1 Release 3.22.0 (31 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 390871 ELF debug info reader confused with multiple .rodata* sections
59 417993 vbit-test fail on s390x with Iop_Add32: spurious dependency on uninit
60 426751 Valgrind reports "still reachable" memory using musl
61 (alpine running inside docker)
62 432801 Valgrind 3.16.1 reports a jump based on uninitialized memory somehow
63 related to clang and signals
64 433857 Add validation to C++17 aligned new/delete alignment size
65 433859 Add mismatched detection to C++ 17 aligned new/delete
66 460192 Add epoll_pwait2
67 461074 DWARF2 CFI reader: unhandled DW_OP_ 0x11 (consts) DW_OP_ 0x92 (bregx)
68 465782 s390x: Valgrind doesn't compile with Clang on s390x
69 466105 aligned_alloc problems, part 2
70 467441 Add mismatched detection to C++ 14 sized delete
71 469049 link failure on ppc64 (big endian) valgrind 3.20
72 469146 massif --ignore-fn does not ignore inlined functions
73 469768 Make it possible to install gdb scripts in a different location
74 470121 Can't run callgrind_control with valgrind 3.21.0 because of perl errors
75 470132 s390x: Assertion failure on VGM instruction
76 470520 Multiple realloc zero errors crash in MC_(eq_Error)
77 470713 Failure on the Yosys project: valgrind: m_libcfile.c:1802
78 (Bool vgPlain_realpath(const HChar *, HChar *)):
79 Assertion 'resolved' failed
80 470830 Don't print actions vgdb me ... continue for vgdb --multi mode
81 470978 s390x: Valgrind cannot start qemu-kvm when "sysctl vm.allocate_pgste=0"
82 471311 gdb --multi mode stdout redirecting to stderr
83 471807 Add support for lazy reading and downloading of DWARF debuginfo
84 472219 Syscall param ppoll(ufds.events) points to uninitialised byte(s)
85 472875 none/tests/s390x/dfp-1 failure
86 472963 Broken regular expression in configure.ac
87 473604 Fix bug472219.c compile failure with Clang 16
88 473677 make check compile failure with Clang 16 based on GCC 13.x
89 473745 must-be-redirected function - strlen
90 473870 FreeBSD 14 applications fail early at startup
91 473944 Handle mold linker split RW PT_LOAD segments correctly
92 474332 aligned_alloc under Valgrind returns nullptr when alignment is not a multiple of sizeof(void *)
93 475650 DRD does not work with C11 threads
94 475652 Missing suppression for __wcsncpy_avx2 (strncpy-avx2.S:308)?
95 476108 vg_replace_malloc DELETE checks size
96 n-i-bz Allow arguments with spaces in .valgrindrc files
97 n-i-bz FreeBSD fixed reading of Valgrind tools own debuginfo
99 To see details of a given bug, visit
100 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
101 where XXXXXX is the bug number as listed above.
103 (3.22.0.RC1: 17 Oct 2023)
104 (3.22.0.RC2: 26 Oct 2023)
106 Release 3.21.0 (28 Apr 2023)
107 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
109 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
110 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
111 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
112 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
113 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
114 AMD64/macOS 10.13 and nanoMIPS/Linux.
116 * ==================== CORE CHANGES ===================
118 * When GDB is used to debug a program running under valgrind using
119 the valgrind gdbserver, GDB will automatically load some
120 python code provided in valgrind defining GDB front end commands
121 corresponding to the valgrind monitor commands.
122 These GDB front end commands accept the same format as
123 the monitor commands directly sent to the Valgrind gdbserver.
124 These GDB front end commands provide a better integration
125 in the GDB command line interface, so as to use for example
126 GDB auto-completion, command specific help, searching for
127 a command or command help matching a regexp, ...
128 For relevant monitor commands, GDB will evaluate arguments
129 to make the use of monitor commands easier.
130 For example, instead of having to print the address of a variable
131 to pass it to a subsequent monitor command, the GDB front end
132 command will evaluate the address argument. It is for example
134 (gdb) memcheck who_points_at &some_struct sizeof(some_struct)
137 $2 = (some_struct_type *) 0x1130a0 <some_struct>
138 (gdb) p sizeof(some_struct)
140 (gdb) monitor who_point_at 0x1130a0 40
142 * The vgdb utility now supports extended-remote protocol when
143 invoked with --multi. In this mode the GDB run command is
144 supported. Which means you don't need to run gdb and valgrind
145 from different terminals. So for example to start your program
146 in gdb and run it under valgrind you can do:
148 (gdb) set remote exec-file prog
150 (gdb) target extended-remote | vgdb --multi
153 * The behaviour of realloc with a size of zero can now
154 be changed for tools that intercept malloc. Those
155 tools are memcheck, helgrind, drd, massif and dhat.
156 Realloc implementations generally do one of two things
157 - free the memory like free() and return NULL
158 (GNU libc and ptmalloc).
159 - either free the memory and then allocate a
160 minimum sized block or just return the
161 original pointer. Return NULL if the
162 allocation of the minimum sized block fails
163 (jemalloc, musl, snmalloc, Solaris, macOS).
164 When Valgrind is configured and built it will
165 try to match the OS and libc behaviour. However
166 if you are using a non-default library to replace
167 malloc and family (e.g., musl on a glibc Linux or
168 tcmalloc on FreeBSD) then you can use a command line
169 option to change the behaviour of Valgrind:
170 --realloc-zero-bytes-frees=yes|no [yes on Linux glibc, no otherwise]
172 * ================== PLATFORM CHANGES =================
174 * Make the address space limit on FreeBSD amd64 128Gbytes
175 (the same as Linux and Solaris, it was 32Gbytes)
177 * ==================== TOOL CHANGES ===================
180 - When doing a delta leak_search, it is now possible to only
181 output the new loss records compared to the previous leak search.
182 This is available in the memcheck monitor command 'leak_search'
183 by specifying the "new" keyword or in your program by using
184 the client request VALGRIND_DO_NEW_LEAK_CHECK.
185 Whenever a "delta" leak search is done (i.e. when specifying
186 "new" or "increased" or "changed" in the monitor command),
187 the new loss records have a "new" marker.
188 - Valgrind now contains python code that defines GDB memcheck
189 front end monitor commands. See CORE CHANGES.
190 - Performs checks for the use of realloc with a size of zero.
191 This is non-portable and a source of errors. If memcheck
192 detects such a usage it will generate an error
193 realloc() with size 0
194 followed by the usual callstacks.
195 A switch has been added to allow this to be turned off:
196 --show-realloc-size-zero=yes|no [yes]
199 - The option ---history-backtrace-size=<number> allows to configure
200 the number of entries to record in the stack traces of "old"
201 accesses. Previously, this number was hardcoded to 8.
202 - Valgrind now contains python code that defines GDB helgrind
203 front end monitor commands. See CORE CHANGES.
206 - `--cache-sim=no` is now the default. The cache simulation is old and
207 unlikely to match any real modern machine. This means only the `Ir`
208 event are gathered by default, but that is by far the most useful
210 - `cg_annotate`, `cg_diff`, and `cg_merge` have been rewritten in
211 Python. As a result, they all have more flexible command line
212 argument handling, e.g. supporting `--show-percs` and
213 `--no-show-percs` forms as well as the existing `--show-percs=yes`
214 and `--show-percs=no`.
215 - `cg_annotate` has some functional changes.
216 - It's much faster, e.g. 3-4x on common cases.
217 - It now supports diffing (with `--diff`, `--mod-filename`, and
218 `--mod-funcname`) and merging (by passing multiple data files).
219 - It now provides more information at the file and function level.
220 There are now "File:function" and "Function:file" sections. These
221 are very useful for programs that use inlining a lot.
222 - Support for user-annotated files and the `-I`/`--include` option
223 has been removed, because it was of little use and blocked other
225 - The `--auto` option is renamed `--annotate`, though the old
226 `--auto=yes`/`--auto=no` forms are still supported.
227 - `cg_diff` and `cg_merge` are now deprecated, because `cg_annotate`
228 now does a better job of diffing and merging.
229 - The Cachegrind output file format has changed very slightly, but in
230 ways nobody is likely to notice.
233 - Valgrind now contains python code that defines GDB callgrind
234 front end monitor commands. See CORE CHANGES.
237 - Valgrind now contains python code that defines GDB massif
238 front end monitor commands. See CORE CHANGES.
241 - A new kind of user request has been added which allows you to
242 override the 1024 byte limit on access count histograms for blocks
243 of memory. The client request is DHAT_HISTOGRAM_MEMORY.
245 * ==================== FIXED BUGS ====================
247 The following bugs have been fixed or resolved. Note that "n-i-bz"
248 stands for "not in bugzilla" -- that is, a bug that was reported to us
249 but never got a bugzilla entry. We encourage you to file bugs in
250 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
251 than mailing the developers (or mailing lists) directly -- bugs that
252 are not entered into bugzilla tend to get forgotten about or ignored.
254 170510 Don't warn about ioctl of size 0 without direction hint
255 241072 List tools in --help output
256 327548 false positive while destroying mutex
257 382034 Testcases build fixes for musl
258 351857 confusing error message about valid command line option
259 374596 inconsistent RDTSCP support on x86_64
260 392331 Spurious lock not held error from inside pthread_cond_timedwait
261 397083 Likely false positive "uninitialised value(s)" for __wmemchr_avx2 and __wmemcmp_avx2_movbe
262 400793 pthread_rwlock_timedwrlock false positive
263 419054 Unhandled syscall getcpu on arm32
264 433873 openat2 syscall unimplemented on Linux
265 434057 Add stdio mode to valgrind's gdbserver
266 435441 valgrind fails to interpose malloc on musl 1.2.2 due to weak symbol name and no libc soname
267 436413 Warn about realloc of size zero
268 439685 compiler warning in callgrind/main.c
269 444110 priv/guest_ppc_toIR.c:36198:31: warning: duplicated 'if' condition.
270 444487 hginfo test detects an extra lock inside data symbol "_rtld_local"
271 444488 Use glibc.pthread.stack_cache_size tunable
272 444568 drd/tests/pth_barrier_thr_cr fails on Fedora 38
273 445743 "The impossible happened: mutex is locked simultaneously by two threads"
274 while using mutexes with priority inheritance and signals
275 449309 Missing loopback device ioctl(s)
276 459476 vgdb: allow address reuse to avoid "address already in use" errorsuse" errors
277 460356 s390: Sqrt32Fx4 -- cannot reduce tree
278 462830 WARNING: unhandled amd64-freebsd syscall: 474
279 463027 broken check for MPX instruction support in assembler
280 464103 Enhancement: add a client request to DHAT to mark memory to be histogrammed
281 464476 Firefox fails to start under Valgrind
282 464609 Valgrind memcheck should support Linux pidfd_open
283 464680 Show issues caused by memory policies like selinux deny_execmem
284 464859 Build failures with GCC-13 (drd tsan_unittest)
285 464969 D language demangling
286 465435 m_libcfile.c:66 (vgPlain_safe_fd): Assertion 'newfd >= VG_(fd_hard_limit)' failed.
287 466104 aligned_alloc problems, part 1
288 467036 Add time cost statistics for Regtest
289 467482 Build failure on aarch64 Alpine
290 467714 fdleak_* and rlimit tests fail when parent process has more than
291 64 descriptors opened
292 467839 Gdbserver: Improve compatibility of library directory name
293 468401 [PATCH] Add a style file for clang-format
294 468556 Build failure for vgdb
295 468606 build: remove "Valgrind relies on GCC" check/output
296 469097 ppc64(be) doesn't support SCV syscall instruction
297 n-i-bz FreeBSD rfork syscall fail with EINVAL or ENOSYS rather than VG_(unimplemented)
299 To see details of a given bug, visit
300 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
301 where XXXXXX is the bug number as listed above.
303 * ==================== KNOWN ISSUES ===================
305 * configure --enable-lto=yes is know to not work in all setups.
306 See bug 469049. Workaround: Build without LTO.
308 (3.21.0.RC1: 14 Apr 2023)
309 (3.21.0.RC2: 21 Apr 2023)
311 Release 3.20.0 (24 Oct 2022)
312 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
314 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
315 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
316 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
317 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
318 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
319 AMD64/macOS 10.13 and nanoMIPS/Linux.
321 * ==================== CORE CHANGES ===================
323 * The option "--vgdb-stop-at=event1,event2,..." accepts the new value abexit.
324 This indicates to invoke gdbserver when your program exits abnormally
325 (i.e. with a non zero exit code).
326 * Fix Rust v0 name demangling.
327 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
328 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
329 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with
331 * The option --enable-debuginfod=<no|yes> [default: yes] has been added on
333 * More DWARF5 support as generated by clang14.
335 * ==================== FIXED BUGS ====================
337 The following bugs have been fixed or resolved. Note that "n-i-bz"
338 stands for "not in bugzilla" -- that is, a bug that was reported to us
339 but never got a bugzilla entry. We encourage you to file bugs in
340 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
341 than mailing the developers (or mailing lists) directly -- bugs that
342 are not entered into bugzilla tend to get forgotten about or ignored.
344 131186 writev reports error in (vector[...])
345 434764 iconv_open causes ld.so v2.28+ to use optimised strncmp
346 446754 Improve error codes from alloc functions under memcheck
347 452274 memcheck crashes with Assertion 'sci->status.what == SsIdle' failed
348 452779 Valgrind fails to build on FreeBSD 13.0 with llvm-devel (15.0.0)
349 453055 shared_timed_mutex drd test fails with "Lock shared failed" message
350 453602 Missing command line option to enable/disable debuginfod
351 452802 Handle lld 9+ split RW PT_LOAD segments correctly
352 454040 s390x: False-positive memcheck:cond in memmem on arch13 systems
353 456171 [PATCH] FreeBSD: Don't record address errors when accessing the 'kern.ps_strings' sysctl struct
354 n-i-bz Implement vgdb invoker on FreeBSD
355 458845 PowerPC: The L field for the dcbf and sync instruction should be
357 458915 Remove register cache to fix 458915 gdbserver causes wrong syscall return
358 459031 Documentation on --error-exitcode incomplete
359 459477 XERROR messages lacks ending '\n' in vgdb
360 462007 Implicit int in none/tests/faultstatus.c
362 To see details of a given bug, visit
363 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
364 where XXXXXX is the bug number as listed above.
366 (3.20.0.RC1: 20 Oct 2022)
369 Release 3.19.0 (11 Apr 2022)
370 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
372 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
373 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
374 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
375 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
376 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
377 AMD64/macOS 10.13 and nanoMIPS/Linux.
379 * ==================== CORE CHANGES ===================
381 * Fix Rust v0 name demangling.
382 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
383 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
384 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with HardenedBSD
386 * ================== PLATFORM CHANGES =================
389 - ignore the "v8.x" architecture levels, only look at actual CPU features
390 present. Fixes mismatch detected between RDMA and atomics features
391 preventing startup on some QEMU configurations.
392 - Implement LD{,A}XP and ST{,L}XP
393 - Fix incorrect code emitted for doubleword CAS.
396 - Fix sys_ipc semtimedop syscall
397 - Fix VFLRX and WFLRX instructions
398 - Fix EXRL instruction with negative offset
401 - Reimplement the vbpermq instruction support to generate less Iops and
402 avoid overflowing internal buffers.
403 - Fix checking for scv support to avoid "Facility 'SCV' unavailable (12),
404 exception" messages in dmsg.
405 - Fix setting condition code for Vector Compare quad word instructions.
406 - Fix fix lxsibzx, lxsihzx and lxsihzx instructions so they only load
407 their respective sized data.
408 - Fix the prefixed stq instruction in PC relative mode.
410 * ==================== TOOL CHANGES ===================
413 - Speed up --track-origins=yes for large (in the range of hundreds to
414 thousands of megabytes) mmap/munmaps.
416 - Several fixes for new versions of libstd++ using new posix try_lock
419 * ==================== FIXED BUGS ====================
421 The following bugs have been fixed or resolved. Note that "n-i-bz"
422 stands for "not in bugzilla" -- that is, a bug that was reported to us
423 but never got a bugzilla entry. We encourage you to file bugs in
424 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
425 than mailing the developers (or mailing lists) directly -- bugs that
426 are not entered into bugzilla tend to get forgotten about or ignored.
428 403802 leak_cpp_interior fails with some reachable blocks different than expected
429 435732 memcheck/tests/leak_cpp_interior fails with gcc11
430 444242 s390x: Valgrind crashes on EXRL with negative offset
431 444399 arm64: unhandled instruction 0xC87F2D89 (LD{,A}XP and ST{,L}XP).
433 444481 gdb_server test failures on s390x
434 444495 dhat/tests/copy fails on s390x
435 444552 memcheck/tests/sem fails on s390x with glibc 2.34
436 444571 PPC, fix the lxsibzx and lxsihzx so they only load their respective
438 444836 PPC, pstq instruction for R=1 is not storing to the correct address.
