugzilla entry. We encourage you to file bugs in
3081 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3082 than mailing the developers (or mailing lists) directly -- bugs that
3083 are not entered into bugzilla tend to get forgotten about or ignored.
3085 To see details of a given bug, visit
3086 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3087 where XXXXXX is the bug number as listed below.
3090 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
3091 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
3092 298281 wcslen causes false(?) uninitialised value warnings
3093 301281 valgrind hangs on OS X when the process calls system()
3094 304035 disInstr(arm): unhandled instruction 0xE1023053
3095 304867 implement MOVBE instruction in x86 mode
3096 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
3097 305042 amd64: implement 0F 7F encoding of movq between two registers
3098 305199 ARM: implement QDADD and QDSUB
3099 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
3100 305513 killed by fatal signal: SIGSEGV
3101 305690 DRD reporting invalid semaphore when sem_trywait fails
3102 305926 Invalid alignment checks for some AVX instructions
3103 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
3104 306310 3.8.0 release tarball missing some files
3105 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
3106 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
3107 n-i-bz shmat of a segment > 4Gb does not work
3108 n-i-bz simulate_control_c script wrong USR1 signal number on mips
3109 n-i-bz vgdb ptrace calls wrong on mips [...]
3110 n-i-bz Fixes for more MPI false positives
3111 n-i-bz exp-sgcheck's memcpy causes programs to segfault
3112 n-i-bz OSX build w/ clang: asserts at startup
3113 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
3114 n-i-bz fix a couple of union tag-vs-field mixups
3115 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
3117 The following bugs were fixed in 3.8.0 but not listed in this NEWS
3120 254088 Valgrind should know about UD2 instruction
3123 304754 NEWS blows TeX's little mind
3125 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
3129 Release 3.8.0 (10 August 2012)
3130 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3131 3.8.0 is a feature release with many improvements and the usual
3132 collection of bug fixes.
3134 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3135 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
3136 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
3137 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
3138 There is initial support for MacOSX 10.8, but it is not usable for
3139 serious work at present.
3141 * ================== PLATFORM CHANGES =================
3143 * Support for MIPS32 platforms running Linux. Valgrind has been
3144 tested on MIPS32 and MIPS32r2 platforms running different Debian
3145 Squeeze and MeeGo distributions. Both little-endian and big-endian
3146 cores are supported. The tools Memcheck, Massif and Lackey have
3147 been tested and are known to work. See README.mips for more details.
3149 * Preliminary support for Android running on x86.
3151 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
3153 * Support for Intel AVX instructions and for AES instructions. This
3154 support is available only for 64 bit code.
3156 * Support for POWER Decimal Floating Point instructions.
3158 * ==================== TOOL CHANGES ====================
3160 * Non-libc malloc implementations are now supported. This is useful
3161 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
3162 Using the new option --soname-synonyms, such tools can be informed
3163 that the malloc implementation is either linked statically into the
3164 executable, or is present in some other shared library different
3165 from libc.so. This makes it possible to process statically linked
3166 programs, and programs using other malloc libraries, for example
3167 TCMalloc or JEMalloc.
3169 * For tools that provide their own replacement for malloc et al, the
3170 option --redzone-size=<number> allows users to specify the size of
3171 the padding blocks (redzones) added before and after each client
3172 allocated block. Smaller redzones decrease the memory needed by
3173 Valgrind. Bigger redzones increase the chance to detect blocks
3174 overrun or underrun. Prior to this change, the redzone size was
3175 hardwired to 16 bytes in Memcheck.
3179 - The leak_check GDB server monitor command now can
3180 control the maximum nr of loss records to output.
3182 - Reduction of memory use for applications allocating
3183 many blocks and/or having many partially defined bytes.
3185 - Addition of GDB server monitor command 'block_list' that lists
3186 the addresses/sizes of the blocks of a leak search loss record.
3188 - Addition of GDB server monitor command 'who_points_at' that lists
3189 the locations pointing at a block.
3191 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
3192 detect an invalid access of these redzones, by marking them
3193 noaccess. Similarly, if a redzone size is given for a memory
3194 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
3195 This still allows to find some bugs if the user has forgotten to
3196 mark the pool superblock noaccess.
3198 - Performance of memory leak check has been improved, especially in
3199 cases where there are many leaked blocks and/or many suppression
3200 rules used to suppress leak reports.
3202 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
3203 more precise analysis, which is important for LLVM/Clang
3204 generated code. This is at the cost of somewhat reduced
3205 performance. Note there is no change to analysis precision or
3206 costs on Linux targets.
3210 - Added even more facilities that can help finding the cause of a data
3211 race, namely the command-line option --ptrace-addr and the macro
3212 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
3214 - Fixed a subtle bug that could cause false positive data race reports.
3216 * ==================== OTHER CHANGES ====================
3218 * The C++ demangler has been updated so as to work well with C++
3219 compiled by up to at least g++ 4.6.
3221 * Tool developers can make replacement/wrapping more flexible thanks
3222 to the new option --soname-synonyms. This was reported above, but
3223 in fact is very general and applies to all function
3224 replacement/wrapping, not just to malloc-family functions.
3226 * Round-robin scheduling of threads can be selected, using the new
3227 option --fair-sched= yes. Prior to this change, the pipe-based
3228 thread serialisation mechanism (which is still the default) could
3229 give very unfair scheduling. --fair-sched=yes improves
3230 responsiveness of interactive multithreaded applications, and
3231 improves repeatability of results from the thread checkers Helgrind
3234 * For tool developers: support to run Valgrind on Valgrind has been
3235 improved. We can now routinely Valgrind on Helgrind or Memcheck.
3237 * gdbserver now shows the float shadow registers as integer
3238 rather than float values, as the shadow values are mostly
3239 used as bit patterns.
3241 * Increased limit for the --num-callers command line flag to 500.
3243 * Performance improvements for error matching when there are many
3244 suppression records in use.
3246 * Improved support for DWARF4 debugging information (bug 284184).
3248 * Initial support for DWZ compressed Dwarf debug info.
3250 * Improved control over the IR optimiser's handling of the tradeoff
3251 between performance and precision of exceptions. Specifically,
3252 --vex-iropt-precise-memory-exns has been removed and replaced by
3253 --vex-iropt-register-updates, with extended functionality. This
3254 allows the Valgrind gdbserver to always show up to date register
3257 * Modest performance gains through the use of translation chaining for
3260 * ==================== FIXED BUGS ====================
3262 The following bugs have been fixed or resolved. Note that "n-i-bz"
3263 stands for "not in bugzilla" -- that is, a bug that was reported to us
3264 but never got a bugzilla entry. We encourage you to file bugs in
3265 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3266 than mailing the developers (or mailing lists) directly -- bugs that
3267 are not entered into bugzilla tend to get forgotten about or ignored.
3269 To see details of a given bug, visit
3270 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3271 where XXXXXX is the bug number as listed below.
3273 197914 Building valgrind from svn now requires automake-1.10
3274 203877 increase to 16Mb maximum allowed alignment for memalign et al
3275 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
3276 247386 make perf does not run all performance tests
3277 270006 Valgrind scheduler unfair
3278 270777 Adding MIPS/Linux port to Valgrind
3279 270796 s390x: Removed broken support for the TS insn
3280 271438 Fix configure for proper SSE4.2 detection
3281 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
3282 273475 Add support for AVX instructions
3283 274078 improved configure logic for mpicc
3284 276993 fix mremap 'no thrash checks'
3285 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
3286 281482 memcheck incorrect byte allocation count in realloc() for silly argument
3287 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
3288 283413 Fix wrong sanity check
3289 283671 Robustize alignment computation in LibVEX_Alloc
3290 283961 Adding support for some HCI IOCTLs
3291 284124 parse_type_DIE: confused by: DWARF 4
3292 284864 == 273475 (Add support for AVX instructions)
3293 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
3294 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
3295 285725 == 273475 (Add support for AVX instructions)
3296 286261 add wrapper for linux I2C_RDWR ioctl
3297 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
3298 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
3299 286384 configure fails "checking for a supported version of gcc"
3300 286497 == 273475 (Add support for AVX instructions)
3301 286596 == 273475 (Add support for AVX instructions)
3302 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
3303 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
3304 287260 Incorrect conditional jump or move depends on uninitialised value(s)
3305 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
3306 287307 == 273475 (Add support for AVX instructions)
3307 287858 VG_(strerror): unknown error
3308 288298 (MacOSX) unhandled syscall shm_unlink
3309 288995 == 273475 (Add support for AVX instructions)
3310 289470 Loading of large Mach-O thin binaries fails.