439 444925 fexecve syscall wrapper not properly implemented
440 445032 valgrind/memcheck crash with SIGSEGV when SIGVTALRM timer used and
442 445211 Fix out of tree builds
443 445300 [PATCH] Fix building tests with Musl
444 445011 SIGCHLD is sent when valgrind uses debuginfod-find
445 445354 arm64 backend: incorrect code emitted for doubleword CAS
446 445415 arm64 front end: alignment checks missing for atomic instructions
447 445504 Using C++ condition_variable results in bogus "mutex is locked simultaneously by two threads" warning
448 445607 Unhandled amd64-freebsd syscall: 247
449 445668 Inline stack frame generation is broken for Rust binaries
450 445916 Demangle Rust v0 symbols with .llvm suffix
451 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false positives
452 446138 DRD/Helgrind with std::timed_mutex::try_lock_until false positives
453 446281 Add a DRD suppression for fwrite
454 446103 Memcheck: `--track-origins=yes` causes extreme slowdowns for large mmap/munmap
455 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false
456 446251 TARGET_SIGNAL_THR added to enum target_signal
457 446823 FreeBSD - missing syscalls when using libzm4
458 447991 s390x: Valgrind indicates illegal instruction on wflrx
459 447995 Valgrind segfault on power10 due to hwcap checking code
460 449483 Powerpc: vcmpgtsq., vcmpgtuq,, vcmpequq. instructions not setting the
461 condition code correctly.
462 449672 ppc64 --track-origins=yes failures because of bad cmov addHRegUse
463 449838 sigsegv liburing the 'impossible' happened for io_uring_setup
464 450025 Powerc: ACC file not implemented as a logical overlay of the VSR
466 450437 Warn for execve syscall with argv or argv[0] being NULL
467 450536 Powerpc: valgrind throws 'facility scv unavailable exception'
468 451626 Syscall param bpf(attr->raw_tracepoint.name) points to unaddressable byte(s)
469 451827 [ppc64le] VEX temporary storage exhausted with several vbpermq instructions
470 451843 valgrind fails to start on a FreeBSD system which enforces W^X
472 To see details of a given bug, visit
473 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
474 where XXXXXX is the bug number as listed above.
476 (3.19.0.RC1: 02 Apr 2022)
477 (3.19.0.RC2: 08 Apr 2022)
480 Release 3.18.0 (15 Oct 2021)
481 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
483 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
484 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
485 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
486 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
487 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
488 AMD64/macOS 10.13 and nanoMIPS/Linux.
490 * ==================== CORE CHANGES ===================
492 * The libiberty demangler has been updated, which brings support for
493 Rust v0 name demangling. [Update: alas, due to a bug, this support
494 isn't working in 3.18.0.]
496 * __libc_freeres isn't called anymore after the program recieves a
497 fatal signal. Causing some internal glibc resources to hang around,
498 but preventing any crashes after the program has ended.
500 * The DWARF reader is now very much faster at startup when just
501 --read-inline-info=yes (the default in most cases) is given.
503 * glibc 2.34, which moved various functions from libpthread.so into
504 libc.so, is now supported.
506 * ================== PLATFORM CHANGES =================
510 - v8.2 scalar and vector FABD, FACGE, FACGT and FADD.
511 - v8.2 FP compare & conditional compare instructions.
512 - Zero variants of v8.2 FP compare instructions.
516 - Support the miscellaneous-instruction-extensions facility 3 and
517 the vector-enhancements facility 2. This enables programs
518 compiled with "-march=arch13" or "-march=z15" to be executed
523 - ISA 3.1 support is now complete
524 - ISA 3.0 support for the darn instruction added.
525 - ISA 3.0 support for the vector system call instruction scv added.
526 - ISA 3.0 support for the copy, paste and cpabort instructions added.
528 * Support for X86/FreeBSD and AMD64/FreeBSD has been added.
530 * ==================== OTHER CHANGES ====================
532 * Memcheck on amd64: minor fixes to remove some false positive
535 * ==================== FIXED BUGS ====================
537 The following bugs have been fixed or resolved. Note that "n-i-bz"
538 stands for "not in bugzilla" -- that is, a bug that was reported to us
539 but never got a bugzilla entry. We encourage you to file bugs in
540 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
541 than mailing the developers (or mailing lists) directly -- bugs that
542 are not entered into bugzilla tend to get forgotten about or ignored.
544 208531 [PATCH]: FreeBSD support for valgrind
545 368960 WARNING: unhandled amd64-linux syscall: 163 (acct)
546 407589 [Linux] Add support for C11 aligned_alloc() and GNU reallocarray()
547 423963 Error in child thread when CLONE_PIDFD is used
548 426148 crash with "impossible happened" when running BPF CO-RE programs
549 429375 PPC ISA 3.1 support is missing, part 9
550 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
551 431306 Update demangler to support Rust v0 name mangling
552 432387 s390x: z15 instructions support
553 433437 FreeBSD support, part 1
554 433438 FreeBSD support, part 2
555 433439 FreeBSD support, part 3
556 433469 FreeBSD support, part 4
557 433473 FreeBSD support, part 5
558 433477 FreeBSD support, part 6
559 433479 FreeBSD support, part 7
560 433504 FreeBSD support, part 8
561 433506 FreeBSD support, part 9
562 433507 FreeBSD support, part 10
563 433508 FreeBSD support, part 11
564 433510 FreeBSD support, part 12
565 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
566 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
567 434296 s390x: False-positive memcheck diagnostics from vector string
569 434840 PPC64 darn instruction not supported
570 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
571 435908 valgrind tries to fetch from deubginfod for files which already
572 have debug information
573 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2
574 439046 valgrind is unusably large when linked with lld
575 439090 Implement close_range(2)
576 439326 Valgrind 3.17.0 won't compile with Intel 2021 oneAPI compilers
577 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
578 440670 unhandled ppc64le-linux syscall: 252 statfs64 and 253 fstatfs64
579 440906 Fix impossible constraint issue in P10 testcase.
580 441512 Remove a unneeded / unnecessary prefix check.
581 441534 Update the expected output for test_isa_3_1_VRT.
582 442061 very slow execution under Fedora 34 (readdwarf3)
583 443031 Gcc -many change requires explicit .machine directives
584 443033 Add support for the ISA 3.0 mcrxrx instruction
585 443034 Sraw, srawi, srad, sradi, mfs
586 443178 Powerpc, test jm-mfspr expected output needs to be updated.
587 443179 Need new test for the lxvx and stxvx instructions on ISA 2.07 and
589 443180 The subnormal test and the ISA 3.0 test generate compiler warnings
590 443314 In the latest GIT version, Valgrind with "--trace-flags" crashes
592 443605 Don't call final_tidyup (__libc_freeres) on FatalSignal
594 To see details of a given bug, visit
595 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
596 where XXXXXX is the bug number as listed below.
598 (3.18.0.RC1: 12 Oct 2021)
599 (3.18.0: 15 Oct 2021)
603 Release 3.17.0 (19 Mar 2021)
604 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
606 3.17.0 fixes a number of bugs and adds some functional changes: support for
607 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
608 some new instructions for Arm64, S390 and POWER. There are also some tool
611 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
612 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
613 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
614 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
615 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
617 * ==================== CORE CHANGES ===================
619 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
622 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
623 debugging information. When a debuginfo file cannot be found locally,
624 Valgrind is able to query debuginfod servers for the file using its
625 build-id. See the user manual for more information about debuginfod support.
627 * ================== PLATFORM CHANGES =================
631 - Inaccuracies resulting from double-rounding in the simulation of
632 floating-point multiply-add/subtract instructions have been fixed. These
633 should now behave exactly as the hardware does.
635 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
636 ongoing. Support for the half-word variants of at least the following
637 instructions has been added:
639 FABS <Vd>.<T>, <Vn>.<T>
641 FNEG <Vd>.<T>, <Vn>.<T>
643 FSQRT <Vd>.<T>, <Vn>.<T>
648 - Implement the new instructions/features that were added to z/Architecture
649 with the vector-enhancements facility 1. Also cover the instructions from
650 the vector-packed-decimal facility that are defined outside the chapter
651 "Vector Decimal Instructions", but not the ones from that chapter itself.
653 For a detailed list of newly supported instructions see the updates to
654 `docs/internals/s390-opcodes.csv'.
656 Since the miscellaneous instruction extensions facility 2 was already
657 added in Valgrind 3.16.0, this completes the support necessary to run
658 general programs built with `--march=z14' under Valgrind. The
659 vector-packed-decimal facility is currently not exploited by the standard
660 toolchain and libraries.
664 - Various bug fixes. Fix for the sync field to limit setting just two of
665 the two bits in the L-field. Fix the write size for the stxsibx and
666 stxsihx instructions. Fix the modsw and modsd instructions.
668 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
669 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
670 conversion instructions are still missing.
672 * ==================== TOOL CHANGES ====================
674 * General tool changes
676 - All the tools and their vgpreload libraries are now installed under
677 libexec because they cannot be executed directly and should be run through
678 the valgrind executable. This should be an internal, not user visible,
679 change, but might impact valgrind packagers.
681 - The --track-fds option now respects -q, --quiet and won't output anything
682 if no file descriptors are leaked. It also won't report the standard stdin
683 (0), stdout (1) or stderr (2) descriptors as being leaked with
684 --trace-fds=yes anymore. To track whether the standard file descriptors
685 are still open at the end of the program run use --trace-fds=all.
689 - DHAT has been extended, with two new modes of operation. The new
690 --mode=copy flag triggers copy profiling, which records calls to memcpy,
691 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
692 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
693 the new dhat/dhat.h file. This is useful for learning more about hot code
694 paths. See the user manual for more information about the new modes.
696 - Because of these changes, DHAT's file format has changed. DHAT output
697 files produced with earlier versions of DHAT will not work with this
698 version of DHAT's viewer, and DHAT output files produced with this version
699 of DHAT will not work with earlier versions of DHAT's viewer.
701 * ==================== FIXED BUGS ====================
703 The following bugs have been fixed or resolved. Note that "n-i-bz"
704 stands for "not in bugzilla" -- that is, a bug that was reported to us
705 but never got a bugzilla entry. We encourage you to file bugs in
706 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
707 than mailing the developers (or mailing lists) directly -- bugs that
708 are not entered into bugzilla tend to get forgotten about or ignored.
710 To see details of a given bug, visit
711 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
712 where XXXXXX is the bug number as listed below.
714 140178 open("/proc/self/exe", ...); doesn't quite work
715 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
716 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
717 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
718 345077 linux syscall execveat support (linux 3.19)
719 361770 Missing F_ADD_SEALS
720 369029 handle linux syscalls sched_getattr and sched_setattr
721 384729 __libc_freeres inhibits cross-platform valgrind
722 388787 Support for C++17 new/delete
723 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
724 396656 Warnings while reading debug info
725 397605 ioctl FICLONE mishandled
726 401416 Compile failure with openmpi 4.0
727 408663 Suppression file for musl libc
728 404076 s390x: z14 vector instructions not implemented
729 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
730 (actually affected all x86 and nanomips regardless of host bitness)
731 413547 regression test does not check for Arm 64 features.
732 414268 Enable AArch64 feature detection and decoding for v8.x instructions
733 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
734 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
735 422261 platform selection fails for unqualified client name
736 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
737 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
738 423195 PPC ISA 3.1 support is missing, part 1
739 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
740 424012 crash with readv/writev having invalid but not NULL arg2 iovec
741 424298 amd64: Implement RDSEED
742 425232 PPC ISA 3.1 support is missing, part 2
743 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
744 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
745 426123 PPC ISA 3.1 support is missing, part 3
746 426144 Fix "condition variable has not been initialized" on Fedora 33.
747 427400 PPC ISA 3.1 support is missing, part 4
748 427401 PPC ISA 3.1 support is missing, part 5
749 427404 PPC ISA 3.1 support is missing, part 6
750 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
751 427787 Support new faccessat2 linux syscall (439)
752 427969 debuginfo section duplicates a section in the main ELF file
753 428035 drd: Unbreak the musl build
754 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
755 428716 cppcheck detects potential leak in VEX/useful/smchash.c
756 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
757 429352 PPC ISA 3.1 support is missing, part 7
758 429354 PPC ISA 3.1 support is missing, part 8
759 429692 unhandled ppc64le-linux syscall: 147 (getsid)
760 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
762 429952 Errors when building regtest with clang
763 430354 ppc stxsibx and stxsihx instructions write too much data
764 430429 valgrind.h doesn't compile on s390x with clang
765 430485 expr_is_guardable doesn't handle Iex_Qop
766 431556 Complete arm64 FADDP v8.2 instruction support
767 432102 Add support for DWARF5 as produced by GCC11
768 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
769 432381 drd: Process STACK_REGISTER client requests
770 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
771 432672 vg_regtest: test-specific environment variables not reset between tests
772 432809 VEX should support REX.W + POPF
773 432861 PPC modsw and modsd give incorrect results for 1 mod 12
774 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
775 432215 Add debuginfod functionality
776 433323 Use pkglibexecdir as vglibdir
777 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
778 433629 valgrind/README has type "abd" instead of "and"
779 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
780 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
781 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
782 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
783 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
785 (3.17.0.RC1: 13 Mar 2021)
786 (3.17.0.RC2: 17 Mar 2021)
787 (3.17.0: 19 Mar 2021)
791 Release 3.16.1 (22 June 2020)
792 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
794 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
795 fixes character encoding problems in the documentation HTML.
797 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
798 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
800 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
804 Release 3.16.0 (27 May 2020)
805 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
807 3.16.0 is a feature release with many improvements and the usual collection of
810 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
811 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
812 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
813 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
814 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
816 * ==================== CORE CHANGES ===================
818 * It is now possible to dynamically change the value of many command line
819 options while your program (or its children) are running under Valgrind.
821 To see the list of dynamically changeable options, run
822 "valgrind --help-dyn-options".
824 You can change the options from the shell by using vgdb to launch
825 the monitor command "v.clo <clo option>...".
826 The same monitor command can be used from a gdb connected
827 to the valgrind gdbserver.
828 Your program can also change the dynamically changeable options using
829 the client request VALGRIND_CLO_CHANGE(option).
831 * ================== PLATFORM CHANGES =================
833 * MIPS: preliminary support for nanoMIPS instruction set has been added.
835 * ==================== TOOL CHANGES ====================
839 - The implicit memcpy done by each call to realloc now counts towards the
840 read and write counts of resized heap blocks, making those counts higher
845 - cg_annotate's --auto and --show-percs options now default to 'yes', because
846 they are usually wanted.
850 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
851 because they are usually wanted.
853 - The command option --collect-systime has been enhanced to specify
854 the unit used to record the elapsed time spent during system calls.
855 The command option now accepts the values no|yes|msec|usec|nsec,
856 where yes is a synonym of msec. When giving the value nsec, the
857 system cpu time of system calls is also recorded.
861 - Several memcheck options are now dynamically changeable.
862 Use valgrind --help-dyn-options to list them.
864 - The release 3.15 introduced a backward incompatible change for
865 some suppression entries related to preadv and pwritev syscalls.
866 When reading a suppression entry using the unsupported 3.14 format,
867 valgrind will now produce a warning to say the suppression entry will not
868 work, and suggest the needed change.
870 - Significantly fewer false positive errors on optimised code generated by
871 Clang and GCC. In particular, Memcheck now deals better with the
872 situation where the compiler will transform C-level "A && B" into "B && A"
873 under certain circumstances (in which the transformation is valid).
874 Handling of integer equality/non-equality checks on partially defined
875 values is also improved on some architectures.
879 - The exprimental Stack and Global Array Checking tool has been removed.
880 It only ever worked on x86 and amd64, and even on those it had a
881 high false positive rate and was slow. An alternative for detecting
882 stack and global array overruns is using the AddressSanitizer (ASAN)
883 facility of the GCC and Clang compilers, which require you to rebuild
884 your code with -fsanitize=address.
886 * ==================== OTHER CHANGES ====================
888 * New and modified GDB server monitor features:
890 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
892 - The gdbserver monitor commands that require an address and an optional
893 length argument now accepts the alternate 'C like' syntax "address[length]".
894 For example, the memcheck command "monitor who_points_at 0x12345678 120"
895 can now also be given as "monitor who_points_at 0x12345678[120]".
897 * ==================== FIXED BUGS ====================
899 The following bugs have been fixed or resolved. Note that "n-i-bz"
900 stands for "not in bugzilla" -- that is, a bug that was reported to us
901 but never got a bugzilla entry. We encourage you to file bugs in
902 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
903 than mailing the developers (or mailing lists) directly -- bugs that
904 are not entered into bugzilla tend to get forgotten about or ignored.