3311 289656 == 273475 (Add support for AVX instructions)
3312 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
3313 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
3314 289839 s390x: Provide support for unicode conversion instructions
3315 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
3316 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
3317 290655 Add support for AESKEYGENASSIST instruction
3318 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
3319 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
3320 291253 ES register not initialised in valgrind simulation
3321 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
3322 291865 s390x: Support the "Compare Double and Swap" family of instructions
3323 292300 == 273475 (Add support for AVX instructions)
3324 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
3325 292493 == 273475 (Add support for AVX instructions)
3326 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
3327 292627 Missing support for some SCSI ioctls
3328 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
3329 292841 == 273475 (Add support for AVX instructions)
3330 292993 implement the getcpu syscall on amd64-linux
3331 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
3332 293088 Add some VEX sanity checks for ppc64 unhandled instructions
3333 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
3334 293754 PCMPxSTRx not implemented for 16-bit characters
3335 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
3336 293808 CLFLUSH not supported by latest VEX for amd64
3337 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
3338 294048 MPSADBW instruction not implemented
3339 294055 regtest none/tests/shell fails when locale is not set to C
3340 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
3341 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
3342 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
3343 294260 disInstr_AMD64: disInstr miscalculated next %rip
3344 294523 --partial-loads-ok=yes causes false negatives
3345 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
3346 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
3347 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
3348 295089 can not annotate source for both helgrind and drd
3349 295221 POWER Processor decimal floating point instruction support missing
3350 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
3351 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
3352 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
3353 295617 ARM - Add some missing syscalls
3354 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
3355 296229 Linux user input device ioctls missing wrappers
3356 296318 ELF Debug info improvements (more than one rx/rw mapping)
3357 296422 Add translation chaining support
3358 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
3359 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
3360 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
3361 297078 gdbserver signal handling problems [..]
3362 297147 drd false positives on newly allocated memory
3363 297329 disallow decoding of IBM Power DFP insns on some machines
3364 297497 POWER Processor decimal floating point instruction support missing
3365 297701 Another alias for strncasecmp_l in libc-2.13.so
3366 297911 'invalid write' not reported when using APIs for custom mem allocators.
3367 297976 s390x: revisit EX implementation
3368 297991 Valgrind interferes with mmap()+ftell()
3369 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
3370 297993 Fix compilation of valgrind with gcc -g3.
3371 298080 POWER Processor DFP support missing, part 3
3372 298227 == 273475 (Add support for AVX instructions)
3373 298335 == 273475 (Add support for AVX instructions)
3374 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
3375 298394 s390x: Don't bail out on an unknown machine model. [..]
3376 298421 accept4() syscall (366) support is missing for ARM
3377 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
3378 298732 valgrind installation problem in ubuntu with kernel version 3.x
3379 298862 POWER Processor DFP instruction support missing, part 4
3380 298864 DWARF reader mis-parses DW_FORM_ref_addr
3381 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
3382 299053 Support DWARF4 DW_AT_high_pc constant form
3383 299104 == 273475 (Add support for AVX instructions)
3384 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
3385 299629 dup3() syscall (358) support is missing for ARM
3386 299694 POWER Processor DFP instruction support missing, part 5
3387 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
3388 299803 == 273475 (Add support for AVX instructions)
3389 299804 == 273475 (Add support for AVX instructions)
3390 299805 == 273475 (Add support for AVX instructions)
3391 300140 ARM - Missing (T1) SMMUL
3392 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
3393 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
3394 300414 FCOM and FCOMP unimplemented for amd64 guest
3395 301204 infinite loop in canonicaliseSymtab with ifunc symbol
3396 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
3397 301265 add x86 support to Android build
3398 301984 configure script doesn't detect certain versions of clang
3399 302205 Fix compiler warnings for POWER VEX code and POWER test cases
3400 302287 Unhandled movbe instruction on Atom processors
3401 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
3402 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
3403 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
3404 302656 == 273475 (Add support for AVX instructions)
3405 302709 valgrind for ARM needs extra tls support for android emulator [..]
3406 302827 add wrapper for CDROM_GET_CAPABILITY
3407 302901 Valgrind crashes with dwz optimized debuginfo
3408 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
3409 303116 Add support for the POWER instruction popcntb
3410 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
3411 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
3412 303466 == 273475 (Add support for AVX instructions)
3413 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
3414 303963 strstr() function produces wrong results under valgrind callgrind
3415 304054 CALL_FN_xx macros need to enforce stack alignment
3416 304561 tee system call not supported
3417 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
3418 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
3419 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
3420 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
3421 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
3422 n-i-bz s390x: Shadow registers can now be examined using vgdb
3424 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
3425 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
3429 Release 3.7.0 (5 November 2011)
3430 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3431 3.7.0 is a feature release with many significant improvements and the
3432 usual collection of bug fixes.
3434 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3435 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
3436 Support for recent distros and toolchain components (glibc 2.14, gcc
3437 4.6, MacOSX 10.7) has been added.
3439 * ================== PLATFORM CHANGES =================
3441 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
3442 analyse 64-bit programs running on z/Architecture. Most user space
3443 instructions up to and including z10 are supported. Valgrind has
3444 been tested extensively on z9, z10, and z196 machines running SLES
3445 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
3446 known to work well. Callgrind, Helgrind, and DRD work reasonably
3447 well on z9 and later models. See README.s390 for more details.
3449 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
3450 64-bit processes are supported. Some complex threaded applications
3451 (Firefox) are observed to hang when run as 32 bit applications,
3452 whereas 64-bit versions run OK. The cause is unknown. Memcheck
3453 will likely report some false errors. In general, expect some rough
3454 spots. This release also supports MacOSX 10.6, but drops support
3457 * Preliminary support for Android (on ARM). Valgrind can now run
3458 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
3459 README.android for more details, plus instructions on how to get
3462 * Support for the IBM Power ISA 2.06 (Power7 instructions)
3464 * General correctness and performance improvements for ARM/Linux, and,
3465 by extension, ARM/Android.
3467 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
3468 instruction set support is under development but is not available in
3471 * Support for AIX5 has been removed.
3473 * ==================== TOOL CHANGES ====================
3475 * Memcheck: some incremental changes:
3477 - reduction of memory use in some circumstances
3479 - improved handling of freed memory, which in some circumstances
3480 can cause detection of use-after-free that would previously have
3483 - fix of a longstanding bug that could cause false negatives (missed
3484 errors) in programs doing vector saturated narrowing instructions.
3486 * Helgrind: performance improvements and major memory use reductions,
3487 particularly for large, long running applications which perform many
3488 synchronisation (lock, unlock, etc) events. Plus many smaller
3491 - display of locksets for both threads involved in a race
3493 - general improvements in formatting/clarity of error messages
3495 - addition of facilities and documentation regarding annotation
3496 of thread safe reference counted C++ classes
3498 - new flag --check-stack-refs=no|yes [yes], to disable race checking
3499 on thread stacks (a performance hack)
3501 - new flag --free-is-write=no|yes [no], to enable detection of races
3502 where one thread accesses heap memory but another one frees it,
3503 without any coordinating synchronisation event
3505 * DRD: enabled XML output; added support for delayed thread deletion
3506 in order to detect races that occur close to the end of a thread
3507 (--join-list-vol); fixed a memory leak triggered by repeated client
3508 memory allocatation and deallocation; improved Darwin support.
3510 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
3512 * exp-sgcheck: this tool has been reduced in scope so as to improve
3513 performance and remove checking that Memcheck does better.
3514 Specifically, the ability to check for overruns for stack and global
3515 arrays is unchanged, but the ability to check for overruns of heap
3516 blocks has been removed. The tool has accordingly been renamed to
3517 exp-sgcheck ("Stack and Global Array Checking").
3519 * ==================== OTHER CHANGES ====================
3521 * GDB server: Valgrind now has an embedded GDB server. That means it
3522 is possible to control a Valgrind run from GDB, doing all the usual
3523 things that GDB can do (single stepping, breakpoints, examining
3524 data, etc). Tool-specific functionality is also available. For
3525 example, it is possible to query the definedness state of variables
3526 or memory from within GDB when running Memcheck; arbitrarily large
3527 memory watchpoints are supported, etc. To use the GDB server, start
3528 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3531 * Improved support for unfriendly self-modifying code: a new option
3532 --smc-check=all-non-file is available. This adds the relevant
3533 consistency checks only to code that originates in non-file-backed
3534 mappings. In effect this confines the consistency checking only to
3535 code that is or might be JIT generated, and avoids checks on code
3536 that must have been compiled ahead of time. This significantly
3537 improves performance on applications that generate code at run time.
3539 * It is now possible to build a working Valgrind using Clang-2.9 on
3542 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3543 These enable and disable error reporting on a per-thread, and
3544 nestable, basis. This is useful for hiding errors in particularly
3545 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3546 now uses this facility.
3548 * Added the --mod-funcname option to cg_diff.
3550 * ==================== FIXED BUGS ====================
3552 The following bugs have been fixed or resolved. Note that "n-i-bz"
3553 stands for "not in bugzilla" -- that is, a bug that was reported to us
3554 but never got a bugzilla entry. We encourage you to file bugs in
3555 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3556 mailing the developers (or mailing lists) directly -- bugs that are
3557 not entered into bugzilla tend to get forgotten about or ignored.
3559 To see details of a given bug, visit
3560 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3561 where XXXXXX is the bug number as listed below.
3563 79311 malloc silly arg warning does not give stack trace
3564 210935 port valgrind.h (not valgrind) to win32 to support client requests
3565 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3566 243404 Port to zSeries
3567 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3568 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3569 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3570 253206 Some fixes for the faultstatus testcase
3571 255223 capget testcase fails when running as root
3572 256703 xlc_dbl_u32.c testcase broken
3573 256726 Helgrind tests have broken inline asm
3574 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3575 264800 testcase compile failure on zseries
3576 265762 make public VEX headers compilable by G++ 3.x
3577 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3578 266753 configure script does not give the user the option to not use QtCore
3579 266931 gen_insn_test.pl is broken
3580 266961 ld-linux.so.2 i?86-linux strlen issues
3581 266990 setns instruction causes false positive
3582 267020 Make directory for temporary files configurable at run-time.
3583 267342 == 267997 (segmentation fault on Mac OS 10.6)
3584 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3585 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3586 267488 regtest: darwin support for 64-bit build
3587 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3588 267630 Add support for IBM Power ISA 2.06 -- stage 1
3589 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3590 267819 Add client request for informing the core about reallocation
3591 267925 laog data structure quadratic for a single sequence of lock
3592 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3593 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3594 268513 missed optimizations in fold_Expr
3595 268619 s390x: fpr - gpr transfer facility
3596 268620 s390x: reconsider "long displacement" requirement
3597 268621 s390x: improve IR generation for XC
3598 268715 s390x: FLOGR is not universally available
3599 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3600 268930 s390x: MHY is not universally available
3601 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3602 269079 Support ptrace system call on ARM
3603 269144 missing "Bad option" error message
3604 269209 conditional load and store facility (z196)
3605 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3606 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3607 269736 s390x: minor code generation tweaks
3608 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3609 269863 s390x: remove unused function parameters
3610 269864 s390x: tweak s390_emit_load_cc
3611 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3612 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3613 270115 s390x: rewrite some testcases
3614 270309 == 267997 (valgrind crash on startup)
3615 270320 add support for Linux FIOQSIZE ioctl() call
3616 270326 segfault while trying to sanitize the environment passed to execle
3617 270794 IBM POWER7 support patch causes regression in none/tests
3618 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3619 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3620 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3621 270959 s390x: invalid use of R0 as base register
3622 271042 VSX configure check fails when it should not
3623 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3624 271259 s390x: fix code confusion
3625 271337 == 267997 (Valgrind segfaults on MacOS X)
3626 271385 s390x: Implement Ist_MBE
3627 271501 s390x: misc cleanups
3628 271504 s390x: promote likely and unlikely
3629 271579 ppc: using wrong enum type
3630 271615 unhandled instruction "popcnt" (arch=amd10h)
3631 271730 Fix bug when checking ioctls: duplicate check
3632 271776 s390x: provide STFLE instruction support
3633 271779 s390x: provide clock instructions like STCK
3634 271799 Darwin: ioctls without an arg report a memory error
3635 271820 arm: fix type confusion
3636 271917 pthread_cond_timedwait failure leads to not-locked false positive
3637 272067 s390x: fix DISP20 macro
3638 272615 A typo in debug output in mc_leakcheck.c
3639 272661 callgrind_annotate chokes when run from paths containing regex chars
3640 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3641 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3642 272967 make documentation build-system more robust
3643 272986 Fix gcc-4.6 warnings with valgrind.h
3644 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3645 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3646 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3647 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3648 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3649 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3650 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3651 273778 exp-ptrcheck: unhandled sysno == 259
3652 274089 exp-ptrcheck: unhandled sysno == 208
3653 274378 s390x: Various dispatcher tweaks
3654 274447 WARNING: unhandled syscall: 340
3655 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3656 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3657 274926 valgrind does not build against linux-3
3658 275148 configure FAIL with glibc-2.14
3659 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3660 275168 Make Valgrind work for MacOSX 10.7 Lion
3661 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3662 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3663 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3664 275308 Fix implementation for ppc64 fres instruc
3665 275339 s390x: fix testcase compile warnings
3666 275517 s390x: Provide support for CKSM instruction
3667 275710 s390x: get rid of redundant address mode calculation
3668 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3669 275852 == 250101 (valgrind uses all swap space and is killed)
3670 276784 Add support for IBM Power ISA 2.06 -- stage 3
3671 276987 gdbsrv: fix tests following recent commits
3672 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3673 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3674 277471 Unhandled syscall: 340
3675 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3676 277653 ARM: support Thumb2 PLD instruction
3677 277663 ARM: NEON float VMUL by scalar incorrect
3678 277689 ARM: tests for VSTn with register post-index are broken
3679 277694 ARM: BLX LR instruction broken in ARM mode
3680 277780 ARM: VMOV.F32 (immediate) instruction is broken
3681 278057 fuse filesystem syscall deadlocks
3682 278078 Unimplemented syscall 280 on ppc32
3683 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3684 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3685 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3686 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3687 279027 Support for MVCL and CLCL instruction
3688 279027 s390x: Provide support for CLCL and MVCL instructions
3689 279062 Remove a redundant check in the insn selector for ppc.
3690 279071 JDK creates PTEST with redundant REX.W prefix
3691 279212 gdbsrv: add monitor cmd v.info scheduler.
3692 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3693 279698 memcheck discards valid-bits for packuswb
3694 279795 memcheck reports uninitialised values for mincore on amd64
3695 279994 Add support for IBM Power ISA 2.06 -- stage 3
3696 280083 mempolicy syscall check errors
3697 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3698 280710 s390x: config files for nightly builds
3699 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3700 280965 Valgrind breaks fcntl locks when program does mmap
3701 281138 WARNING: unhandled syscall: 340
3702 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3703 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3704 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3705 281468 s390x: handle do_clone and gcc clones in call traces
3706 281488 ARM: VFP register corruption
3707 281828 == 275284 (false memmove warning: "Source and destination overlap")
3708 281883 s390x: Fix system call wrapper for "clone".