906 To see details of a given bug, visit
907 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
908 where XXXXXX is the bug number as listed below.
910 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
911 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
912 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
913 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
914 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
915 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
916 400872 Add nanoMIPS support to Valgrind
917 403212 drd/tests/trylock hangs on FreeBSD
918 404406 s390x: z14 miscellaneous instructions not implemented
919 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
920 406561 mcinfcallWSRU gdbserver_test fails on ppc64
921 406824 Unsupported baseline
922 407218 Add support for the copy_file_range syscall
923 407307 Intercept stpcpy also in ld.so for arm64
924 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
926 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
927 408009 Expose rdrand and f16c even on avx if host cpu supports them
928 408091 Missing pkey syscalls
929 408414 Add support for missing for preadv2 and pwritev2 syscalls
930 409141 Valgrind hangs when SIGKILLed
931 409206 Support for Linux PPS and PTP ioctls
932 409367 exit_group() after signal to thread waiting in futex() causes hangs
933 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
934 409780 References to non-existent configure.in
935 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
936 410599 Non-deterministic behaviour of pth_self_kill_15_other test
937 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
938 411134 Allow the user to change a set of command line options during execution
939 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
940 412344 Problem setting mips flags with specific paths
941 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
942 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
943 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
944 413603 callgrind_annotate/cg_annotate truncate function names at '#'
945 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
946 415136 ARMv8.1 Compare-and-Swap instructions are not supported
947 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
948 416239 valgrind crashes when handling clock_adjtime
949 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
950 416286 DRD reports "conflicting load" error on std::mutex::lock()
951 416301 s390x: "compare and signal" not supported
952 416387 finit_module and bpf syscalls are unhandled on arm64
953 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
954 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
955 416753 new 32bit time syscalls for 2038+
956 417075 pwritev(vector[...]) suppression ignored
957 417075 is not fixed, but incompatible supp entries are detected
958 and a warning is produced for these.
959 417187 [MIPS] Conditional branch problem since 'grail' changes
960 417238 Test memcheck/tests/vbit-test fails on mips64 BE
961 417266 Make memcheck/tests/linux/sigqueue usable with musl
962 417281 s390x: /bin/true segfaults with "grail" enabled
963 417427 commit to fix vki_siginfo_t definition created numerous regression
965 417452 s390_insn_store_emit: dst->tag for HRcVec128
966 417578 Add suppressions for glibc DTV leaks
967 417906 clone with CLONE_VFORK and no CLONE_VM fails
968 418004 Grail code additions break ppc64.
969 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
970 418997 s390x: Support Iex_ITE for float and vector types
971 419503 s390x: Avoid modifying registers returned from isel functions
972 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
973 421570 std_mutex fails on Arm v8.1 h/w
974 434035 vgdb might crash if valgrind is killed
975 n-i-bz Fix minor one time leaks in dhat.
976 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
977 n-i-bz Add support for the Linux io_uring system calls
978 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
979 n-i-bz Fix non-glibc build of test suite with s390x_features
980 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
981 423195 PPC ISA 3.1 support is missing, part 1
983 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
984 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
985 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
989 Release 3.15.0 (12 April 2019)
990 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
992 3.15.0 is a feature release with many improvements and the usual collection of
995 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
996 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
997 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
998 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
999 support for X86/macOS 10.13 and AMD64/macOS 10.13.
1001 * ==================== CORE CHANGES ===================
1003 * The XTree Massif output format now makes use of the information obtained
1004 when specifying --read-inline-info=yes.
1006 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
1008 * ==================== TOOL CHANGES ====================
1012 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
1013 it has been promoted from an experimental tool to a regular tool. Run it
1014 with --tool=dhat instead of --tool=exp-dhat.
1016 - DHAT now prints only minimal data when the program ends, instead writing
1017 the bulk of the profiling data to a file. As a result, the --show-top-n
1018 and --sort-by options have been removed.
1020 - Profile results can be viewed with the new viewer, dh_view.html. When
1021 a run ends, a short message is printed, explaining how to view the result.
1023 - See the documentation for more details.
1027 - cg_annotate has a new option, --show-percs, which prints percentages next
1028 to all event counts.
1032 - callgrind_annotate has a new option, --show-percs, which prints percentages
1033 next to all event counts.
1035 - callgrind_annotate now inserts commas in call counts, and
1036 sort the caller/callee lists in the call tree.
1040 - The default value for --read-inline-info is now "yes" on
1041 Linux/Android/Solaris. It is still "no" on other OS.
1045 - The option --xtree-leak=yes (to output leak result in xtree format)
1046 automatically activates the option --show-leak-kinds=all, as xtree
1047 visualisation tools such as kcachegrind can in any case select what kind
1048 of leak to visualise.
1050 - There has been further work to avoid false positives. In particular,
1051 integer equality on partially defined inputs (C == and !=) is now handled
1054 * ==================== OTHER CHANGES ====================
1056 * The new option --show-error-list=no|yes displays, at the end of the run, the
1057 list of detected errors and the used suppressions. Prior to this change,
1058 showing this information could only be done by specifying "-v -v", but that
1059 also produced a lot of other possibly-non-useful messages. The option -s is
1060 equivalent to --show-error-list=yes.
1062 * ==================== FIXED BUGS ====================
1064 The following bugs have been fixed or resolved. Note that "n-i-bz"
1065 stands for "not in bugzilla" -- that is, a bug that was reported to us
1066 but never got a bugzilla entry. We encourage you to file bugs in
1067 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1068 than mailing the developers (or mailing lists) directly -- bugs that
1069 are not entered into bugzilla tend to get forgotten about or ignored.
1071 To see details of a given bug, visit
1072 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1073 where XXXXXX is the bug number as listed below.
1075 385411 s390x: z13 vector floating-point instructions not implemented
1076 397187 z13 vector register support for vgdb gdbserver
1077 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
1078 398870 Please add support for instruction vcvtps2ph
1079 399287 amd64 front end: Illegal Instruction vcmptrueps
1080 399301 Use inlined frames in Massif XTree output.
1081 399322 Improve callgrind_annotate output
1082 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
1083 400164 helgrind test encounters mips x-compiler warnings and assembler error
1084 400490 s390x: VRs allocated as if separate from FPRs
1085 400491 s390x: Operand of LOCH treated as unsigned integer
1086 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
1087 options, which specify a mips64 processor
1088 401112 LLVM 5.0 generates comparison against partially initialized data
1089 401277 More bugs in z13 support
1090 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
1091 401578 drd: crashes sometimes on fork()
1092 401627 memcheck errors with glibc avx2 optimized wcsncmp
1093 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
1094 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
1095 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
1097 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
1098 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
1099 402123 invalid assembler opcodes for mips32r2
1100 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
1101 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
1102 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
1103 402351 mips64 libvexmultiarch_test fails on s390x
1104 402369 Overhaul DHAT
1105 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
1106 402480 Do not use %rsp in clobber list
1107 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
1108 402515 Implement new option --show-error-list=no|yes / -s
1109 402519 POWER 3.0 addex instruction incorrectly implemented
1110 402781 Redo the cache used to process indirect branch targets
1111 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
1112 403552 s390x: wrong facility bit checked for vector facility
1113 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
1114 404638 Add VG_(replaceIndexXA)
1115 404843 s390x: backtrace sometimes ends prematurely
1116 404888 autotools cleanup series
1117 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
1118 405182 Valgrind fails to build with Clang
1119 405205 filter_libc: remove the line holding the futex syscall error entirely
1120 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
1121 the upper and lower 32-bits of the 64-bit result
1122 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
1123 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
1124 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
1126 405403 s390x disassembler cannot be used on x86
1127 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
1128 405458 MIPS mkFormVEC arguments swapped?
1129 405716 drd: Fix an integer overflow in the stack margin calculation
1130 405722 Support arm64 core dump
1131 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
1132 of the 64-bit destination field.
1133 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
1134 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
1135 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
1137 406256 PPC64, vector floating point instructions don't handle subnormal
1138 according to VSCR[NJ] bit setting.
1139 406352 cachegrind/callgrind fails ann tests because of missing a.c
1140 406354 dhat is broken on x86 (32bit)
1141 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
1142 406357 gdbserver_tests fails because of gdb output change
1143 406360 memcheck/tests/libstdc++.supp needs more supression variants
1144 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
1145 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
1146 407340 PPC64, does not support the vlogefp, vexptefp instructions.
1147 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
1148 n-i-bz Fix callgrind_annotate non deterministic order for equal total
1149 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
1150 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
1151 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
1153 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
1154 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
1155 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
1159 Release 3.14.0 (9 October 2018)
1160 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1162 3.14.0 is a feature release with many improvements and the usual collection of
1165 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1166 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1167 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1168 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
1169 support for X86/macOS 10.13, AMD64/macOS 10.13.
1171 * ==================== CORE CHANGES ===================
1173 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
1174 debug info for unloaded code. This allows saved stack traces (e.g. for
1175 memory leaks) to include file/line info for code that has been dlclose'd (or
1176 similar). See the user manual for more information and known limitations.
1178 * Ability to specify suppressions based on source file name and line number.
1180 * Majorly overhauled register allocator. No end-user changes, but the JIT
1181 generates code a bit more quickly now.
1183 * ================== PLATFORM CHANGES =================
1185 * Preliminary support for macOS 10.13 has been added.
1187 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
1189 * mips: support for MIPS SIMD architecture (MSA) has been added.
1191 * mips: support for MIPS N32 ABI has been added.
1193 * s390: partial support for vector instructions (integer and string) has been
1196 * ==================== TOOL CHANGES ====================
1198 * Helgrind: Addition of a flag
1199 --delta-stacktrace=no|yes [yes on linux amd64/x86]
1200 which specifies how full history stack traces should be computed.
1201 Setting this to =yes can speed up Helgrind by 25% when using
1202 --history-level=full.
1204 * Memcheck: reduced false positive rate for optimised code created by Clang 6
1205 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
1206 blocks more carefully to determine where it can avoid expensive definedness
1207 checks without loss of precision. This is controlled by the flag
1208 --expensive-definedness-checks=no|auto|yes [auto].
1210 * ==================== OTHER CHANGES ====================
1212 * Valgrind is now buildable with link-time optimisation (LTO). A new
1213 configure option --enable-lto=yes allows building Valgrind with LTO. If the
1214 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
1215 Note that if you are doing Valgrind development, --enable-lto=yes massively
1216 slows down the build process.
1218 * ==================== FIXED BUGS ====================
1220 The following bugs have been fixed or resolved. Note that "n-i-bz"
1221 stands for "not in bugzilla" -- that is, a bug that was reported to us
1222 but never got a bugzilla entry. We encourage you to file bugs in
1223 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1224 than mailing the developers (or mailing lists) directly -- bugs that
1225 are not entered into bugzilla tend to get forgotten about or ignored.
1227 To see details of a given bug, visit
1228 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1229 where XXXXXX is the bug number as listed below.
1231 79362 Debug info is lost for .so files when they are dlclose'd
1232 208052 strlcpy error when n = 0
1233 255603 exp-sgcheck Assertion '!already_present' failed
1234 338252 building valgrind with -flto (link time optimisation) fails
1235 345763 MIPS N32 ABI support
1236 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
1237 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
1238 372347 Replacement problem of the additional c++14/c++17 new/delete operators
1239 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1240 376257 helgrind history full speed up using a cached stack
1241 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
1243 379748 Fix missing pselect syscall (OS X 10.11)
1244 379754 Fix missing syscall ulock_wait (OS X 10.12)
1245 380397 s390x: __GI_strcspn() replacemenet needed
1246 381162 possible array overrun in VEX register allocator
1247 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
1248 381274 powerpc too chatty even with --sigill-diagnostics=no
1249 381289 epoll_pwait can have a NULL sigmask
1250 381553 VEX register allocator v3
1251 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
1252 381769 Use ucontext_t instead of struct ucontext
1253 381805 arm32 needs ld.so index hardwire for new glibc security fixes
1254 382256 gz compiler flag test doesn't work for gold
1255 382407 vg_perf needs "--terse" command line option
1256 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
1257 382563 MIPS MSA ASE support
1258 382998 xml-socket doesn't work
1259 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
1260 383723 Fix missing kevent_qos syscall (macOS 10.11)
1261 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
1262 384096 Mention AddrCheck at Memcheck's command line option [..]
1263 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
1264 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
1265 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
1266 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
1267 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
1268 384337 performance improvements to VEX register allocator v2 and v3
1269 384526 reduce number of spill insns generated by VEX register allocator v3
1270 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
1271 384631 Sanitise client args as printed with -v
1272 384633 Add a simple progress-reporting facility
1273 384987 VEX regalloc: allocate caller-save registers for short lived vregs
1274 385055 PPC VEX temporary storage exhausted
1275 385182 PPC64 is missing support for the DSCR
1276 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
1277 385207 PPC64, generate_store_FPRF() generates too many Iops
1278 385208 PPC64, xxperm instruction exhausts temporary memory
1279 385210 PPC64, vpermr instruction could exhaust temporary memory
1280 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
1281 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
1282 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
1283 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
1284 385408 s390x: z13 vector "support" instructions not implemented
1285 385409 s390x: z13 vector integer instructions not implemented
1286 385410 s390x: z13 vector string instructions not implemented
1287 385412 s390x: new non-vector z13 instructions not implemented
1288 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
1289 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
1290 385939 Optionally exit on the first error
1291 386318 valgrind.org/info/tools.html is missing SGCheck
1292 386425 running valgrind + wine on armv7l gives illegal opcode
1293 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
1294 387410 MIPSr6 support
1295 387664 Memcheck: make expensive-definedness-checks be the default
1296 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
1297 387766 asm shifts cause false positive "Conditional jump or move depends
1298 on uninitialised value"
1299 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
1300 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
1301 388786 Support bpf syscall in amd64 Linux
1302 388862 Add replacements for wmemchr and wcsnlen on Linux
1303 389065 valgrind meets gcc flag -Wlogical-op
1304 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
1305 390471 suppression by specification of source-file line number
1306 390723 make xtree dump files world wide readable, similar to log files
1307 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
1308 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
1309 392118 unhandled amd64-linux syscall: 332 (statx)
1310 392449 callgrind not clearing the number of calls properly
1311 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
1312 lxssp, stxssp and stxvl instructions.
1313 393023 callgrind_control risks using the wrong vgdb
1314 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
1315 393099 posix_memalign() invalid write if alignment == 0
1316 393146 failing assert "is_DebugInfo_active(di)"
1317 395709 PPC64 is missing support for the xvnegsp instruction
1318 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
1320 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
1321 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
1322 396839 s390x: Trap instructions not implemented
1323 396887 arch_prctl should return EINVAL on unknown option
1324 == 397286 crash before launching binary (Unsupported arch_prctl option)
1325 == 397393 valgrind: the 'impossible' happened: (Archlinux)
1326 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
1327 396906 compile tests failure on mips32-linux: broken inline asm in tests on
1329 397012 glibc ld.so uses arch_prctl on i386
1330 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
1331 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
1332 397424 glibc 2.27 and gdb_server tests
1333 398028 Assertion `cfsi_fits` failing in simple C program
1334 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
1336 n-i-bz Fix missing workq_ops operations (macOS)
1337 n-i-bz fix bug in strspn replacement
1338 n-i-bz Add support for the Linux BLKFLSBUF ioctl
1339 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
1340 n-i-bz Fix possible stack trashing by semctl syscall wrapping
1341 n-i-bz Add support for the Linux membarrier() system call
1342 n-i-bz x86 front end: recognise and handle UD2 correctly
1343 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
1344 correctly aligned before entering the handler.
1346 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
1347 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
1348 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
1352 Release 3.13.0 (15 June 2017)
1353 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1355 3.13.0 is a feature release with many improvements and the usual collection of
1358 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1359 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1360 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1361 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
1363 * ==================== CORE CHANGES ===================
1365 * The translation cache size has been increased to keep up with the demands of
1366 large applications. The maximum number of sectors has increased from 24 to
1367 48. The default number of sectors has increased from 16 to 32 on all
1368 targets except Android, where the increase is from 6 to 12.
1370 * The amount of memory that Valgrind can use has been increased from 64GB to
1371 128GB. In particular this means your application can allocate up to about
1372 60GB when running on Memcheck.
1374 * Valgrind's default load address has been changed from 0x3800'0000 to
1375 0x5800'0000, so as to make it possible to load larger executables. This
1376 should make it possible to load executables of size at least 1200MB.
1378 * A massive spaceleak caused by reading compressed debuginfo files has been
1379 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
1382 * The C++ demangler has been updated.
1384 * Support for demangling Rust symbols has been added.