3709 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3710 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3711 282238 SLES10: make check fails
3712 282979 strcasestr needs replacement with recent(>=2.12) glibc
3713 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3714 283243 Regression in ppc64 memcheck tests
3715 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3716 283427 re-connect epoll_pwait syscall on ARM linux
3717 283600 gdbsrv: android: port vgdb.c
3718 283709 none/tests/faultstatus needs to account for page size
3719 284305 filter_gdb needs enhancement to work on ppc64
3720 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3721 284472 Thumb2 ROR.W encoding T2 not implemented
3722 284621 XML-escape process command line in XML output
3723 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3724 that have non-power-of-2 sizes (also AMDs)
3725 n-i-bz don't be spooked by libraries mashed by elfhack
3726 n-i-bz don't be spooked by libxul.so linked with gold
3727 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3729 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3730 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3731 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3735 Release 3.6.1 (16 February 2011)
3736 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3737 3.6.1 is a bug fix release. It adds support for some SSE4
3738 instructions that were omitted in 3.6.0 due to lack of time. Initial
3739 support for glibc-2.13 has been added. A number of bugs causing
3740 crashing or assertion failures have been fixed.
3742 The following bugs have been fixed or resolved. Note that "n-i-bz"
3743 stands for "not in bugzilla" -- that is, a bug that was reported to us
3744 but never got a bugzilla entry. We encourage you to file bugs in
3745 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3746 mailing the developers (or mailing lists) directly -- bugs that are
3747 not entered into bugzilla tend to get forgotten about or ignored.
3749 To see details of a given bug, visit
3750 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3751 where XXXXXX is the bug number as listed below.
3753 188572 Valgrind on Mac should suppress setenv() mem leak
3754 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3755 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3756 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3757 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3758 254420 memory pool tracking broken
3759 254957 Test code failing to compile due to changes in memcheck.h
3760 255009 helgrind/drd: crash on chmod with invalid parameter
3761 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3762 255355 helgrind/drd: crash on threaded programs doing fork
3764 255418 (SSE4.x) rint call compiled with ICC
3765 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3766 255888 closing valgrindoutput tag outputted to log-stream on error
3767 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3768 255966 Slowness when using mempool annotations
3769 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3770 256600 super-optimized strcasecmp() false positive
3771 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3772 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3773 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3774 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3775 257276 Missing case in memcheck --track-origins=yes
3776 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3777 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3778 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3779 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3780 263099 callgrind_annotate counts Ir improperly [...]
3781 263877 undefined coprocessor instruction on ARMv7
3782 265964 configure FAIL with glibc-2.13
3783 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3784 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3785 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3786 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3787 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3789 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3793 Release 3.6.0 (21 October 2010)
3794 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3795 3.6.0 is a feature release with many significant improvements and the
3796 usual collection of bug fixes.
3798 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3799 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3800 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3802 -------------------------
3804 Here are some highlights. Details are shown further down:
3806 * Support for ARM/Linux.
3808 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3810 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3812 * Support for the SSE4.2 instruction set.
3814 * Enhancements to the Callgrind profiler, including the ability to
3815 handle CPUs with three levels of cache.
3817 * A new experimental heap profiler, DHAT.
3819 * A huge number of bug fixes and small enhancements.
3821 -------------------------
3823 Here are details of the above changes, together with descriptions of
3824 many other changes, and a list of fixed bugs.
3826 * ================== PLATFORM CHANGES =================
3828 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3829 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3830 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3832 This requires a CPU capable of running the ARMv7-A instruction set
3833 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3834 of the user space instruction set, including ARM and Thumb integer
3835 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3836 Cachegrind and Massif tools work properly; other tools work to
3839 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3840 with support for recent releases of the underlying toolchain
3841 components, notably gcc-4.5 and glibc-2.12.
3843 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3844 support also works much better on OS X 10.5, and is as solid as
3847 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3848 64-bit mode. In 32-bit mode, support is only available up to and
3849 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3850 supported in 64-bit mode, and 32-bit mode does in fact support the
3851 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3852 10.6 on 32-bit targets.
3854 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3855 and including version 2.05 is supported.
3857 * ==================== TOOL CHANGES ====================
3859 * Cachegrind has a new processing script, cg_diff, which finds the
3860 difference between two profiles. It's very useful for evaluating
3861 the performance effects of a change in a program.
3863 Related to this change, the meaning of cg_annotate's (rarely-used)
3864 --threshold option has changed; this is unlikely to affect many
3865 people, if you do use it please see the user manual for details.
3867 * Callgrind now can do branch prediction simulation, similar to
3868 Cachegrind. In addition, it optionally can count the number of
3869 executed global bus events. Both can be used for a better
3870 approximation of a "Cycle Estimation" as derived event (you need to
3871 update the event formula in KCachegrind yourself).
3873 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3874 rather than the L2 cache. This is to accommodate machines with
3875 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3876 cache configuration of such a machine it will run the simulation as
3877 if the L2 cache isn't present. This means the results are less
3878 likely to match the true result for the machine, but
3879 Cachegrind/Callgrind's results are already only approximate, and
3880 should not be considered authoritative. The results are still
3881 useful for giving a general idea about a program's locality.
3883 * Massif has a new option, --pages-as-heap, which is disabled by
3884 default. When enabled, instead of tracking allocations at the level
3885 of heap blocks (as allocated with malloc/new/new[]), it instead
3886 tracks memory allocations at the level of memory pages (as mapped by
3887 mmap, brk, etc). Each mapped page is treated as its own block.
3888 Interpreting the page-level output is harder than the heap-level
3889 output, but this option is useful if you want to account for every
3890 byte of memory used by a program.
3892 * DRD has two new command-line options: --free-is-write and
3893 --trace-alloc. The former allows to detect reading from already freed
3894 memory, and the latter allows tracing of all memory allocations and
3897 * DRD has several new annotations. Custom barrier implementations can
3898 now be annotated, as well as benign races on static variables.
3900 * DRD's happens before / happens after annotations have been made more
3901 powerful, so that they can now also be used to annotate e.g. a smart
3902 pointer implementation.
3904 * Helgrind's annotation set has also been drastically improved, so as
3905 to provide to users a general set of annotations to describe locks,
3906 semaphores, barriers and condition variables. Annotations to
3907 describe thread-safe reference counted heap objects have also been
3910 * Memcheck has a new command-line option, --show-possibly-lost, which
3911 is enabled by default. When disabled, the leak detector will not
3912 show possibly-lost blocks.
3914 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3915 has been added. DHAT keeps track of allocated heap blocks, and also
3916 inspects every memory reference to see which block (if any) is being
3917 accessed. This gives a lot of insight into block lifetimes,
3918 utilisation, turnover, liveness, and the location of hot and cold
3919 fields. You can use DHAT to do hot-field profiling.
3921 * ==================== OTHER CHANGES ====================
3923 * Improved support for unfriendly self-modifying code: the extra
3924 overhead incurred by --smc-check=all has been reduced by
3925 approximately a factor of 5 as compared with 3.5.0.
3927 * Ability to show directory names for source files in error messages.
3928 This is combined with a flexible mechanism for specifying which
3929 parts of the paths should be shown. This is enabled by the new flag
3932 * A new flag, --require-text-symbol, which will stop the run if a
3933 specified symbol is not found it a given shared object when it is
3934 loaded into the process. This makes advanced working with function
3935 intercepting and wrapping safer and more reliable.
3937 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3938 and control of Valgrind is now available for the tools Memcheck and
3939 Helgrind. XML output from Valgrind is available for Memcheck,
3940 Helgrind and exp-Ptrcheck.
3942 * More reliable stack unwinding on amd64-linux, particularly in the
3943 presence of function wrappers, and with gcc-4.5 compiled code.
3945 * Modest scalability (performance improvements) for massive
3946 long-running applications, particularly for those with huge amounts
3949 * Support for analyzing programs running under Wine with has been
3950 improved. The header files <valgrind/valgrind.h>,
3951 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3952 Windows-programs compiled with MinGW or one of the Microsoft Visual
3955 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3956 The 32-bit simulator was not affected. This did not occur often,
3957 but when it did would usually crash the program under test.
3960 * A large number of bugs were fixed. These are shown below.
3962 * A number of bugs were investigated, and were candidates for fixing,
3963 but are not fixed in 3.6.0, due to lack of developer time. They may
3964 get fixed in later releases. They are:
3966 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3967 212419 false positive "lock order violated" (A+B vs A)
3968 213685 Undefined value propagates past dependency breaking instruction
3969 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3970 237920 valgrind segfault on fork failure
3971 242137 support for code compiled by LLVM-2.8
3972 242423 Another unknown Intel cache config value
3973 243232 Inconsistent Lock Orderings report with trylock
3974 243483 ppc: callgrind triggers VEX assertion failure
3975 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3976 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3978 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3979 249435 Analyzing wine programs with callgrind triggers a crash
3980 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3981 250065 Handling large allocations
3982 250101 huge "free" memory usage due to m_mallocfree.c
3983 "superblocks fragmentation"
3984 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3985 252091 Callgrind on ARM does not detect function returns correctly
3986 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3987 254420 memory pool tracking broken
3988 n-i-bz support for adding symbols for JIT generated code
3991 The following bugs have been fixed or resolved. Note that "n-i-bz"
3992 stands for "not in bugzilla" -- that is, a bug that was reported to us
3993 but never got a bugzilla entry. We encourage you to file bugs in
3994 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3995 mailing the developers (or mailing lists) directly -- bugs that are
3996 not entered into bugzilla tend to get forgotten about or ignored.
3998 To see details of a given bug, visit
3999 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4000 where XXXXXX is the bug number as listed below.
4002 135264 dcbzl instruction missing
4004 153699 Valgrind should report unaligned reads with movdqa
4006 190429 Valgrind reports lost of errors in ld.so
4007 with x86_64 2.9.90 glibc
4008 197266 valgrind appears to choke on the xmms instruction
4010 197988 Crash when demangling very large symbol names
4011 202315 unhandled syscall: 332 (inotify_init1)
4012 203256 Add page-level profiling to Massif
4013 205093 dsymutil=yes needs quotes, locking (partial fix)
4014 205241 Snow Leopard 10.6 support (partial fix)
4015 206600 Leak checker fails to upgrade indirect blocks when their
4016 parent becomes reachable
4017 210935 port valgrind.h (not valgrind) to win32 so apps run under
4018 wine can make client requests
4019 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
4020 within Linux ip-stack checksum functions
4021 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
4023 213685 Undefined value propagates past dependency breaking instruction
4025 215914 Valgrind inserts bogus empty environment variable
4027 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
4028 222545 shmat fails under valgind on some arm targets
4029 222560 ARM NEON support
4032 232509 Docs build fails with formatting inside <title></title> elements
4034 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
4035 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
4036 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
4037 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
4038 237485 symlink (syscall 57) is not supported on Mac OS
4039 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
4041 238208 is_just_below_ESP doesn't take into account red-zone
4042 238345 valgrind passes wrong $0 when executing a shell script
4043 238679 mq_timedreceive syscall doesn't flag the reception buffer
4045 238696 fcntl command F_DUPFD_CLOEXEC not supported
4046 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4047 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4048 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
4050 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
4056 242606 unhandled syscall: setegid (in Ptrcheck)
4057 242814 Helgrind "Impossible has happened" during
4058 QApplication::initInstance();
4059 243064 Valgrind attempting to read debug information from iso
4060 243270 Make stack unwinding in Valgrind wrappers more reliable
4061 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
4062 sysno = 277 (mq_open)
4063 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
4064 244493 ARM VFP d16-d31 registers support
4065 244670 add support for audit_session_self syscall on Mac OS 10.6
4066 244921 The xml report of helgrind tool is not well format
4067 244923 In the xml report file, the <preamble> not escape the
4068 xml char, eg '<','&','>'
4069 245535 print full path names in plain text reports
4070 245925 x86-64 red zone handling problem
4071 246258 Valgrind not catching integer underruns + new [] s
4072 246311 reg/reg cmpxchg doesn't work on amd64
4073 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
4074 246888 Improve Makefile.vex.am
4075 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
4076 to [f]chmod_extended
4077 247526 IBM POWER6 (ISA 2.05) support is incomplete
4078 247561 Some leak testcases fails due to reachable addresses in
4080 247875 sizeofIRType to handle Ity_I128
4081 247894 [PATCH] unhandled syscall sys_readahead
4082 247980 Doesn't honor CFLAGS passed to configure
4083 248373 darwin10.supp is empty in the trunk
4084 248822 Linux FIBMAP ioctl has int parameter instead of long
4085 248893 [PATCH] make readdwarf.c big endianess safe to enable
4086 unwinding on big endian systems
4087 249224 Syscall 336 not supported (SYS_proc_info)
4089 249775 Incorrect scheme for detecting NEON capabilities of host CPU
4090 249943 jni JVM init fails when using valgrind
4091 249991 Valgrind incorrectly declares AESKEYGENASSIST support
4093 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
4094 250799 frexp$fenv_access_off function generates SIGILL
4095 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
4096 251251 support pclmulqdq insn
4097 251362 valgrind: ARM: attach to debugger either fails or provokes
4099 251674 Unhandled syscall 294
4102 254257 Add support for debugfiles found by build-id
4103 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
4104 254646 Wrapped functions cause stack misalignment on OS X
4105 (and possibly Linux)
4106 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
4108 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
4112 Release 3.5.0 (19 August 2009)
4113 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4114 3.5.0 is a feature release with many significant improvements and the
4115 usual collection of bug fixes. The main improvement is that Valgrind
4116 now works on Mac OS X.
4118 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
4119 and X86/Darwin. Support for recent distros and toolchain components
4120 (glibc 2.10, gcc 4.5) has been added.
4122 -------------------------
4124 Here is a short summary of the changes. Details are shown further
4127 * Support for Mac OS X (10.5.x).
4129 * Improvements and simplifications to Memcheck's leak checker.
4131 * Clarification and simplifications in various aspects of Valgrind's
4134 * XML output for Helgrind and Ptrcheck.
4136 * Performance and stability improvements for Helgrind and DRD.
4138 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
4140 * A new experimental tool, BBV, useful for computer architecture
4143 * Improved Wine support, including ability to read Windows PDB
4146 -------------------------
4148 Here are details of the above changes, followed by descriptions of
4149 many other minor changes, and a list of fixed bugs.
4152 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
4153 called "Darwin" because that is the name of the OS core, which is the
4154 level that Valgrind works at.)
4158 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
4159 because it would require work and 10.4 is only becoming less common.
4161 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
4162 fairly well. For 10.5.x, 32-bit programs are the default even on
4163 64-bit machines, so it handles most current programs.
4165 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
4166 officially supported, but simple programs at least will probably work.
4167 However, start-up is slow.
4169 - PowerPC machines are not supported.
4171 Things that don't work:
4173 - The Ptrcheck tool.
4175 - Objective-C garbage collection.
4179 - If you have Rogue Amoeba's "Instant Hijack" program installed,
4180 Valgrind will fail with a SIGTRAP at start-up. See
4181 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
4186 - You will likely find --dsymutil=yes a useful option, as error
4187 messages may be imprecise without it.
4189 - Mac OS X support is new and therefore will be less robust than the
4190 Linux support. Please report any bugs you find.
4192 - Threaded programs may run more slowly than on Linux.
4194 Many thanks to Greg Parker for developing this port over several years.
4197 * Memcheck's leak checker has been improved.
4199 - The results for --leak-check=summary now match the summary results
4200 for --leak-check=full. Previously they could differ because
4201 --leak-check=summary counted "indirectly lost" blocks and
4202 "suppressed" blocks as "definitely lost".
4204 - Blocks that are only reachable via at least one interior-pointer,
4205 but are directly pointed to by a start-pointer, were previously
4206 marked as "still reachable". They are now correctly marked as
4209 - The default value for the --leak-resolution option has been
4210 changed from "low" to "high". In general, this means that more
4211 leak reports will be produced, but each leak report will describe
4212 fewer leaked blocks.
4214 - With --leak-check=full, "definitely lost" and "possibly lost"
4215 leaks are now considered as proper errors, ie. they are counted
4216 for the "ERROR SUMMARY" and affect the behaviour of
4217 --error-exitcode. These leaks are not counted as errors if
4218 --leak-check=summary is specified, however.
4220 - Documentation for the leak checker has been improved.
4223 * Various aspects of Valgrind's text output have changed.