1386 * A new representation of stack traces, the "XTree", has been added. An XTree
1387 is a tree of stacktraces with data associated with the stacktraces. This is
1388 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
1389 consumption of your program. Reporting is controlled by the new options
1390 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
1392 A report can also be produced on demand using the gdbserver monitor command
1393 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
1394 format' and 'massif format. The existing visualisers for these formats (e.g.
1395 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
1396 analyse these reports.
1398 Memcheck can also produce XTree leak reports using the Callgrind file
1399 format. For more details, see the user manual.
1401 * ================== PLATFORM CHANGES =================
1403 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
1405 * amd64: fixes for JIT failure problems on long AVX2 code blocks
1407 * amd64 and x86: support for CET prefixes has been added
1409 * arm32: a few missing ARMv8 instructions have been implemented
1411 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
1412 Store-Conditional instructions has been added. This is to deal with
1413 processor implementations that implement the LL/SC specifications strictly
1414 and as a result cause Valgrind to hang in certain situations. The
1415 alternative implementation is automatically enabled at startup, as required.
1416 You can use the option --sim-hints=fallback-llsc to force-enable it if you
1419 * Support for OSX 10.12 has been improved.
1421 * On Linux, clone handling has been improved to honour CLONE_VFORK that
1422 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
1423 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
1424 CLONE_VM exact semantics will (still) not work.
1426 * The TileGX/Linux port has been removed because it appears to be both unused
1429 * ==================== TOOL CHANGES ====================
1433 - Memcheck should give fewer false positives when running optimised
1434 Clang/LLVM generated code.
1436 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1438 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
1439 to produce the end of execution leak report in a xtree callgrind format
1442 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
1443 the leak report in an xtree file.
1447 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1449 - For some workloads (typically, for big applications), Massif memory
1450 consumption and CPU consumption has decreased significantly.
1454 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1456 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
1457 for Ada gnat compiled applications.
1459 * ==================== OTHER CHANGES ====================
1461 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
1462 append the inner guest stacktrace to the inner host stacktrace. This helps
1463 to investigate the errors reported by the outer, when they are caused by the
1464 inner guest program (such as an inner regtest). See README_DEVELOPERS for
1467 * To allow fast detection of callgrind files by desktop environments and file
1468 managers, the format was extended to have an optional first line that
1469 uniquely identifies the format ("# callgrind format"). Callgrind creates
1470 this line now, as does the new xtree functionality.
1472 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
1473 have a new %n format letter that is replaced by a sequence number.
1475 * "--version -v" now shows the SVN revision numbers from which Valgrind was
1478 * ==================== FIXED BUGS ====================
1480 The following bugs have been fixed or resolved. Note that "n-i-bz"
1481 stands for "not in bugzilla" -- that is, a bug that was reported to us
1482 but never got a bugzilla entry. We encourage you to file bugs in
1483 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1484 than mailing the developers (or mailing lists) directly -- bugs that
1485 are not entered into bugzilla tend to get forgotten about or ignored.
1487 To see details of a given bug, visit
1488 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1489 where XXXXXX is the bug number as listed below.
1491 162848 --log-file output isn't split when a program forks
1492 340777 Illegal instruction on mips (ar71xx)
1493 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
1494 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
1495 to a different stack.
1496 344139 x86 stack-seg overrides, needed by the Wine people
1497 344524 store conditional of guest applications always fail - observed on
1499 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
1500 352395 Please provide SVN revision info in --version -v
1501 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
1502 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
1503 INVALID_POSIX_THREADID' failed
1504 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
1505 barrier implementation
1506 358697 valgrind.h: Some code remains even when defining NVALGRIND
1507 359202 Add musl libc configure/compile
1508 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
1509 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
1510 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
1511 362223 assertion failed when .valgrindrc is a directory instead of a file
1512 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
1513 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
1514 368507 can't malloc chunks larger than about 34GB
1515 368529 Android arm target link error, missing atexit and pthread_atfork
1516 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
1517 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
1518 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
1519 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
1520 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
1521 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
1522 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
1523 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
1524 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
1525 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
1526 370028 Reduce the number of compiler warnings on MIPS platforms
1527 370635 arm64 missing syscall getcpu
1528 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
1529 371227 Clean AArch64 syscall table
1530 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
1531 371471 Valgrind complains about non legit memory leaks on placement new (C++)
1532 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
1533 371503 disInstr(arm64): unhandled instruction 0xF89F0000
1534 371869 support '%' in symbol Z-encoding
1535 371916 execution tree xtree concept
1536 372120 c++ demangler demangles symbols which are not c++
1537 372185 Support of valgrind on ARMv8 with 32 bit executable
1538 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1539 372195 Power PC, xxsel instruction is not always recognized.
1540 372504 Hanging on exit_group
1541 372600 process loops forever when fatal signals are arriving quickly
1542 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1543 373046 Stacks registered by core are never deregistered
1544 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1545 373086 Implement additional Xen hypercalls
1546 373192 Calling posix_spawn in glibc 2.24 completely broken
1547 373488 Support for fanotify API on ARM64 architecture
1548 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1549 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1550 373938 const IRExpr arguments for matchIRExpr()
1551 374719 some spelling fixes
1552 374963 increase valgrind's load address to prevent mmap failure
1553 375514 valgrind_get_tls_addr() does not work in case of static TLS
1554 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1555 for ML_(find_rx_mapping)()
1556 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1557 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1558 == 377159 "vex: the `impossible' happened" still present
1559 == 375150 Assertion 'tres.status == VexTransOK' failed
1560 == 378068 valgrind crashes on AVX2 function in FFmpeg
1561 376142 Segfaults on MIPS Cavium Octeon boards
1562 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1563 376455 Solaris: unhandled syscall lgrpsys(180)
1564 376518 Solaris: unhandled fast trap getlgrp(6)
1565 376611 ppc64 and arm64 don't know about prlimit64 syscall
1566 376729 PPC64, remove R2 from the clobber list
1568 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1569 to be wrongly marked as addressable
1570 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1571 PIE enabled by default
1572 377376 memcheck/tests/linux/getregset fails with glibc2.24
1573 377427 PPC64, lxv instruction failing on odd destination register
1574 377478 PPC64: ISA 3.0 setup fixes
1575 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1576 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1577 and 2 args for FUTEX_TRYLOCK_PI
1578 377717 Fix massive space leak when reading compressed debuginfo sections
1579 377891 Update Xen 4.6 domctl wrappers
1580 377930 fcntl syscall wrapper is missing flock structure check
1581 378524 libvexmultiarch_test regression on s390x and ppc64
1582 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1583 378673 Update libiberty demangler
1584 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1585 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1586 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1587 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1588 (task_register_dyld_image_infos)
1589 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1590 (task_register_dyld_shared_cache_image_info)
1591 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1592 379473 MIPS: add support for rdhwr cycle counter register
1593 379504 remove TileGX/Linux port
1594 379525 Support more x86 nop opcodes
1595 379838 disAMode(x86): not an addr!
1596 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1597 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1598 379895 clock_gettime does not execute POST syscall wrapper
1599 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1600 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1601 380200 xtree generated callgrind files refer to files without directory name
1602 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1603 380397 s390x: __GI_strcspn() replacement needed
1604 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1606 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1607 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1608 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1612 Release 3.12.0 (20 October 2016)
1613 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1615 3.12.0 is a feature release with many improvements and the usual
1616 collection of bug fixes.
1618 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1619 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1620 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1621 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1622 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1623 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1625 * ================== PLATFORM CHANGES =================
1627 * POWER: Support for ISA 3.0 has been added
1629 * mips: support for O32 FPXX ABI has been added.
1630 * mips: improved recognition of different processors
1631 * mips: determination of page size now done at run time
1633 * amd64: Partial support for AMD FMA4 instructions.
1635 * arm, arm64: Support for v8 crypto and CRC instructions.
1637 * Improvements and robustification of the Solaris port.
1639 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1641 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1642 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1643 where possible. Valgrind's support for 32-bit x86 has stagnated in
1644 recent years and has fallen far behind that for 64-bit x86
1645 instructions. By contrast 64-bit x86 is well supported, up to and
1648 * ==================== TOOL CHANGES ====================
1652 - Added meta mempool support for describing a custom allocator which:
1653 - Auto-frees all chunks assuming that destroying a pool destroys all
1655 - Uses itself to allocate other memory blocks
1657 - New flag --ignore-range-below-sp to ignore memory accesses below
1658 the stack pointer, if you really have to. The related flag
1659 --workaround-gcc296-bugs=yes is now deprecated. Use
1660 --ignore-range-below-sp=1024-1 as a replacement.
1664 - Improved thread startup time significantly on non-Linux platforms.
1668 - Added collection of the metric "tot-blocks-allocd"
1670 * ==================== OTHER CHANGES ====================
1672 * Replacement/wrapping of malloc/new related functions is now done not just
1673 for system libraries by default, but for any globally defined malloc/new
1674 related function (both in shared libraries and statically linked alternative
1675 malloc implementations). The dynamic (runtime) linker is excluded, though.
1676 To only intercept malloc/new related functions in
1677 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1678 "nouserintercepts" can be any non-existing library name).
1679 This new functionality is not implemented for MacOS X.
1681 * The maximum number of callers in a suppression entry is now equal to
1682 the maximum size for --num-callers (500).
1683 Note that --gen-suppressions=yes|all similarly generates suppressions
1684 containing up to --num-callers frames.
1686 * New and modified GDB server monitor features:
1688 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1689 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1692 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1693 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1696 * Valgrind is able to read compressed debuginfo sections in two formats:
1697 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1698 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1700 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1701 for the most common use case (x86_64-linux, Memcheck) has been
1704 * Improved performance for programs that do a lot of discarding of
1705 instruction address ranges of 8KB or less.
1707 * The C++ symbol demangler has been updated.
1709 * More robustness against invalid syscall parameters on Linux.
1711 * ==================== FIXED BUGS ====================
1713 The following bugs have been fixed or resolved. Note that "n-i-bz"
1714 stands for "not in bugzilla" -- that is, a bug that was reported to us
1715 but never got a bugzilla entry. We encourage you to file bugs in
1716 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1717 than mailing the developers (or mailing lists) directly -- bugs that
1718 are not entered into bugzilla tend to get forgotten about or ignored.
1720 To see details of a given bug, visit
1721 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1722 where XXXXXX is the bug number as listed below.
1724 191069 Exiting due to signal not reported in XML output
1725 199468 Suppressions: stack size limited to 25
1726 while --num-callers allows more frames
1727 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1728 278744 cvtps2pd with redundant RexW
1729 303877 valgrind doesn't support compressed debuginfo sections.
1730 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1731 348345 Assertion fails for negative lineno
1732 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1733 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1734 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1735 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1736 352197 mips: mmap2() not wrapped correctly for page size > 4096
1737 353083 arm64 doesn't implement various xattr system calls
1738 353084 arm64 doesn't support sigpending system call
1739 353137 www: update info for Supported Platforms
1740 353138 www: update "The Valgrind Developers" page
1741 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1744 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1745 353398 WARNING: unhandled amd64-solaris syscall: 207
1746 353660 XML in auxwhat tag not escaping reserved symbols properly
1747 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1748 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1749 353802 ELF debug info reader confused with multiple .rodata sections
1750 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1751 353917 unhandled amd64-solaris syscall fchdir(120)
1752 353920 unhandled amd64-solaris syscall: 170
1753 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1754 354392 unhandled amd64-solaris syscall: 171
1755 354797 Vbit test does not include Iops for Power 8 instruction support
1756 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1760 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1761 355188 valgrind should intercept all malloc related global functions
1762 355454 do not intercept malloc related symbols from the runtime linker
1763 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1764 356044 Dwarf line info reader misinterprets is_stmt register
1765 356112 mips: replace addi with addiu
1766 356393 valgrind (vex) crashes because isZeroU happened
1769 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1770 356678 arm64-linux: unhandled syscall 232 (mincore)
1771 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1772 356823 Unsupported ARM instruction: stlex
1773 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1774 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1775 357673 crash if I try to run valgrind with a binary link with libcurl
1776 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1777 357871 pthread_spin_destroy not properly wrapped
1778 357887 Calls to VG_(fclose) do not close the file descriptor
1779 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1780 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1781 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1782 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1783 359181 Buffer Overflow during Demangling
1784 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1785 359289 s390x: popcnt (B9E1) not implemented
1786 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1787 359503 Add missing syscalls for aarch64 (arm64)
1788 359645 "You need libc6-dbg" help message could be more helpful
1789 359703 s390: wire up separate socketcalls system calls
1790 359724 getsockname might crash - deref_UInt should call safe_to_deref
1791 359733 amd64 implement ld.so strchr/index override like x86
1792 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1793 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1795 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1796 359871 Incorrect mask handling in ppoll
1797 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1798 360008 Contents of Power vr registers contents is not printed correctly when
1799 the --vgdb-shadow-registers=yes option is used
1800 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1801 non-zero shadow bits
1802 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1803 360425 arm64 unsupported instruction ldpsw
1805 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1806 360571 Error about the Android Runtime reading below the stack pointer on ARM
1807 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1808 360749 kludge for multiple .rodata sections on Solaris no longer needed
1809 360752 raise the number of reserved fds in m_main.c from 10 to 12
1810 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1811 361226 s390x: risbgn (EC59) not implemented
1812 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1813 361354 ppc64[le]: wire up separate socketcalls system calls
1814 361615 Inconsistent termination for multithreaded process terminated by signal
1815 361926 Unhandled Solaris syscall: sysfs(84)
1816 362009 V dumps core on unimplemented functionality before threads are created
1817 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1818 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1819 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1820 362953 Request for an update to the Valgrind Developers page
1821 363680 add renameat2() support
1822 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1823 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1824 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1825 364058 clarify in manual limitations of array overruns detections
1826 364413 pselect sycallwrapper mishandles NULL sigmask
1827 364728 Power PC, missing support for several HW registers in
1828 get_otrack_shadow_offset_wrk()
1829 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1830 365273 Invalid write to stack location reported after signal handler runs
1831 365912 ppc64BE segfault during jm-insns test (RELRO)
1832 366079 FPXX Support for MIPS32 Valgrind
1833 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1834 366344 Multiple unhandled instruction for Aarch64
1835 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1836 367995 Integration of memcheck with custom memory allocator
1837 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1838 368412 False positive result for altivec capability check
1839 368416 Add tc06_two_races_xml.exp output for ppc64
1840 368419 Perf Events ioctls not implemented
1841 368461 mmapunmap test fails on ppc64
1842 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1843 369000 AMD64 fma4 instructions unsupported.
1844 369169 ppc64 fails jm_int_isa_2_07 test
1845 369175 jm_vec_isa_2_07 test crashes on ppc64
1846 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1847 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1848 369359 msghdr_foreachfield can crash when handling bad iovec
1849 369360 Bad sigprocmask old or new sets can crash valgrind
1850 369361 vmsplice syscall wrapper crashes on bad iovec
1851 369362 Bad sigaction arguments crash valgrind
1852 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1853 369402 Bad set/get_thread_area pointer crashes valgrind
1854 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1855 369446 valgrind crashes on unknown fcntl command
1856 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1857 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1858 370265 ISA 3.0 HW cap stuff needs updating
1859 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1860 372195 Power PC, xxsel instruction is not always recognized
1862 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1863 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1864 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1865 n-i-bz Fix ppoll_alarm exclusion on OS X
1866 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1867 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1868 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1869 n-i-bz amd64: memcheck false positive with shr %edx
1870 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1871 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1872 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1873 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1875 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1876 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1877 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1881 Release 3.11.0 (22 September 2015)
1882 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1884 3.11.0 is a feature release with many improvements and the usual
1885 collection of bug fixes.
1887 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1888 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1889 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1890 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1891 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1892 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1894 * ================== PLATFORM CHANGES =================
1896 * Support for Solaris/x86 and Solaris/amd64 has been added.
1898 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1900 * Preliminary support for the Tilera TileGX architecture has been added.
1902 * s390x: It is now required for the host to have the "long displacement"
1903 facility. The oldest supported machine model is z990.
1905 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1906 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1908 * The JIT's register allocator is significantly faster, making the JIT
1909 as a whole somewhat faster, so JIT-intensive activities, for example
1910 program startup, are modestly faster, around 5%.
1912 * There have been changes to the default settings of several command
1913 line flags, as detailed below.
1915 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1916 capable hosts, the simulated CPUID will now indicate AVX2 support.
1918 * ==================== TOOL CHANGES ====================
1922 - The default value for --leak-check-heuristics has been changed from
1923 "none" to "all". This helps to reduce the number of possibly
1924 lost blocks, in particular for C++ applications.
1926 - The default value for --keep-stacktraces has been changed from
1927 "malloc-then-free" to "malloc-and-free". This has a small cost in
1928 memory (one word per malloc-ed block) but allows Memcheck to show the
1929 3 stacktraces of a dangling reference: where the block was allocated,
1930 where it was freed, and where it is acccessed after being freed.