4225 - Valgrind's start-up message has changed. It is shorter but also
4226 includes the command being run, which makes it easier to use
4227 --trace-children=yes. An example:
4229 - Valgrind's shut-down messages have also changed. This is most
4230 noticeable with Memcheck, where the leak summary now occurs before
4231 the error summary. This change was necessary to allow leaks to be
4232 counted as proper errors (see the description of the leak checker
4233 changes above for more details). This was also necessary to fix a
4234 longstanding bug in which uses of suppressions against leaks were
4235 not "counted", leading to difficulties in maintaining suppression
4236 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
4238 - Behavior of -v has changed. In previous versions, -v printed out
4239 a mixture of marginally-user-useful information, and tool/core
4240 statistics. The statistics printing has now been moved to its own
4241 flag, --stats=yes. This means -v is less verbose and more likely
4242 to convey useful end-user information.
4244 - The format of some (non-XML) stack trace entries has changed a
4245 little. Previously there were six possible forms:
4247 0x80483BF: really (a.c:20)
4248 0x80483BF: really (in /foo/a.out)
4250 0x80483BF: (within /foo/a.out)
4251 0x80483BF: ??? (a.c:20)
4254 The third and fourth of these forms have been made more consistent
4255 with the others. The six possible forms are now:
4257 0x80483BF: really (a.c:20)
4258 0x80483BF: really (in /foo/a.out)
4259 0x80483BF: really (in ???)
4260 0x80483BF: ??? (in /foo/a.out)
4261 0x80483BF: ??? (a.c:20)
4264 Stack traces produced when --xml=yes is specified are different
4268 * Helgrind and Ptrcheck now support XML output, so they can be used
4269 from GUI tools. Also, the XML output mechanism has been
4272 - The XML format has been overhauled and generalised, so it is more
4273 suitable for error reporting tools in general. The Memcheck
4274 specific aspects of it have been removed. The new format, which
4275 is an evolution of the old format, is described in
4276 docs/internals/xml-output-protocol4.txt.
4278 - Memcheck has been updated to use the new format.
4280 - Helgrind and Ptrcheck are now able to emit output in this format.
4282 - The XML output mechanism has been overhauled. XML is now output
4283 to its own file descriptor, which means that:
4285 * Valgrind can output text and XML independently.
4287 * The longstanding problem of XML output being corrupted by
4288 unexpected un-tagged text messages is solved.
4290 As before, the destination for text output is specified using
4291 --log-file=, --log-fd= or --log-socket=.
4293 As before, XML output for a tool is enabled using --xml=yes.
4295 Because there's a new XML output channel, the XML output
4296 destination is now specified by --xml-file=, --xml-fd= or
4299 Initial feedback has shown this causes some confusion. To
4300 clarify, the two envisaged usage scenarios are:
4302 (1) Normal text output. In this case, do not specify --xml=yes
4303 nor any of --xml-file=, --xml-fd= or --xml-socket=.
4305 (2) XML output. In this case, specify --xml=yes, and one of
4306 --xml-file=, --xml-fd= or --xml-socket= to select the XML
4307 destination, one of --log-file=, --log-fd= or --log-socket=
4308 to select the destination for any remaining text messages,
4309 and, importantly, -q.
4311 -q makes Valgrind completely silent on the text channel,
4312 except in the case of critical failures, such as Valgrind
4313 itself segfaulting, or failing to read debugging information.
4314 Hence, in this scenario, it suffices to check whether or not
4315 any output appeared on the text channel. If yes, then it is
4316 likely to be a critical error which should be brought to the
4317 attention of the user. If no (the text channel produced no
4318 output) then it can be assumed that the run was successful.
4320 This allows GUIs to make the critical distinction they need to
4321 make (did the run fail or not?) without having to search or
4322 filter the text output channel in any way.
4324 It is also recommended to use --child-silent-after-fork=yes in
4328 * Improvements and changes in Helgrind:
4330 - XML output, as described above
4332 - Checks for consistent association between pthread condition
4333 variables and their associated mutexes are now performed.
4335 - pthread_spinlock functions are supported.
4337 - Modest performance improvements.
4339 - Initial (skeletal) support for describing the behaviour of
4340 non-POSIX synchronisation objects through ThreadSanitizer
4341 compatible ANNOTATE_* macros.
4343 - More controllable tradeoffs between performance and the level of
4344 detail of "previous" accesses in a race. There are now three
4347 * --history-level=full. This is the default, and was also the
4348 default in 3.4.x. It shows both stacks involved in a race, but
4349 requires a lot of memory and can be very slow in programs that
4350 do many inter-thread synchronisation events.
4352 * --history-level=none. This only shows the later stack involved
4353 in a race. This can be much faster than --history-level=full,
4354 but makes it much more difficult to find the other access
4355 involved in the race.
4357 The new intermediate setting is
4359 * --history-level=approx
4361 For the earlier (other) access, two stacks are presented. The
4362 earlier access is guaranteed to be somewhere in between the two
4363 program points denoted by those stacks. This is not as useful
4364 as showing the exact stack for the previous access (as per
4365 --history-level=full), but it is better than nothing, and it's
4366 almost as fast as --history-level=none.
4369 * New features and improvements in DRD:
4371 - The error messages printed by DRD are now easier to interpret.
4372 Instead of using two different numbers to identify each thread
4373 (Valgrind thread ID and DRD thread ID), DRD does now identify
4374 threads via a single number (the DRD thread ID). Furthermore
4375 "first observed at" information is now printed for all error
4376 messages related to synchronization objects.
4378 - Added support for named semaphores (sem_open() and sem_close()).
4380 - Race conditions between pthread_barrier_wait() and
4381 pthread_barrier_destroy() calls are now reported.
4383 - Added support for custom allocators through the macros
4384 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
4385 in <valgrind/valgrind.h>). An alternative for these two macros is
4386 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
4389 - Added support for annotating non-POSIX synchronization objects
4390 through several new ANNOTATE_*() macros.
4392 - OpenMP: added support for the OpenMP runtime (libgomp) included
4393 with gcc versions 4.3.0 and 4.4.0.
4397 - Added two new command-line options (--first-race-only and
4398 --segment-merging-interval).
4401 * Genuinely atomic support for x86/amd64/ppc atomic instructions
4403 Valgrind will now preserve (memory-access) atomicity of LOCK-
4404 prefixed x86/amd64 instructions, and any others implying a global
4405 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
4407 This means that Valgrinded processes will "play nicely" in
4408 situations where communication with other processes, or the kernel,
4409 is done through shared memory and coordinated with such atomic
4410 instructions. Prior to this change, such arrangements usually
4411 resulted in hangs, races or other synchronisation failures, because
4412 Valgrind did not honour atomicity of such instructions.
4415 * A new experimental tool, BBV, has been added. BBV generates basic
4416 block vectors for use with the SimPoint analysis tool, which allows
4417 a program's overall behaviour to be approximated by running only a
4418 fraction of it. This is useful for computer architecture
4419 researchers. You can run BBV by specifying --tool=exp-bbv (the
4420 "exp-" prefix is short for "experimental"). BBV was written by
4424 * Modestly improved support for running Windows applications under
4425 Wine. In particular, initial support for reading Windows .PDB debug
4426 information has been added.
4429 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
4430 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
4434 * The Valgrind client requests VALGRIND_PRINTF and
4435 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
4436 the string was always printed immediately on its own line. Now, the
4437 string will be added to a buffer but not printed until a newline is
4438 encountered, or other Valgrind output is printed (note that for
4439 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
4440 "other Valgrind output"). This allows you to use multiple
4441 VALGRIND_PRINTF calls to build up a single output line, and also to
4442 print multiple output lines with a single request (by embedding
4443 multiple newlines in the string).
4446 * The graphs drawn by Massif's ms_print program have changed slightly:
4448 - The half-height chars '.' and ',' are no longer drawn, because
4449 they are confusing. The --y option can be used if the default
4450 y-resolution is not high enough.
4452 - Horizontal lines are now drawn after the top of a snapshot if
4453 there is a gap until the next snapshot. This makes it clear that
4454 the memory usage has not dropped to zero between snapshots.