1932 - The default value for --partial-loads-ok has been changed from "no" to
1933 "yes", so as to avoid false positive errors resulting from some kinds
1934 of vectorised loops.
1936 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1937 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1938 than get_vbits when you need to associate byte data value with
1939 their corresponding validity bits.
1941 - The 'block_list' monitor command has been enhanced:
1942 o it can print a range of loss records
1943 o it now accepts an optional argument 'limited <max_blocks>'
1944 to control the number of blocks printed.
1945 o if a block has been found using a heuristic, then
1946 'block_list' now shows the heuristic after the block size.
1947 o the loss records/blocks to print can be limited to the blocks
1948 found via specified heuristics.
1950 - The C helper functions used to instrument loads on
1951 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1952 replaced by handwritten assembly sequences. This gives speedups
1953 in the region of 0% to 7% for those targets only.
1955 - A new command line option, --expensive-definedness-checks=yes|no,
1956 has been added. This is useful for avoiding occasional invalid
1957 uninitialised-value errors in optimised code. Watch out for
1958 runtime degradation, as this can be up to 25%. As always, though,
1959 the slowdown is highly application specific. The default setting
1964 - A new monitor command 'all_snapshots <filename>' dumps all
1965 snapshots taken so far.
1969 - Significant memory reduction and moderate speedups for
1970 --history-level=full for applications accessing a lot of memory
1971 with many different stacktraces.
1973 - The default value for --conflict-cache-size=N has been doubled to
1974 2000000. Users that were not using the default value should
1975 preferably also double the value they give.
1977 The default was changed due to the changes in the "full history"
1978 implementation. Doubling the value gives on average a slightly more
1979 complete history and uses similar memory (or significantly less memory
1980 in the worst case) than the previous implementation.
1982 - The Helgrind monitor command 'info locks' now accepts an optional
1983 argument 'lock_addr', which shows information about the lock at the
1986 - When using --history-level=full, the new Helgrind monitor command
1987 'accesshistory <addr> [<len>]' will show the recorded accesses for
1988 <len> (or 1) bytes at <addr>.
1990 * ==================== OTHER CHANGES ====================
1992 * The default value for the --smc-check option has been changed from
1993 "stack" to "all-non-file" on targets that provide automatic D-I
1994 cache coherence (x86, amd64 and s390x). The result is to provide,
1995 by default, transparent support for JIT generated and self-modifying
1996 code on all targets.
1998 * Mac OS X only: the default value for the --dsymutil option has been
1999 changed from "no" to "yes", since any serious usage on Mac OS X
2000 always required it to be "yes".
2002 * The command line options --db-attach and --db-command have been removed.
2003 They were deprecated in 3.10.0.
2005 * When a process dies due to a signal, Valgrind now shows the signal
2006 and the stacktrace at default verbosity (i.e. verbosity 1).
2008 * The address description logic used by Memcheck and Helgrind now
2009 describes addresses in anonymous segments, file mmap-ed segments,
2010 shared memory segments and the brk data segment.
2012 * The new option --error-markers=<begin>,<end> can be used to mark the
2013 begin/end of errors in textual output mode, to facilitate
2014 searching/extracting errors in output files that mix valgrind errors
2015 with program output.
2017 * The new option --max-threads=<number> can be used to change the number
2018 of threads valgrind can handle. The default is 500 threads which
2019 should be more than enough for most applications.
2021 * The new option --valgrind-stacksize=<number> can be used to change the
2022 size of the private thread stacks used by Valgrind. This is useful
2023 for reducing memory use or increasing the stack size if Valgrind
2024 segfaults due to stack overflow.
2026 * The new option --avg-transtab-entry-size=<number> can be used to specify
2027 the expected instrumented block size, either to reduce memory use or
2028 to avoid excessive retranslation.
2030 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
2032 * New and modified GDB server monitor features:
2034 - When a signal is reported in GDB, you can now use the GDB convenience
2035 variable $_siginfo to examine detailed signal information.
2037 - Valgrind's gdbserver now allows the user to change the signal
2038 to deliver to the process. So, use 'signal SIGNAL' to continue execution
2039 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
2040 continue without passing the signal to the process.
2042 - With GDB >= 7.10, the command 'target remote'
2043 will automatically load the executable file of the process running
2044 under Valgrind. This means you do not need to specify the executable
2045 file yourself, GDB will discover it itself. See GDB documentation about
2046 'qXfer:exec-file:read' packet for more info.
2048 * ==================== FIXED BUGS ====================
2050 The following bugs have been fixed or resolved. Note that "n-i-bz"
2051 stands for "not in bugzilla" -- that is, a bug that was reported to us
2052 but never got a bugzilla entry. We encourage you to file bugs in
2053 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2054 than mailing the developers (or mailing lists) directly -- bugs that
2055 are not entered into bugzilla tend to get forgotten about or ignored.
2057 To see details of a given bug, visit
2058 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2059 where XXXXXX is the bug number as listed below.
2061 116002 VG_(printf): Problems with justification of strings and integers
2062 155125 avoid cutting away file:lineno after long function name
2063 197259 Unsupported arch_prtctl PR_SET_GS option
2064 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
2065 201216 Fix Valgrind does not support pthread_sigmask() on OS X
2066 201435 Fix Darwin: -v does not show kernel version
2067 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
2068 211256 Fixed an outdated comment regarding the default platform.
2069 211529 Incomplete call stacks for code compiled by newer versions of MSVC
2070 211926 Avoid compilation warnings in valgrind.h with -pedantic
2071 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
2073 226609 Crediting upstream authors in man page
2074 231257 Valgrind omits path when executing script from shebang line
2075 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
2076 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
2077 269360 s390x: Fix addressing mode selection for compare-and-swap
2078 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
2080 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
2081 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
2082 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
2083 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
2084 330147 libmpiwrap PMPI_Get_count returns undefined value
2085 333051 mmap of huge pages fails due to incorrect alignment
2087 334802 valgrind does not always explain why a given option is bad
2088 335618 mov.w rN, pc/sp (ARM32)
2089 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
2094 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
2095 338602 AVX2 bit in CPUID missing
2096 338606 Strange message for scripts with invalid interpreter
2097 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
2098 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
2099 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
2101 339156 gdbsrv not called for fatal signal
2102 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
2103 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
2104 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
2105 339442 Fix testsuite build failure on OS X 10.9
2106 339542 Enable compilation with Intel's ICC compiler
2107 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
2108 339688 Mac-specific ASM does not support .version directive (cpuid,
2109 tronical and pushfpopf tests)
2110 339745 Valgrind crash when check Marmalade app (partial fix)
2111 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
2112 339778 Linux/TileGx platform support for Valgrind
2113 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
2114 339789 Fix none/tests/execve test on Mac OS X 10.9
2115 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
2116 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
2117 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
2118 340392 Allow user to select more accurate definedness checking in memcheck
2119 to avoid invalid complaints on optimised code
2120 340430 Fix some grammatical weirdness in the manual.
2121 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
2122 341419 Signal handler ucontext_t not filled out correctly on OS X
2123 341539 VG_(describe_addr) should not describe address as belonging to client
2124 segment if it is past the heap end
2125 341613 Enable building of manythreads and thread-exits tests on Mac OS X
2126 341615 Fix none/tests/darwin/access_extended test on Mac OS X
2127 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
2128 341789 aarch64: shmat fails with valgrind on ARMv8
2129 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
2130 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
2131 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
2132 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
2133 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
2134 342221 socket connect false positive uninit memory for unknown af family
2135 342353 Allow dumping full massif output while valgrind is still running
2136 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
2140 342603 Add I2C_SMBUS ioctl support
2141 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
2142 342683 Mark memory past the initial brk limit as unaddressable
2143 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
2144 342795 Internal glibc __GI_mempcpy call should be intercepted
2145 342841 s390x: Support instructions fiebr(a) and fidbr(a)
2146 343012 Unhandled syscall 319 (memfd_create)
2147 343069 Patch updating v4l2 API support
2148 343173 helgrind crash during stack unwind
2149 343219 fix GET_STARTREGS for arm
2150 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
2151 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
2152 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
2153 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
2154 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
2155 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
2156 343597 ppc64le: incorrect use of offseof macro
2157 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
2158 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
2159 343732 Unhandled syscall 144 (setgid) on aarch64
2160 343733 Unhandled syscall 187 (msgctl and related) on aarch64
2161 343802 s390x: False positive "conditional jump or move depends on [..]
2162 343902 --vgdb=yes doesn't break when --xml=yes is used
2163 343967 Don't warn about setuid/setgid/setcap executable for directories
2164 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
2165 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
2166 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
2167 344054 www - update info for Solaris/illumos
2168 344416 'make regtest' does not work cleanly on OS X
2169 344235 Remove duplicate include of pub_core_aspacemgr.h
2170 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
2171 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
2172 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
2173 344314 callgrind_annotate ... warnings about commands containing newlines
2174 344318 socketcall should wrap recvmmsg and sendmmsg
2175 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
2176 344416 Fix 'make regtest' does not work cleanly on OS X
2177 344499 Fix compilation for Linux kernel >= 4.0.0
2178 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
2179 unix:349 (__pthread_fchdir)
2180 344559 Garbage collection of unused segment names in address space manager
2181 344560 Fix stack traces missing penultimate frame on OS X
2182 344621 Fix memcheck/tests/err_disable4 test on OS X
2183 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
2184 344702 Fix missing libobjc suppressions on OS X 10.10
2186 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
2187 344939 Fix memcheck/tests/xml1 on OS X 10.10
2188 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
2189 345079 Fix build problems in VEX/useful/test_main.c
2190 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
2191 345177 arm64: prfm (reg) not implemented
2192 345215 Performance improvements for the register allocator
2193 345248 add support for Solaris OS in valgrind
2194 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
2195 345394 Fix memcheck/tests/strchr on OS X
2196 345637 Fix memcheck/tests/sendmsg on OS X
2197 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
2198 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
2199 345887 Fix an assertion in the address space manager
2200 345928 amd64: callstack only contains current function for small stacks
2201 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
2202 345987 MIPS64: Implement cavium LHX instruction
2203 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
2204 346185 Fix typo saving altivec register v24
2205 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
2206 and LibVEX_GuestPPC64_get_CR()
2207 346270 Regression tests none/tests/jm_vec/isa_2_07 and
2208 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
2209 346307 fuse filesystem syscall deadlocks
2210 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
2211 346411 MIPS: SysRes::_valEx handling is incorrect
2212 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
2213 346474 PPC64 Power 8, spr TEXASRU register not supported
2214 346487 Compiler generates "note" about a future ABI change for PPC64
2215 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
2216 and causing spurious "invalid read of size 8" warnings
2217 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
2218 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
2219 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
2220 347322 Power PC regression test cleanup
2221 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
2223 347389 unhandled syscall: 373 (Linux ARM syncfs)
2224 347686 Patch set to cleanup PPC64 regtests
2225 347978 Remove bash dependencies where not needed
2226 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
2227 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
2229 348102 Patch updating v4l2 API support
2230 348247 amd64 front end: jno jumps wrongly when overflow is not set
2231 348269 Improve mmap MAP_HUGETLB support.
2232 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
2233 348345 Assertion fails for negative lineno
2234 348377 Unsupported ARM instruction: yield
2235 348565 Fix detection of command line option availability for clang
2236 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
2237 348728 Fix broken check for VIDIOC_G_ENC_INDEX
2238 348748 Fix redundant condition
2239 348890 Fix clang warning about unsupported --param inline-unit-growth=900
2240 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
2241 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
2242 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
2243 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
2244 349626 Implemented additional Xen hypercalls
2245 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
2246 349790 Clean up of the hardware capability checking utilities.
2247 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
2248 349874 Fix typos in source code
2249 349879 memcheck: add handwritten assembly for helperc_LOADV*
2250 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
2251 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
2252 350202 Add limited param to 'monitor block_list'
2253 350290 s390x: Support instructions fixbr(a)
2254 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
2255 350809 Fix none/tests/async-sigs for Solaris
2256 350811 Remove reference to --db-attach which has been removed.
2257 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
2258 350854 hard-to-understand code in VG_(load_ELF)()
2259 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2260 351386 Solaris: Cannot run ld.so.1 under Valgrind
2261 351474 Fix VG_(iseqsigset) as obvious
2262 351531 Typo in /include/vki/vki-xen-physdev.h header guard
2263 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
2264 351858 ldsoexec support on Solaris
2265 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
2266 352130 helgrind reports false races for printfs using mempcpy on FILE* state
2267 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
2268 352320 arm64 crash on none/tests/nestedfs
2269 352765 Vbit test fails on Power 6
2270 352768 The mbar instruction is missing from the Power PC support
2271 352769 Power PC program priority register (PPR) is not supported
2272 n-i-bz Provide implementations of certain compiler builtins to support
2273 compilers that may not provide those
2274 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
2275 n-i-bz Fix compilation on distros with glibc < 2.5
2276 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
2277 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
2278 n-i-bz mremap did not work properly on shared memory
2279 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
2280 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
2282 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
2283 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
2284 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
2288 Release 3.10.1 (25 November 2014)
2289 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2290 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
2291 and backports fixes for all reported missing AArch64 ARMv8 instructions
2292 and syscalls from the trunk. If you package or deliver 3.10.0 for others
2293 to use, you might want to consider upgrading to 3.10.1 instead.
2295 The following bugs have been fixed or resolved. Note that "n-i-bz"
2296 stands for "not in bugzilla" -- that is, a bug that was reported to us
2297 but never got a bugzilla entry. We encourage you to file bugs in
2298 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2299 than mailing the developers (or mailing lists) directly -- bugs that
2300 are not entered into bugzilla tend to get forgotten about or ignored.
2302 To see details of a given bug, visit
2303 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2304 where XXXXXX is the bug number as listed below.
2306 335440 arm64: ld1 (single structure) is not implemented
2307 335713 arm64: unhanded instruction: prfm (immediate)
2308 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
2309 339182 ppc64: AvSplat ought to load destination vector register with [..]
2310 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
2311 339433 ppc64 lxvw4x instruction uses four 32-byte loads
2312 339645 Use correct tag names in sys_getdents/64 wrappers
2313 339706 Fix false positive for ioctl(TIOCSIG) on linux
2314 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
2315 339853 arm64 times syscall unknown
2316 339855 arm64 unhandled getsid/setsid syscalls
2317 339858 arm64 dmb sy not implemented
2318 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
2319 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
2320 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
2322 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
2323 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
2324 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
2325 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
2326 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
2327 340509 arm64: unhandled instruction fcvtas
2328 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
2329 340632 arm64: unhandled instruction fcvtas
2330 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
2331 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
2332 340788 warning: unhandled syscall: 318 (getrandom)
2333 340807 disInstr(arm): unhandled instruction: 0xEE989B20
2334 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
2335 340922 arm64: unhandled getgroups/setgroups syscalls
2336 350251 Fix typo in VEX utility program (test_main.c).
2337 350407 arm64: unhandled instruction ucvtf (vector, integer)
2338 350809 none/tests/async-sigs breaks when run under cron on Solaris
2339 350811 update README.solaris after r15445
2340 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
2341 350854 strange code in VG_(load_ELF)()
2342 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2343 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
2344 n-i-bz Add missing ]] to terminate CDATA.
2345 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
2346 n-i-bz Enable sys_fadvise64_64 on arm32.
2347 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
2348 n-i-bz Add test cases for all known arm64 load/store instructions.
2349 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
2350 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
2351 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
2352 n-i-bz arm64: Implement "fcvtpu w, s".
2353 n-i-bz arm64: implement ADDP and various others
2354 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
2355 n-i-bz arm64: enable FCVT{A,N}S X,S.
2357 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
2361 Release 3.10.0 (10 September 2014)
2362 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2364 3.10.0 is a feature release with many improvements and the usual
2365 collection of bug fixes.
2367 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
2368 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
2369 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
2370 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
2371 significantly improved relative to the 3.9.0 release.
2373 * ================== PLATFORM CHANGES =================
2375 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
2376 is mostly complete, and is usable, but some SIMD instructions are as
2379 * Support for little-endian variant of the 64-bit POWER architecture.
2381 * Support for Android on MIPS32.
2383 * Support for 64bit FPU on MIPS32 platforms.
2385 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
2387 * Configuration for and running on Android targets has changed.
2388 See README.android in the source tree for details.
2390 * ================== DEPRECATED FEATURES =================
2392 * --db-attach is now deprecated and will be removed in the next
2393 valgrind feature release. The built-in GDB server capabilities are
2394 superior and should be used instead. Learn more here:
2395 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
2397 * ==================== TOOL CHANGES ====================
2401 - Client code can now selectively disable and re-enable reporting of
2402 invalid address errors in specific ranges using the new client
2403 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
2404 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
2406 - Leak checker: there is a new leak check heuristic called
2407 "length64". This is used to detect interior pointers pointing 8
2408 bytes inside a block, on the assumption that the first 8 bytes
2409 holds the value "block size - 8". This is used by
2410 sqlite3MemMalloc, for example.
2412 - Checking of system call parameters: if a syscall parameter
2413 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
2414 several fields not initialised, an error is now reported for each
2415 field. Previously, an error was reported only for the first
2416 uninitialised field.
2418 - Mismatched alloc/free checking: a new flag
2419 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
2420 such checks if necessary.
2424 - Improvements to error messages:
2426 o Race condition error message involving heap allocated blocks also
2427 show the thread number that allocated the raced-on block.
2429 o All locks referenced by an error message are now announced.
2430 Previously, some error messages only showed the lock addresses.
2432 o The message indicating where a lock was first observed now also
2433 describes the address/location of the lock.
2435 - Helgrind now understands the Ada task termination rules and
2436 creates a happens-before relationship between a terminated task
2437 and its master. This avoids some false positives and avoids a big
2438 memory leak when a lot of Ada tasks are created and terminated.
2439 The interceptions are only activated with forthcoming releases of
2440 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
2442 - A new GDB server monitor command "info locks" giving the list of
2443 locks, their location, and their status.
2447 - callgrind_control now supports the --vgdb-prefix argument,
2448 which is needed if valgrind was started with this same argument.
2450 * ==================== OTHER CHANGES ====================
2452 * Unwinding through inlined function calls. Stack unwinding can now
2453 make use of Dwarf3 inlined-unwind information if it is available.
2454 The practical effect is that inlined calls become visible in stack
2455 traces. The suppression matching machinery has been adjusted
2456 accordingly. This is controlled by the new option
2457 --read-inline-info=yes|no. Currently this is enabled by default
2458 only on Linux and Android targets and only for the tools Memcheck,
2461 * Valgrind can now read EXIDX unwind information on 32-bit ARM
2462 targets. If an object contains both CFI and EXIDX unwind
2463 information, Valgrind will prefer the CFI over the EXIDX. This
2464 facilitates unwinding through system libraries on arm-android
2467 * Address description logic has been improved and is now common
2468 between Memcheck and Helgrind, resulting in better address
2469 descriptions for some kinds of error messages.
2471 * Error messages about dubious arguments (eg, to malloc or calloc) are
2472 output like other errors. This means that they can be suppressed
2473 and they have a stack trace.
2475 * The C++ demangler has been updated for better C++11 support.
2477 * New and modified GDB server monitor features:
2479 - Thread local variables/storage (__thread) can now be displayed.
2481 - The GDB server monitor command "v.info location <address>"
2482 displays information about an address. The information produced
2483 depends on the tool and on the options given to valgrind.
2484 Possibly, the following are described: global variables, local
2485 (stack) variables, allocated or freed blocks, ...
2487 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
2488 ask the GDB server to stop at the start of program execution, at
2489 the end of the program execution and on Valgrind internal errors.
2491 - A new monitor command "v.info stats" shows various Valgrind core
2492 and tool statistics.
2494 - A new monitor command "v.set hostvisibility" allows the GDB server
2495 to provide access to Valgrind internal host status/memory.
2497 * A new option "--aspace-minaddr=<address>" can in some situations
2498 allow the use of more memory by decreasing the address above which
2499 Valgrind maps memory. It can also be used to solve address
2500 conflicts with system libraries by increasing the default value.
2501 See user manual for details.
2503 * The amount of memory used by Valgrind to store debug info (unwind
2504 info, line number information and symbol data) has been
2505 significantly reduced, even though Valgrind now reads more
2506 information in order to support unwinding of inlined function calls.
2508 * Dwarf3 handling with --read-var-info=yes has been improved:
2510 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
2512 - Code compiled with
2513 -ffunction-sections -fdata-sections -Wl,--gc-sections
2514 no longer causes assertion failures.
2516 * Improved checking for the --sim-hints= and --kernel-variant=
2517 options. Unknown strings are now detected and reported to the user
2520 * The semantics of stack start/end boundaries in the valgrind.h
2521 VALGRIND_STACK_REGISTER client request has been clarified and
2522 documented. The convention is that start and end are respectively
2523 the lowest and highest addressable bytes of the stack.
2525 * ==================== FIXED BUGS ====================
2527 The following bugs have been fixed or resolved. Note that "n-i-bz"
2528 stands for "not in bugzilla" -- that is, a bug that was reported to us
2529 but never got a bugzilla entry. We encourage you to file bugs in
2530 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2531 than mailing the developers (or mailing lists) directly -- bugs that
2532 are not entered into bugzilla tend to get forgotten about or ignored.
2534 To see details of a given bug, visit
2535 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2536 where XXXXXX is the bug number as listed below.
2538 175819 Support for ipv6 socket reporting with --track-fds
2539 232510 make distcheck fails
2540 249435 Analyzing wine programs with callgrind triggers a crash
2541 278972 support for inlined function calls in stacktraces and suppression
2543 291310 FXSAVE instruction marks memory as undefined on amd64
2544 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2545 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2546 315199 vgcore file for threaded app does not show which thread crashed
2547 315952 tun/tap ioctls are not supported
2548 323178 Unhandled instruction: PLDW register (ARM)
2549 323179 Unhandled instruction: PLDW immediate (ARM)
2550 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2551 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2552 325124 [MIPSEL] Compilation error
2553 325477 Phase 4 support for IBM Power ISA 2.07
2554 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2555 325628 Phase 5 support for IBM Power ISA 2.07
2556 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2557 325751 Missing the two privileged Power PC Transactional Memory Instructions
2558 325816 Phase 6 support for IBM Power ISA 2.07
2559 325856 Make SGCheck fail gracefully on unsupported platforms
2560 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2561 326436 DRD: False positive in libstdc++ std::list::push_back
2562 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2563 326462 Refactor vgdb to isolate invoker stuff into separate module
2564 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2565 326623 DRD: false positive conflict report in a field assignment
2566 326724 Valgrind does not compile on OSX 1.9 Mavericks
2567 326816 Intercept for __strncpy_sse2_unaligned missing?
2568 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2569 326983 Clear direction flag after tests on amd64.
2570 327212 Do not prepend the current directory to absolute path names.
2571 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2572 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2573 327284 s390x: Fix translation of the risbg instruction
2574 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2575 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2576 327916 DW_TAG_typedef may have no name
2577 327943 s390x: add a redirection for the 'index' function
2578 328100 XABORT not implemented
2579 328205 Implement additional Xen hypercalls
2580 328454 add support Backtraces with ARM unwind tables (EXIDX)
2581 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2582 328711 valgrind.1 manpage "memcheck options" section is badly generated
2583 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2584 329612 Incorrect handling of AT_BASE for image execution
2585 329694 clang warns about using uninitialized variable
2586 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2587 330228 mmap must align to VKI_SHMLBA on mips32
2588 330257 LLVM does not support `-mno-dynamic-no-pic` option
2589 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2590 330459 --track-fds=yes doesn't track eventfds
2591 330469 Add clock_adjtime syscall support
2592 330594 Missing sysalls on PowerPC / uClibc
2593 330622 Add test to regression suite for POWER instruction: dcbzl
2594 330939 Support for AMD's syscall instruction on x86
2596 330941 Typo in PRE(poll) syscall wrapper
2597 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2598 331254 Fix expected output for memcheck/tests/dw4
2599 331255 Fix race condition in test none/tests/coolo_sigaction
2600 331257 Fix type of jump buffer in test none/tests/faultstatus
2601 331305 configure uses bash specific syntax
2602 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2603 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2604 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2605 331829 Unexpected ioctl opcode sign extension
2606 331830 ppc64: WARNING: unhandled syscall: 96/97
2607 331839 drd/tests/sem_open specifies invalid semaphore name
2608 331847 outcome of drd/tests/thread_name is nondeterministic
2609 332037 Valgrind cannot handle Thumb "add pc, reg"
2610 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2611 consistency checks enabled
2612 332263 intercepts for pthread_rwlock_timedrdlock and
2613 pthread_rwlock_timedwrlock are incorrect
2614 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2616 332276 Implement additional Xen hypercalls
2617 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2618 332765 Fix ms_print to create temporary files in a proper directory
2619 333072 drd: Add semaphore annotations
2620 333145 Tests for missaligned PC+#imm access for arm
2621 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2622 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2623 333248 WARNING: unhandled syscall: unix:443
2624 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2625 333501 cachegrind: assertion: Cache set count is not a power of two.
2628 333666 Recognize MPX instructions and bnd prefix.
2629 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2630 333817 Valgrind reports the memory areas written to by the SG_IO
2632 334049 lzcnt fails silently (x86_32)
2633 334384 Valgrind does not have support Little Endian support for
2635 334585 recvmmsg unhandled (+patch) (arm)
2636 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2637 334727 Build fails with -Werror=format-security
2638 334788 clarify doc about --log-file initial program directory
2639 334834 PPC64 Little Endian support, patch 2
2640 334836 PPC64 Little Endian support, patch 3 testcase fixes
2641 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2642 335034 Unhandled ioctl: HCIGETDEVLIST
2643 335155 vgdb, fix error print statement.
2644 335262 arm64: movi 8bit version is not supported
2645 335263 arm64: dmb instruction is not implemented
2646 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2647 335496 arm64: sbc/abc instructions are not implemented
2648 335554 arm64: unhandled instruction: abs
2649 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2650 335735 arm64: unhandled instruction: cnt
2651 335736 arm64: unhandled instruction: uaddlv
2652 335848 arm64: unhandled instruction: {s,u}cvtf
2653 335902 arm64: unhandled instruction: sli
2654 335903 arm64: unhandled instruction: umull (vector)
2655 336055 arm64: unhandled instruction: mov (element)
2656 336062 arm64: unhandled instruction: shrn{,2}
2657 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2658 336189 arm64: unhandled Instruction: mvn
2659 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2660 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2661 336772 Make moans about unknown ioctls more informative
2662 336957 Add a section about the Solaris/illumos port on the webpage
2663 337094 ifunc wrapper is broken on ppc64
2664 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2665 337528 leak check heuristic for block prefixed by length as 64bit number
2666 337740 Implement additional Xen hypercalls
2667 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2668 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2669 337871 deprecate --db-attach
2670 338023 Add support for all V4L2/media ioctls
2671 338024 inlined functions are not shown if DW_AT_ranges is used
2672 338106 Add support for 'kcmp' syscall
2673 338115 DRD: computed conflict set differs from actual after fork
2674 338160 implement display of thread local storage in gdbsrv
2675 338205 configure.ac and check for -Wno-tautological-compare
2676 338300 coredumps are missing one byte of every segment
2677 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2678 338499 --sim-hints parsing broken due to wrong order in tokens
2679 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2680 338681 Unable to unwind through clone thread created on i386-linux
2681 338698 race condition between gdbsrv and vgdb on startup
2682 338703 helgrind on arm-linux gets false positives in dynamic loader
2683 338791 alt dwz files can be relative of debug/main file
2684 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2685 338932 build V-trunk with gcc-trunk
2686 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2687 345079 Fix build problems in VEX/useful/test_main.c
2688 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2689 n-i-bz s390x: Fix memory corruption for multithreaded applications
2690 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2691 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2692 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2693 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2694 n-i-bz Add command-line option --default-suppressions
2695 n-i-bz Add support for BLKDISCARDZEROES ioctl
2696 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2697 n-i-bz Add support for sys_pivot_root and sys_unshare
2699 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2700 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2701 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2705 Release 3.9.0 (31 October 2013)
2706 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2707 3.9.0 is a feature release with many improvements and the usual
2708 collection of bug fixes.
2710 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2711 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2712 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2713 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2715 * ================== PLATFORM CHANGES =================
2717 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2718 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2720 * Support for MIPS DSP ASE on MIPS32 platforms.
2722 * Support for s390x Decimal Floating Point instructions on hosts that
2723 have the DFP facility installed.
2725 * Support for POWER8 (Power ISA 2.07) instructions
2727 * Support for Intel AVX2 instructions. This is available only on 64
2730 * Initial support for Intel Transactional Synchronization Extensions,
2733 * Initial support for Hardware Transactional Memory on POWER.
2735 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2736 run large GUI apps tolerably well.
2738 * ==================== TOOL CHANGES ====================
2742 - Improvements in handling of vectorised code, leading to
2743 significantly fewer false error reports. You need to use the flag
2744 --partial-loads-ok=yes to get the benefits of these changes.
2746 - Better control over the leak checker. It is now possible to
2747 specify which leak kinds (definite/indirect/possible/reachable)
2748 should be displayed, which should be regarded as errors, and which
2749 should be suppressed by a given leak suppression. This is done
2750 using the options --show-leak-kinds=kind1,kind2,..,
2751 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2752 "match-leak-kinds:" line in suppression entries, respectively.
2754 Note that generated leak suppressions contain this new line and
2755 are therefore more specific than in previous releases. To get the
2756 same behaviour as previous releases, remove the "match-leak-kinds:"
2757 line from generated suppressions before using them.
2759 - Reduced "possible leak" reports from the leak checker by the use
2760 of better heuristics. The available heuristics provide detection
2761 of valid interior pointers to std::stdstring, to new[] allocated
2762 arrays with elements having destructors and to interior pointers
2763 pointing to an inner part of a C++ object using multiple
2764 inheritance. They can be selected individually using the
2765 option --leak-check-heuristics=heur1,heur2,...
2767 - Better control of stacktrace acquisition for heap-allocated
2768 blocks. Using the --keep-stacktraces option, it is possible to
2769 control independently whether a stack trace is acquired for each
2770 allocation and deallocation. This can be used to create better
2771 "use after free" errors or to decrease Valgrind's resource
2772 consumption by recording less information.
2774 - Better reporting of leak suppression usage. The list of used
2775 suppressions (shown when the -v option is given) now shows, for
2776 each leak suppressions, how many blocks and bytes it suppressed
2777 during the last leak search.
2781 - False errors resulting from the use of statically initialised
2782 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2785 - False errors resulting from the use of pthread_cond_waits that
2786 timeout, have been removed.
2788 * ==================== OTHER CHANGES ====================
2790 * Some attempt to tune Valgrind's space requirements to the expected
2791 capabilities of the target:
2793 - The default size of the translation cache has been reduced from 8
2794 sectors to 6 on Android platforms, since each sector occupies
2795 about 40MB when using Memcheck.
2797 - The default size of the translation cache has been increased to 16
2798 sectors on all other platforms, reflecting the fact that large
2799 applications require instrumentation and storage of huge amounts
2800 of code. For similar reasons, the number of memory mapped
2801 segments that can be tracked has been increased by a factor of 6.
2803 - In all cases, the maximum number of sectors in the translation
2804 cache can be controlled by the new flag --num-transtab-sectors.
2806 * Changes in how debug info (line numbers, etc) is read:
2808 - Valgrind no longer temporarily mmaps the entire object to read
2809 from it. Instead, reading is done through a small fixed sized
2810 buffer. This avoids virtual memory usage spikes when Valgrind
2811 reads debuginfo from large shared objects.
2813 - A new experimental remote debug info server. Valgrind can read
2814 debug info from a different machine (typically, a build host)
2815 where debuginfo objects are stored. This can save a lot of time
2816 and hassle when running Valgrind on resource-constrained targets
2817 (phones, tablets) when the full debuginfo objects are stored
2818 somewhere else. This is enabled by the --debuginfo-server=
2821 - Consistency checking between main and debug objects can be
2822 disabled using the --allow-mismatched-debuginfo option.
2824 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2825 scanning can recover stack traces in some cases when the normal
2826 unwind mechanisms fail. Stack scanning is best described as "a
2827 nasty, dangerous and misleading hack" and so is disabled by default.
2828 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2831 * Detection and merging of recursive stack frame cycles. When your
2832 program has recursive algorithms, this limits the memory used by
2833 Valgrind for recorded stack traces and avoids recording
2834 uninteresting repeated calls. This is controlled by the command
2835 line option --merge-recursive-frame and by the monitor command
2836 "v.set merge-recursive-frames".
2838 * File name and line numbers for used suppressions. The list of used
2839 suppressions (shown when the -v option is given) now shows, for each
2840 used suppression, the file name and line number where the suppression
2843 * New and modified GDB server monitor features:
2845 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2846 that can be used to execute gdbserver monitor commands from the
2849 - A new monitor command, "v.info open_fds", that gives the list of
2850 open file descriptors and additional details.
2852 - An optional message in the "v.info n_errs_found" monitor command,
2853 for example "v.info n_errs_found test 1234 finished", allowing a
2854 comment string to be added to the process output, perhaps for the
2855 purpose of separating errors of different tests or test phases.