4457 * Something that happened in 3.4.0, but wasn't clearly announced: the
4458 option --read-var-info=yes can be used by some tools (Memcheck,
4459 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
4460 variable type and location information. This makes those tools
4461 start up more slowly and increases memory consumption, but
4462 descriptions of data addresses in error messages become more
4466 * exp-Omega, an experimental instantaneous leak-detecting tool, was
4467 disabled in 3.4.0 due to a lack of interest and maintenance,
4468 although the source code was still in the distribution. The source
4469 code has now been removed from the distribution. For anyone
4470 interested, the removal occurred in SVN revision r10247.
4473 * Some changes have been made to the build system.
4475 - VEX/ is now integrated properly into the build system. This means
4476 that dependency tracking within VEX/ now works properly, "make
4477 install" will work without requiring "make" before it, and
4478 parallel builds (ie. 'make -j') now work (previously a
4479 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
4480 was effectively ignored).
4482 - The --with-vex configure option has been removed. It was of
4483 little use and removing it simplified the build system.
4485 - The location of some install files has changed. This should not
4486 affect most users. Those who might be affected:
4488 * For people who use Valgrind with MPI programs, the installed
4489 libmpiwrap.so library has moved from
4490 $(INSTALL)/<platform>/libmpiwrap.so to
4491 $(INSTALL)/libmpiwrap-<platform>.so.
4493 * For people who distribute standalone Valgrind tools, the
4494 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
4495 have moved to $(INSTALL)/libcoregrind-<platform>.a.
4497 These changes simplify the build system.
4499 - Previously, all the distributed suppression (*.supp) files were
4500 installed. Now, only default.supp is installed. This should not
4501 affect users as the other installed suppression files were not
4502 read; the fact that they were installed was a mistake.
4505 * KNOWN LIMITATIONS:
4507 - Memcheck is unusable with the Intel compiler suite version 11.1,
4508 when it generates code for SSE2-and-above capable targets. This
4509 is because of icc's use of highly optimised inlined strlen
4510 implementations. It causes Memcheck to report huge numbers of
4511 false errors even in simple programs. Helgrind and DRD may also
4514 Versions 11.0 and earlier may be OK, but this has not been
4518 The following bugs have been fixed or resolved. Note that "n-i-bz"
4519 stands for "not in bugzilla" -- that is, a bug that was reported to us
4520 but never got a bugzilla entry. We encourage you to file bugs in
4521 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4522 mailing the developers (or mailing lists) directly -- bugs that are
4523 not entered into bugzilla tend to get forgotten about or ignored.
4525 To see details of a given bug, visit
4526 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4527 where XXXXXX is the bug number as listed below.
4529 84303 How about a LockCheck tool?
4530 91633 dereference of null ptr in vgPlain_st_basetype
4531 97452 Valgrind doesn't report any pthreads problems
4532 100628 leak-check gets assertion failure when using
4533 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4534 108528 NPTL pthread cleanup handlers not called
4535 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4536 110128 mallinfo is not implemented...
4537 110770 VEX: Generated files not always updated when making valgrind
4538 111102 Memcheck: problems with large (memory footprint) applications
4539 115673 Vex's decoder should never assert
4540 117564 False positive: Syscall param clone(child_tidptr) contains
4541 uninitialised byte(s)
4542 119404 executing ssh from inside valgrind fails
4543 133679 Callgrind does not write path names to sources with dwarf debug
4545 135847 configure.in problem with non gnu compilers (and possible fix)
4546 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4547 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4548 136230 memcheck reports "possibly lost", should be "still reachable"
4549 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4550 137904 Valgrind reports a memory leak when using POSIX threads,
4552 139076 valgrind VT_GETSTATE error
4553 142228 complaint of elf_dynamic_do_rela in trivial usage
4554 145347 spurious warning with USBDEVFS_REAPURB
4555 148441 (wine) can't find memory leak in Wine, win32 binary
4557 148742 Leak-check fails assert on exit
4558 149878 add (proper) check for calloc integer overflow
4559 150606 Call graph is broken when using callgrind control
4560 152393 leak errors produce an exit code of 0. I need some way to
4561 cause leak errors to result in a nonzero exit code.
4562 157154 documentation (leak-resolution doc speaks about num-callers
4563 def=4) + what is a loss record
4564 159501 incorrect handling of ALSA ioctls
4565 162020 Valgrinding an empty/zero-byte file crashes valgrind
4566 162482 ppc: Valgrind crashes while reading stabs information
4567 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4568 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4569 163560 VEX/test_main.c is missing from valgrind-3.3.1
4570 164353 malloc_usable_size() doesn't return a usable size
4571 165468 Inconsistent formatting in memcheck manual -- please fix
4572 169505 main.c:286 (endOfInstr):
4573 Assertion 'ii->cost_offset == *cost_offset' failed
4574 177206 Generate default.supp during compile instead of configure
4575 177209 Configure valt_load_address based on arch+os
4576 177305 eventfd / syscall 323 patch lost
4577 179731 Tests fail to build because of inlining of non-local asm labels
4578 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4579 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4580 181594 Bogus warning for empty text segment
4581 181707 dwarf doesn't require enumerations to have name
4582 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4583 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4584 Assertion '!already_present' failed.
4585 185359 exp-ptrcheck: unhandled syscall getresuid()
4586 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4587 185816 Valgrind is unable to handle debug info for files with split
4588 debug info that are prelinked afterwards
4589 185980 [darwin] unhandled syscall: sem_open
4590 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4591 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4592 186790 Suppression pattern used for leaks are not reported
4593 186796 Symbols with length>200 in suppression files are ignored
4594 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4595 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4596 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4597 188046 bashisms in the configure script
4598 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4599 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4600 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4601 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4602 assertion fail "!lock->heldBy"
4603 188427 Add support for epoll_create1 (with patch)
4604 188530 Support for SIOCGSTAMPNS
4605 188560 Include valgrind.spec in the tarball
4606 188572 Valgrind on Mac should suppress setenv() mem leak
4607 189054 Valgrind fails to build because of duplicate non-local asm labels
4608 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4609 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4610 189763 drd assertion failure: s_threadinfo[tid].is_recording
4611 190219 unhandled syscall: 328 (x86-linux)
4612 190391 dup of 181394; see above
4613 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4614 190820 No debug information on powerpc-linux
4615 191095 PATCH: Improve usbdevfs ioctl handling
4616 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4618 191189 --xml=yes should obey --gen-suppressions=all
4619 191192 syslog() needs a suppression on macosx
4620 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4621 191761 getrlimit on MacOSX
4622 191992 multiple --fn-skip only works sometimes; dependent on order
4623 192634 V. reports "aspacem sync_check_mapping_callback:
4624 segment mismatch" on Darwin
4625 192954 __extension__ missing on 2 client requests
4626 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4627 194474 "INSTALL" file has different build instructions than "README"
4628 194671 Unhandled syscall (sem_wait?) from mac valgrind
4629 195069 memcheck: reports leak (memory still reachable) for
4631 195169 drd: (vgDrd_barrier_post_wait):
4632 Assertion 'r->sg[p->post_iteration]' failed.
4633 195268 valgrind --log-file doesn't accept ~/...
4634 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4635 195860 WARNING: unhandled syscall: unix:223
4636 196528 need a error suppression for pthread_rwlock_init under os x?