2857 - A new monitor command "v.info execontext" that shows information
2858 about the stack traces recorded by Valgrind.
2860 - A new monitor command "v.do expensive_sanity_check_general" to run
2861 some internal consistency checks.
2863 * New flag --sigill-diagnostics to control whether a diagnostic
2864 message is printed when the JIT encounters an instruction it can't
2865 translate. The actual behavior -- delivery of SIGILL to the
2866 application -- is unchanged.
2868 * The maximum amount of memory that Valgrind can use on 64 bit targets
2869 has been increased from 32GB to 64GB. This should make it possible
2870 to run applications on Memcheck that natively require up to about 35GB.
2872 * ==================== FIXED BUGS ====================
2874 The following bugs have been fixed or resolved. Note that "n-i-bz"
2875 stands for "not in bugzilla" -- that is, a bug that was reported to us
2876 but never got a bugzilla entry. We encourage you to file bugs in
2877 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2878 than mailing the developers (or mailing lists) directly -- bugs that
2879 are not entered into bugzilla tend to get forgotten about or ignored.
2881 To see details of a given bug, visit
2882 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2883 where XXXXXX is the bug number as listed below.
2885 123837 system call: 4th argument is optional, depending on cmd
2886 135425 memcheck should tell you where Freed blocks were Mallocd
2887 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2888 207815 Adds some of the drm ioctls to syswrap-linux.c
2889 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2890 252955 Impossible to compile with ccache
2891 253519 Memcheck reports auxv pointer accesses as invalid reads.
2892 263034 Crash when loading some PPC64 binaries
2893 269599 Increase deepest backtrace
2894 274695 s390x: Support "compare to/from logical" instructions (z196)
2895 275800 s390x: Autodetect cache info (part 2)
2896 280271 Valgrind reports possible memory leaks on still-reachable std::string
2897 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2898 289578 Backtraces with ARM unwind tables (stack scan flags)
2899 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2900 304832 ppc32: build failure
2901 305431 Use find_buildid shdr fallback for separate .debug files
2902 305728 Add support for AVX2 instructions
2903 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2904 306035 s390x: Fix IR generation for LAAG and friends
2905 306054 s390x: Condition code computation for convert-to-int/logical
2906 306098 s390x: alternate opcode form for convert to/from fixed
2907 306587 Fix cache line detection from auxiliary vector for PPC.
2908 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2909 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2910 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2911 307101 sys_capget second argument can be NULL
2912 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2913 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2914 307113 s390x: DFP support
2915 307141 valgrind does't work in mips-linux system
2916 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2917 307285 x86_amd64 feature test for avx in test suite is wrong
2918 307290 memcheck overlap testcase needs memcpy version filter
2919 307463 Please add "&limit=0" to the "all open bugs" link
2920 307465 --show-possibly-lost=no should reduce the error count / exit code
2921 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2922 307729 pkgconfig support broken valgrind.pc
2923 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2924 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2925 308089 Unhandled syscall on ppc64: prctl
2926 308135 PPC32 MPC8xx has 16 bytes cache size
2927 308321 testsuite memcheck filter interferes with gdb_filter
2929 308341 vgdb should report process exit (or fatal signal)
2930 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2931 308495 Remove build dependency on installed Xen headers
2932 308573 Internal error on 64-bit instruction executed in 32-bit mode
2934 308627 pmovmskb validity bit propagation is imprecise
2935 308644 vgdb command for having the info for the track-fds option
2936 308711 give more info about aspacemgr and arenas in out_of_memory
2937 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2938 308718 ARM implement SMLALBB family of instructions
2939 308886 Missing support for PTRACE_SET/GETREGSET
2940 308930 syscall name_to_handle_at (303 on amd64) not handled
2941 309229 V-bit tester does not report number of tests generated
2942 309323 print unrecognized instuction on MIPS
2943 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2944 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2945 309430 Self hosting ppc64 encounters a vassert error on operand type
2946 309600 valgrind is a bit confused about 0-sized sections
2947 309823 Generate errors for still reachable blocks
2948 309921 PCMPISTRI validity bit propagation is imprecise
2949 309922 none/tests/ppc64/test_dfp5 sometimes fails
2950 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2951 310424 --read-var-info does not properly describe static variables
2952 310792 search additional path for debug symbols
2953 310931 s390x: Message-security assist (MSA) instruction extension [..]
2954 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2955 311318 ARM: "128-bit constant is not implemented" error message
2956 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2957 311690 V crashes because it redirects branches inside of a redirected function
2958 311880 x86_64: make regtest hangs at shell_valid1
2959 311922 WARNING: unhandled syscall: 170
2961 312171 ppc: insn selection for DFP
2962 312571 Rounding mode call wrong for the DFP Iops [..]
2963 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2964 312913 Dangling pointers error should also report the alloc stack trace
2965 312980 Building on Mountain Lion generates some compiler warnings
2966 313267 Adding MIPS64/Linux port to Valgrind
2969 313811 Buffer overflow in assert_fail
2970 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2971 314269 ppc: dead code in insn selection
2972 314718 ARM: implement integer divide instruction (sdiv and udiv)
2973 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2974 315441 sendmsg syscall should ignore unset msghdr msg_flags
2975 315534 msgrcv inside a thread causes valgrind to hang (block)
2976 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2977 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2978 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2979 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2980 316144 valgrind.1 manpage contains unknown ??? strings [..]
2981 316145 callgrind command line options in manpage reference (unknown) [..]
2982 316145 callgrind command line options in manpage reference [..]
2983 316181 drd: Fixed a 4x slowdown for certain applications
2984 316503 Valgrind does not support SSE4 "movntdqa" instruction
2985 316535 Use of |signed int| instead of |size_t| in valgrind messages
2986 316696 fluidanimate program of parsec 2.1 stuck
2987 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2988 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2989 317186 "Impossible happens" when occurs VCVT instruction on ARM
2990 317318 Support for Threading Building Blocks "scalable_malloc"
2991 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2992 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2993 317463 bmi testcase IR SANITY CHECK FAILURE
2994 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2995 318050 libmpiwrap fails to compile with out-of-source build
2996 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2997 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2998 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2999 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
3000 318932 Add missing PPC64 and PPC32 system call support
3001 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
3002 319395 Crash with unhandled instruction on STRT (Thumb) instructions
3003 319494 VEX Makefile-gcc standalone build update after r2702
3004 319505 [MIPSEL] Crash: unhandled UNRAY operator.
3005 319858 disInstr(thumb): unhandled instruction on instruction STRBT
3006 319932 disInstr(thumb): unhandled instruction on instruction STRHT
3007 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
3008 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
3009 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
3010 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
3011 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
3012 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
3013 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
3014 320895 add fanotify support (patch included)
3015 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
3016 321065 Valgrind updates for Xen 4.3
3017 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
3018 321363 Unhandled instruction: SSAX (ARM + Thumb)
3019 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
3020 321466 Unhandled instruction: SHASX (ARM + Thumb)
3021 321467 Unhandled instruction: SHSAX (ARM + Thumb)
3022 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
3023 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
3024 321620 Unhandled instruction: UASX (ARM + Thumb)
3025 321621 Unhandled instruction: USAX (ARM + Thumb)
3026 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
3027 321693 Unhandled instruction: LDRSBT (Thumb)
3028 321694 Unhandled instruction: UQASX (ARM + Thumb)
3029 321696 Unhandled instruction: UQSAX (Thumb + ARM)
3030 321697 Unhandled instruction: UHASX (ARM + Thumb)
3031 321703 Unhandled instruction: UHSAX (ARM + Thumb)
3032 321704 Unhandled instruction: REVSH (ARM + Thumb)
3033 321730 Add cg_diff and cg_merge man pages
3034 321738 Add vgdb and valgrind-listener man pages
3036 321891 Unhandled instruction: LDRHT (Thumb)
3037 321960 pthread_create() then alloca() causing invalid stack write errors
3038 321969 ppc32 and ppc64 don't support [lf]setxattr
3039 322254 Show threadname together with tid if set by application
3040 322294 Add initial support for IBM Power ISA 2.07
3041 322368 Assertion failure in wqthread_hijack under OS X 10.8
3042 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
3043 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
3044 322851 0bXXX binary literal syntax is not standard
3045 323035 Unhandled instruction: LDRSHT(Thumb)
3046 323036 Unhandled instruction: SMMLS (ARM and Thumb)
3047 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
3048 323175 Unhandled instruction: SMLALD (ARM + Thumb)
3049 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
3050 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
3051 323437 Phase 2 support for IBM Power ISA 2.07
3052 323713 Support mmxext (integer sse) subset on i386 (athlon)
3053 323803 Transactional memory instructions are not supported for Power
3054 323893 SSE3 not available on amd cpus in valgrind
3055 323905 Probable false positive from Valgrind/drd on close()
3056 323912 valgrind.h header isn't compatible for mingw64
3057 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
3058 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
3059 324181 mmap does not handle MAP_32BIT
3060 324227 memcheck false positive leak when a thread calls exit+block [..]
3061 324421 Support for fanotify API on ARM architecture
3062 324514 gdbserver monitor cmd output behaviour consistency [..]
3063 324518 ppc64: Emulation of dcbt instructions does not handle [..]
3064 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
3065 324582 When access is made to freed memory, report both allocation [..]
3066 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
3067 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
3068 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
3069 324834 Unhandled instructions in Microsoft C run-time for x86_64
3070 324894 Phase 3 support for IBM Power ISA 2.07
3071 326091 drd: Avoid false race reports from optimized strlen() impls
3072 326113 valgrind libvex hwcaps error on AMD64
3073 n-i-bz Some wrong command line options could be ignored
3074 n-i-bz patch to allow fair-sched on android
3075 n-i-bz report error for vgdb snapshot requested before execution
3076 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
3078 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
3082 Release 3.8.1 (19 September 2012)
3083 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3084 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
3085 that occur moderately frequently in real use cases, adds support for
3086 some missing instructions on ARM, and fixes a deadlock condition on
3087 MacOSX. If you package or deliver 3.8.0 for others to use, you might
3088 want to consider upgrading to 3.8.1 instead.
3090 The following bugs have been fixed or resolved. Note that "n-i-bz"
3091 stands for "not in bugzilla" -- that is, a bug that was reported to us
3092 but never got a bugzilla entry. We encourage you to file bugs in
3093 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3094 than mailing the developers (or mailing lists) directly -- bugs that
3095 are not entered into bugzilla tend to get forgotten about or ignored.
3097 To see details of a given bug, visit
3098 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3099 where XXXXXX is the bug number as listed below.
3102 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
3103 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
3104 298281 wcslen causes false(?) uninitialised value warnings
3105 301281 valgrind hangs on OS X when the process calls system()
3106 304035 disInstr(arm): unhandled instruction 0xE1023053
3107 304867 implement MOVBE instruction in x86 mode
3108 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
3109 305042 amd64: implement 0F 7F encoding of movq between two registers
3110 305199 ARM: implement QDADD and QDSUB
3111 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
3112 305513 killed by fatal signal: SIGSEGV
3113 305690 DRD reporting invalid semaphore when sem_trywait fails
3114 305926 Invalid alignment checks for some AVX instructions
3115 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
3116 306310 3.8.0 release tarball missing some files
3117 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
3118 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
3119 n-i-bz shmat of a segment > 4Gb does not work
3120 n-i-bz simulate_control_c script wrong USR1 signal number on mips
3121 n-i-bz vgdb ptrace calls wrong on mips [...]
3122 n-i-bz Fixes for more MPI false positives
3123 n-i-bz exp-sgcheck's memcpy causes programs to segfault
3124 n-i-bz OSX build w/ clang: asserts at startup
3125 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
3126 n-i-bz fix a couple of union tag-vs-field mixups
3127 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
3129 The following bugs were fixed in 3.8.0 but not listed in this NEWS
3132 254088 Valgrind should know about UD2 instruction
3135 304754 NEWS blows TeX's little mind
3137 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
3141 Release 3.8.0 (10 August 2012)
3142 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3143 3.8.0 is a feature release with many improvements and the usual
3144 collection of bug fixes.
3146 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3147 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
3148 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
3149 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
3150 There is initial support for MacOSX 10.8, but it is not usable for
3151 serious work at present.
3153 * ================== PLATFORM CHANGES =================
3155 * Support for MIPS32 platforms running Linux. Valgrind has been
3156 tested on MIPS32 and MIPS32r2 platforms running different Debian
3157 Squeeze and MeeGo distributions. Both little-endian and big-endian
3158 cores are supported. The tools Memcheck, Massif and Lackey have
3159 been tested and are known to work. See README.mips for more details.
3161 * Preliminary support for Android running on x86.
3163 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
3165 * Support for Intel AVX instructions and for AES instructions. This
3166 support is available only for 64 bit code.
3168 * Support for POWER Decimal Floating Point instructions.
3170 * ==================== TOOL CHANGES ====================
3172 * Non-libc malloc implementations are now supported. This is useful
3173 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
3174 Using the new option --soname-synonyms, such tools can be informed
3175 that the malloc implementation is either linked statically into the
3176 executable, or is present in some other shared library different
3177 from libc.so. This makes it possible to process statically linked
3178 programs, and programs using other malloc libraries, for example
3179 TCMalloc or JEMalloc.
3181 * For tools that provide their own replacement for malloc et al, the
3182 option --redzone-size=<number> allows users to specify the size of
3183 the padding blocks (redzones) added before and after each client
3184 allocated block. Smaller redzones decrease the memory needed by
3185 Valgrind. Bigger redzones increase the chance to detect blocks
3186 overrun or underrun. Prior to this change, the redzone size was
3187 hardwired to 16 bytes in Memcheck.
3191 - The leak_check GDB server monitor command now can
3192 control the maximum nr of loss records to output.
3194 - Reduction of memory use for applications allocating
3195 many blocks and/or having many partially defined bytes.
3197 - Addition of GDB server monitor command 'block_list' that lists
3198 the addresses/sizes of the blocks of a leak search loss record.
3200 - Addition of GDB server monitor command 'who_points_at' that lists
3201 the locations pointing at a block.
3203 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
3204 detect an invalid access of these redzones, by marking them
3205 noaccess. Similarly, if a redzone size is given for a memory
3206 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
3207 This still allows to find some bugs if the user has forgotten to
3208 mark the pool superblock noaccess.
3210 - Performance of memory leak check has been improved, especially in
3211 cases where there are many leaked blocks and/or many suppression
3212 rules used to suppress leak reports.
3214 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
3215 more precise analysis, which is important for LLVM/Clang
3216 generated code. This is at the cost of somewhat reduced
3217 performance. Note there is no change to analysis precision or
3218 costs on Linux targets.
3222 - Added even more facilities that can help finding the cause of a data
3223 race, namely the command-line option --ptrace-addr and the macro
3224 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
3226 - Fixed a subtle bug that could cause false positive data race reports.
3228 * ==================== OTHER CHANGES ====================
3230 * The C++ demangler has been updated so as to work well with C++
3231 compiled by up to at least g++ 4.6.
3233 * Tool developers can make replacement/wrapping more flexible thanks
3234 to the new option --soname-synonyms. This was reported above, but
3235 in fact is very general and applies to all function
3236 replacement/wrapping, not just to malloc-family functions.
3238 * Round-robin scheduling of threads can be selected, using the new
3239 option --fair-sched= yes. Prior to this change, the pipe-based
3240 thread serialisation mechanism (which is still the default) could
3241 give very unfair scheduling. --fair-sched=yes improves
3242 responsiveness of interactive multithreaded applications, and
3243 improves repeatability of results from the thread checkers Helgrind
3246 * For tool developers: support to run Valgrind on Valgrind has been
3247 improved. We can now routinely Valgrind on Helgrind or Memcheck.
3249 * gdbserver now shows the float shadow registers as integer
3250 rather than float values, as the shadow values are mostly
3251 used as bit patterns.
3253 * Increased limit for the --num-callers command line flag to 500.
3255 * Performance improvements for error matching when there are many
3256 suppression records in use.
3258 * Improved support for DWARF4 debugging information (bug 284184).
3260 * Initial support for DWZ compressed Dwarf debug info.
3262 * Improved control over the IR optimiser's handling of the tradeoff
3263 between performance and precision of exceptions. Specifically,
3264 --vex-iropt-precise-memory-exns has been removed and replaced by
3265 --vex-iropt-register-updates, with extended functionality. This
3266 allows the Valgrind gdbserver to always show up to date register
3269 * Modest performance gains through the use of translation chaining for
3272 * ==================== FIXED BUGS ====================
3274 The following bugs have been fixed or resolved. Note that "n-i-bz"
3275 stands for "not in bugzilla" -- that is, a bug that was reported to us
3276 but never got a bugzilla entry. We encourage you to file bugs in
3277 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3278 than mailing the developers (or mailing lists) directly -- bugs that
3279 are not entered into bugzilla tend to get forgotten about or ignored.