4637 197227 Support aio_* syscalls on Darwin
4638 197456 valgrind should reject --suppressions=(directory)
4639 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4640 197591 unhandled syscall 27 (mincore)
4641 197793 Merge DCAS branch to the trunk == 85756, 142103
4642 197794 Avoid duplicate filenames in Vex
4643 197898 make check fails on current SVN
4644 197901 make check fails also under exp-ptrcheck in current SVN
4645 197929 Make --leak-resolution=high the default
4646 197930 Reduce spacing between leak reports
4647 197933 Print command line of client at start-up, and shorten preamble
4648 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4649 198395 add BBV to the distribution as an experimental tool
4650 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4651 198649 callgrind_annotate doesn't cumulate counters
4652 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4653 199977 Valgrind complains about an unrecognized instruction in the
4654 atomic_incs test program
4655 200029 valgrind isn't able to read Fedora 12 debuginfo
4656 200760 darwin unhandled syscall: unix:284
4657 200827 DRD doesn't work on Mac OS X
4658 200990 VG_(read_millisecond_timer)() does not work correctly
4659 201016 Valgrind does not support pthread_kill() on Mac OS
4660 201169 Document --read-var-info
4661 201323 Pre-3.5.0 performance sanity checking
4662 201384 Review user manual for the 3.5.0 release
4663 201585 mfpvr not implemented on ppc
4664 201708 tests failing because x86 direction flag is left set
4665 201757 Valgrind doesn't handle any recent sys_futex additions
4666 204377 64-bit valgrind can not start a shell script
4667 (with #!/path/to/shell) if the shell is a 32-bit executable
4668 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4669 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4670 about memory allocation events with memory access tracing enabled
4671 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4673 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4677 Release 3.4.1 (28 February 2009)
4678 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4679 3.4.1 is a bug-fix release that fixes some regressions and assertion
4680 failures in debug info reading in 3.4.0, most notably incorrect stack
4681 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4682 other debug info problems are also fixed. A number of bugs in the
4683 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4685 In view of the fact that 3.4.0 contains user-visible regressions
4686 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4687 encouraged to ship 3.4.1 in preference to 3.4.0.
4689 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4690 bugzilla" -- that is, a bug that was reported to us but never got a
4691 bugzilla entry. We encourage you to file bugs in bugzilla
4692 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4693 developers (or mailing lists) directly -- bugs that are not entered
4694 into bugzilla tend to get forgotten about or ignored.
4696 n-i-bz Fix various bugs reading icc-11 generated debug info
4697 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4698 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4699 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4700 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4701 179618 exp-ptrcheck crashed / exit prematurely
4702 179624 helgrind: false positive races with pthread_create and
4703 recv/open/close/read
4704 134207 pkg-config output contains @VG_PLATFORM@
4705 176926 floating point exception at valgrind startup with PPC 440EPX
4706 181594 Bogus warning for empty text segment
4707 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4708 181707 Dwarf3 doesn't require enumerations to have name
4709 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4710 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4711 Assertion '!already_present' failed.
4712 185359 exp-ptrcheck unhandled syscall getresuid()
4714 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4715 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4719 Release 3.4.0 (2 January 2009)
4720 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4721 3.4.0 is a feature release with many significant improvements and the
4722 usual collection of bug fixes. This release supports X86/Linux,
4723 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4724 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4726 3.4.0 brings some significant tool improvements. Memcheck can now
4727 report the origin of uninitialised values, the thread checkers
4728 Helgrind and DRD are much improved, and we have a new experimental
4729 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4730 global arrays. In detail:
4732 * Memcheck is now able to track the origin of uninitialised values.
4733 When it reports an uninitialised value error, it will try to show
4734 the origin of the value, as either a heap or stack allocation.
4735 Origin tracking is expensive and so is not enabled by default. To
4736 use it, specify --track-origins=yes. Memcheck's speed will be
4737 essentially halved, and memory usage will be significantly
4738 increased. Nevertheless it can drastically reduce the effort
4739 required to identify the root cause of uninitialised value errors,
4740 and so is often a programmer productivity win, despite running more
4743 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4744 3.4.0, will be released shortly.
4746 * Helgrind's race detection algorithm has been completely redesigned
4747 and reimplemented, to address usability and scalability concerns:
4749 - The new algorithm has a lower false-error rate: it is much less
4750 likely to report races that do not really exist.
4752 - Helgrind will display full call stacks for both accesses involved
4753 in a race. This makes it easier to identify the root causes of
4756 - Limitations on the size of program that can run have been removed.
4758 - Performance has been modestly improved, although that is very
4761 - Direct support for Qt4 threading has been added.
4763 - pthread_barriers are now directly supported.
4765 - Helgrind works well on all supported Linux targets.
4767 * The DRD thread debugging tool has seen major improvements:
4769 - Greatly improved performance and significantly reduced memory
4772 - Support for several major threading libraries (Boost.Thread, Qt4,
4773 glib, OpenMP) has been added.
4775 - Support for atomic instructions, POSIX semaphores, barriers and
4776 reader-writer locks has been added.
4778 - Works now on PowerPC CPUs too.
4780 - Added support for printing thread stack usage at thread exit time.
4782 - Added support for debugging lock contention.
4784 - Added a manual for Drd.
4786 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4787 checks for misuses of pointers. In that sense it is a bit like
4788 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4789 detect overruns of stack and global arrays, it can detect
4790 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4791 detect accesses heap blocks that have been freed a very long time
4792 ago (millions of blocks in the past).
4794 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4795 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4796 of the main Valgrind documentation. As this is an experimental
4797 tool, we would be particularly interested in hearing about your
4798 experiences with it.
4800 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4801 longer built by default, although the code remains in the repository
4802 and the tarball. This is due to three factors: a perceived lack of
4803 users, a lack of maintenance, and concerns that it may not be
4804 possible to achieve reliable operation using the existing design.
4806 * As usual, support for the latest Linux distros and toolchain
4807 components has been added. It should work well on Fedora Core 10,
4808 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4809 state) is supported, as is glibc-2.9. The C++ demangler has been
4810 updated so as to work well with C++ compiled by even the most recent
4813 * You can now use frame-level wildcards in suppressions. This was a
4814 frequently-requested enhancement. A line "..." in a suppression now
4815 matches zero or more frames. This makes it easier to write
4816 suppressions which are precise yet insensitive to changes in
4819 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4821 * Very basic support for IBM Power6 has been added (64-bit processes only).
4823 * Valgrind is now cross-compilable. For example, it is possible to
4824 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4825 on a ppc32/64-linux target.
4827 * You can set the main thread's stack size at startup using the
4828 new --main-stacksize= flag (subject of course to ulimit settings).
4829 This is useful for running apps that need a lot of stack space.
4831 * The limitation that you can't use --trace-children=yes together
4832 with --db-attach=yes has been removed.
4834 * The following bugs have been fixed. Note that "n-i-bz" stands for
4835 "not in bugzilla" -- that is, a bug that was reported to us but
4836 never got a bugzilla entry. We encourage you to file bugs in
4837 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4838 mailing the developers (or mailing lists) directly.
4840 n-i-bz Make return types for some client requests 64-bit clean
4841 n-i-bz glibc 2.9 support
4842 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4843 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4844 n-i-bz Building in an env without gdb gives bogus gdb attach
4845 92456 Tracing the origin of uninitialised memory
4846 106497 Valgrind does not demangle some C++ template symbols
4848 151612 Suppression with "..." (frame-level wildcards in .supp files)
4849 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4850 159285 unhandled syscall:25 (stime, on x86-linux)
4851 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4852 160954 ppc build of valgrind crashes with illegal instruction (isel)
4853 160956 mallinfo implementation, w/ patch
4854 162092 Valgrind fails to start gnome-system-monitor
4855 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4856 163794 assertion failure with "--track-origins=yes"
4857 163933 sigcontext.err and .trapno must be set together
4858 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4859 164476 Missing kernel module loading system calls
4860 164669 SVN regression: mmap() drops posix file locks
4861 166581 Callgrind output corruption when program forks
4862 167288 Patch file for missing system calls on Cell BE
4863 168943 unsupported scas instruction pentium
4864 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4865 172417 x86->IR: 0x82 ...
4866 172563 amd64->IR: 0xD9 0xF5 - fprem1
4867 173099 .lds linker script generation error
4868 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4869 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4871 174908 --log-file value not expanded correctly for core file
4872 175044 Add lookup_dcookie for amd64
4873 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4875 Developer-visible changes:
4877 * Valgrind's debug-info reading machinery has been majorly overhauled.
4878 It can now correctly establish the addresses for ELF data symbols,
4879 which is something that has never worked properly before now.
4881 Also, Valgrind can now read DWARF3 type and location information for
4882 stack and global variables. This makes it possible to use the
4883 framework to build tools that rely on knowing the type and locations
4884 of stack and global variables, for example exp-Ptrcheck.
4886 Reading of such information is disabled by default, because most
4887 tools don't need it, and because it is expensive in space and time.
4888 However, you can force Valgrind to read it, using the
4889 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4890 make use of such information, if present, to provide source-level
4891 descriptions of data addresses in the error messages they create.
4893 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4894 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).