3281 To see details of a given bug, visit
3282 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3283 where XXXXXX is the bug number as listed below.
3285 197914 Building valgrind from svn now requires automake-1.10
3286 203877 increase to 16Mb maximum allowed alignment for memalign et al
3287 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
3288 247386 make perf does not run all performance tests
3289 270006 Valgrind scheduler unfair
3290 270777 Adding MIPS/Linux port to Valgrind
3291 270796 s390x: Removed broken support for the TS insn
3292 271438 Fix configure for proper SSE4.2 detection
3293 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
3294 273475 Add support for AVX instructions
3295 274078 improved configure logic for mpicc
3296 276993 fix mremap 'no thrash checks'
3297 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
3298 281482 memcheck incorrect byte allocation count in realloc() for silly argument
3299 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
3300 283413 Fix wrong sanity check
3301 283671 Robustize alignment computation in LibVEX_Alloc
3302 283961 Adding support for some HCI IOCTLs
3303 284124 parse_type_DIE: confused by: DWARF 4
3304 284864 == 273475 (Add support for AVX instructions)
3305 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
3306 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
3307 285725 == 273475 (Add support for AVX instructions)
3308 286261 add wrapper for linux I2C_RDWR ioctl
3309 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
3310 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
3311 286384 configure fails "checking for a supported version of gcc"
3312 286497 == 273475 (Add support for AVX instructions)
3313 286596 == 273475 (Add support for AVX instructions)
3314 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
3315 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
3316 287260 Incorrect conditional jump or move depends on uninitialised value(s)
3317 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
3318 287307 == 273475 (Add support for AVX instructions)
3319 287858 VG_(strerror): unknown error
3320 288298 (MacOSX) unhandled syscall shm_unlink
3321 288995 == 273475 (Add support for AVX instructions)
3322 289470 Loading of large Mach-O thin binaries fails.
3323 289656 == 273475 (Add support for AVX instructions)
3324 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
3325 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
3326 289839 s390x: Provide support for unicode conversion instructions
3327 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
3328 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
3329 290655 Add support for AESKEYGENASSIST instruction
3330 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
3331 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
3332 291253 ES register not initialised in valgrind simulation
3333 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
3334 291865 s390x: Support the "Compare Double and Swap" family of instructions
3335 292300 == 273475 (Add support for AVX instructions)
3336 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
3337 292493 == 273475 (Add support for AVX instructions)
3338 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
3339 292627 Missing support for some SCSI ioctls
3340 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
3341 292841 == 273475 (Add support for AVX instructions)
3342 292993 implement the getcpu syscall on amd64-linux
3343 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
3344 293088 Add some VEX sanity checks for ppc64 unhandled instructions
3345 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
3346 293754 PCMPxSTRx not implemented for 16-bit characters
3347 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
3348 293808 CLFLUSH not supported by latest VEX for amd64
3349 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
3350 294048 MPSADBW instruction not implemented
3351 294055 regtest none/tests/shell fails when locale is not set to C
3352 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
3353 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
3354 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
3355 294260 disInstr_AMD64: disInstr miscalculated next %rip
3356 294523 --partial-loads-ok=yes causes false negatives
3357 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
3358 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
3359 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
3360 295089 can not annotate source for both helgrind and drd
3361 295221 POWER Processor decimal floating point instruction support missing
3362 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
3363 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
3364 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
3365 295617 ARM - Add some missing syscalls
3366 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
3367 296229 Linux user input device ioctls missing wrappers
3368 296318 ELF Debug info improvements (more than one rx/rw mapping)
3369 296422 Add translation chaining support
3370 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
3371 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
3372 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
3373 297078 gdbserver signal handling problems [..]
3374 297147 drd false positives on newly allocated memory
3375 297329 disallow decoding of IBM Power DFP insns on some machines
3376 297497 POWER Processor decimal floating point instruction support missing
3377 297701 Another alias for strncasecmp_l in libc-2.13.so
3378 297911 'invalid write' not reported when using APIs for custom mem allocators.
3379 297976 s390x: revisit EX implementation
3380 297991 Valgrind interferes with mmap()+ftell()
3381 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
3382 297993 Fix compilation of valgrind with gcc -g3.
3383 298080 POWER Processor DFP support missing, part 3
3384 298227 == 273475 (Add support for AVX instructions)
3385 298335 == 273475 (Add support for AVX instructions)
3386 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
3387 298394 s390x: Don't bail out on an unknown machine model. [..]
3388 298421 accept4() syscall (366) support is missing for ARM
3389 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
3390 298732 valgrind installation problem in ubuntu with kernel version 3.x
3391 298862 POWER Processor DFP instruction support missing, part 4
3392 298864 DWARF reader mis-parses DW_FORM_ref_addr
3393 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
3394 299053 Support DWARF4 DW_AT_high_pc constant form
3395 299104 == 273475 (Add support for AVX instructions)
3396 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
3397 299629 dup3() syscall (358) support is missing for ARM
3398 299694 POWER Processor DFP instruction support missing, part 5
3399 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
3400 299803 == 273475 (Add support for AVX instructions)
3401 299804 == 273475 (Add support for AVX instructions)
3402 299805 == 273475 (Add support for AVX instructions)
3403 300140 ARM - Missing (T1) SMMUL
3404 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
3405 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
3406 300414 FCOM and FCOMP unimplemented for amd64 guest
3407 301204 infinite loop in canonicaliseSymtab with ifunc symbol
3408 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
3409 301265 add x86 support to Android build
3410 301984 configure script doesn't detect certain versions of clang
3411 302205 Fix compiler warnings for POWER VEX code and POWER test cases
3412 302287 Unhandled movbe instruction on Atom processors
3413 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
3414 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
3415 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
3416 302656 == 273475 (Add support for AVX instructions)
3417 302709 valgrind for ARM needs extra tls support for android emulator [..]
3418 302827 add wrapper for CDROM_GET_CAPABILITY
3419 302901 Valgrind crashes with dwz optimized debuginfo
3420 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
3421 303116 Add support for the POWER instruction popcntb
3422 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
3423 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
3424 303466 == 273475 (Add support for AVX instructions)
3425 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
3426 303963 strstr() function produces wrong results under valgrind callgrind
3427 304054 CALL_FN_xx macros need to enforce stack alignment
3428 304561 tee system call not supported
3429 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
3430 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
3431 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
3432 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
3433 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
3434 n-i-bz s390x: Shadow registers can now be examined using vgdb
3436 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
3437 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
3441 Release 3.7.0 (5 November 2011)
3442 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3443 3.7.0 is a feature release with many significant improvements and the
3444 usual collection of bug fixes.
3446 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3447 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
3448 Support for recent distros and toolchain components (glibc 2.14, gcc
3449 4.6, MacOSX 10.7) has been added.
3451 * ================== PLATFORM CHANGES =================
3453 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
3454 analyse 64-bit programs running on z/Architecture. Most user space
3455 instructions up to and including z10 are supported. Valgrind has
3456 been tested extensively on z9, z10, and z196 machines running SLES
3457 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
3458 known to work well. Callgrind, Helgrind, and DRD work reasonably
3459 well on z9 and later models. See README.s390 for more details.
3461 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
3462 64-bit processes are supported. Some complex threaded applications
3463 (Firefox) are observed to hang when run as 32 bit applications,
3464 whereas 64-bit versions run OK. The cause is unknown. Memcheck
3465 will likely report some false errors. In general, expect some rough
3466 spots. This release also supports MacOSX 10.6, but drops support
3469 * Preliminary support for Android (on ARM). Valgrind can now run
3470 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
3471 README.android for more details, plus instructions on how to get
3474 * Support for the IBM Power ISA 2.06 (Power7 instructions)
3476 * General correctness and performance improvements for ARM/Linux, and,
3477 by extension, ARM/Android.
3479 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
3480 instruction set support is under development but is not available in
3483 * Support for AIX5 has been removed.
3485 * ==================== TOOL CHANGES ====================
3487 * Memcheck: some incremental changes:
3489 - reduction of memory use in some circumstances
3491 - improved handling of freed memory, which in some circumstances
3492 can cause detection of use-after-free that would previously have
3495 - fix of a longstanding bug that could cause false negatives (missed
3496 errors) in programs doing vector saturated narrowing instructions.
3498 * Helgrind: performance improvements and major memory use reductions,
3499 particularly for large, long running applications which perform many
3500 synchronisation (lock, unlock, etc) events. Plus many smaller
3503 - display of locksets for both threads involved in a race
3505 - general improvements in formatting/clarity of error messages
3507 - addition of facilities and documentation regarding annotation
3508 of thread safe reference counted C++ classes
3510 - new flag --check-stack-refs=no|yes [yes], to disable race checking
3511 on thread stacks (a performance hack)
3513 - new flag --free-is-write=no|yes [no], to enable detection of races
3514 where one thread accesses heap memory but another one frees it,
3515 without any coordinating synchronisation event
3517 * DRD: enabled XML output; added support for delayed thread deletion
3518 in order to detect races that occur close to the end of a thread
3519 (--join-list-vol); fixed a memory leak triggered by repeated client
3520 memory allocatation and deallocation; improved Darwin support.
3522 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
3524 * exp-sgcheck: this tool has been reduced in scope so as to improve
3525 performance and remove checking that Memcheck does better.
3526 Specifically, the ability to check for overruns for stack and global
3527 arrays is unchanged, but the ability to check for overruns of heap
3528 blocks has been removed. The tool has accordingly been renamed to
3529 exp-sgcheck ("Stack and Global Array Checking").
3531 * ==================== OTHER CHANGES ====================
3533 * GDB server: Valgrind now has an embedded GDB server. That means it
3534 is possible to control a Valgrind run from GDB, doing all the usual
3535 things that GDB can do (single stepping, breakpoints, examining
3536 data, etc). Tool-specific functionality is also available. For
3537 example, it is possible to query the definedness state of variables
3538 or memory from within GDB when running Memcheck; arbitrarily large
3539 memory watchpoints are supported, etc. To use the GDB server, start
3540 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3543 * Improved support for unfriendly self-modifying code: a new option
3544 --smc-check=all-non-file is available. This adds the relevant
3545 consistency checks only to code that originates in non-file-backed
3546 mappings. In effect this confines the consistency checking only to
3547 code that is or might be JIT generated, and avoids checks on code
3548 that must have been compiled ahead of time. This significantly
3549 improves performance on applications that generate code at run time.
3551 * It is now possible to build a working Valgrind using Clang-2.9 on
3554 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3555 These enable and disable error reporting on a per-thread, and
3556 nestable, basis. This is useful for hiding errors in particularly
3557 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3558 now uses this facility.
3560 * Added the --mod-funcname option to cg_diff.
3562 * ==================== FIXED BUGS ====================
3564 The following bugs have been fixed or resolved. Note that "n-i-bz"
3565 stands for "not in bugzilla" -- that is, a bug that was reported to us
3566 but never got a bugzilla entry. We encourage you to file bugs in
3567 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3568 mailing the developers (or mailing lists) directly -- bugs that are
3569 not entered into bugzilla tend to get forgotten about or ignored.
3571 To see details of a given bug, visit
3572 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3573 where XXXXXX is the bug number as listed below.
3575 79311 malloc silly arg warning does not give stack trace
3576 210935 port valgrind.h (not valgrind) to win32 to support client requests
3577 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3578 243404 Port to zSeries
3579 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3580 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3581 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3582 253206 Some fixes for the faultstatus testcase
3583 255223 capget testcase fails when running as root
3584 256703 xlc_dbl_u32.c testcase broken
3585 256726 Helgrind tests have broken inline asm
3586 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3587 264800 testcase compile failure on zseries
3588 265762 make public VEX headers compilable by G++ 3.x
3589 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3590 266753 configure script does not give the user the option to not use QtCore
3591 266931 gen_insn_test.pl is broken
3592 266961 ld-linux.so.2 i?86-linux strlen issues
3593 266990 setns instruction causes false positive
3594 267020 Make directory for temporary files configurable at run-time.
3595 267342 == 267997 (segmentation fault on Mac OS 10.6)
3596 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3597 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3598 267488 regtest: darwin support for 64-bit build
3599 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3600 267630 Add support for IBM Power ISA 2.06 -- stage 1
3601 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3602 267819 Add client request for informing the core about reallocation
3603 267925 laog data structure quadratic for a single sequence of lock
3604 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3605 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3606 268513 missed optimizations in fold_Expr
3607 268619 s390x: fpr - gpr transfer facility
3608 268620 s390x: reconsider "long displacement" requirement
3609 268621 s390x: improve IR generation for XC
3610 268715 s390x: FLOGR is not universally available
3611 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3612 268930 s390x: MHY is not universally available
3613 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3614 269079 Support ptrace system call on ARM
3615 269144 missing "Bad option" error message
3616 269209 conditional load and store facility (z196)
3617 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3618 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3619 269736 s390x: minor code generation tweaks
3620 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3621 269863 s390x: remove unused function parameters
3622 269864 s390x: tweak s390_emit_load_cc
3623 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3624 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3625 270115 s390x: rewrite some testcases
3626 270309 == 267997 (valgrind crash on startup)
3627 270320 add support for Linux FIOQSIZE ioctl() call
3628 270326 segfault while trying to sanitize the environment passed to execle
3629 270794 IBM POWER7 support patch causes regression in none/tests
3630 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3631 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3632 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3633 270959 s390x: invalid use of R0 as base register
3634 271042 VSX configure check fails when it should not
3635 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3636 271259 s390x: fix code confusion
3637 271337 == 267997 (Valgrind segfaults on MacOS X)
3638 271385 s390x: Implement Ist_MBE
3639 271501 s390x: misc cleanups
3640 271504 s390x: promote likely and unlikely
3641 271579 ppc: using wrong enum type
3642 271615 unhandled instruction "popcnt" (arch=amd10h)
3643 271730 Fix bug when checking ioctls: duplicate check
3644 271776 s390x: provide STFLE instruction support
3645 271779 s390x: provide clock instructions like STCK
3646 271799 Darwin: ioctls without an arg report a memory error
3647 271820 arm: fix type confusion
3648 271917 pthread_cond_timedwait failure leads to not-locked false positive
3649 272067 s390x: fix DISP20 macro
3650 272615 A typo in debug output in mc_leakcheck.c
3651 272661 callgrind_annotate chokes when run from paths containing regex chars
3652 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3653 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3654 272967 make documentation build-system more robust
3655 272986 Fix gcc-4.6 warnings with valgrind.h
3656 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3657 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3658 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3659 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3660 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3661 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3662 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3663 273778 exp-ptrcheck: unhandled sysno == 259
3664 274089 exp-ptrcheck: unhandled sysno == 208
3665 274378 s390x: Various dispatcher tweaks
3666 274447 WARNING: unhandled syscall: 340
3667 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3668 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3669 274926 valgrind does not build against linux-3
3670 275148 configure FAIL with glibc-2.14
3671 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3672 275168 Make Valgrind work for MacOSX 10.7 Lion
3673 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3674 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3675 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3676 275308 Fix implementation for ppc64 fres instruc
3677 275339 s390x: fix testcase compile warnings
3678 275517 s390x: Provide support for CKSM instruction
3679 275710 s390x: get rid of redundant address mode calculation
3680 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3681 275852 == 250101 (valgrind uses all swap space and is killed)
3682 276784 Add support for IBM Power ISA 2.06 -- stage 3
3683 276987 gdbsrv: fix tests following recent commits
3684 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3685 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3686 277471 Unhandled syscall: 340
3687 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3688 277653 ARM: support Thumb2 PLD instruction
3689 277663 ARM: NEON float VMUL by scalar incorrect
3690 277689 ARM: tests for VSTn with register post-index are broken
3691 277694 ARM: BLX LR instruction broken in ARM mode
3692 277780 ARM: VMOV.F32 (immediate) instruction is broken
3693 278057 fuse filesystem syscall deadlocks
3694 278078 Unimplemented syscall 280 on ppc32
3695 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3696 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3697 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3698 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3699 279027 Support for MVCL and CLCL instruction
3700 279027 s390x: Provide support for CLCL and MVCL instructions
3701 279062 Remove a redundant check in the insn selector for ppc.
3702 279071 JDK creates PTEST with redundant REX.W prefix
3703 279212 gdbsrv: add monitor cmd v.info scheduler.
3704 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3705 279698 memcheck discards valid-bits for packuswb
3706 279795 memcheck reports uninitialised values for mincore on amd64
3707 279994 Add support for IBM Power ISA 2.06 -- stage 3
3708 280083 mempolicy syscall check errors
3709 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3710 280710 s390x: config files for nightly builds
3711 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3712 280965 Valgrind breaks fcntl locks when program does mmap
3713 281138 WARNING: unhandled syscall: 340
3714 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3715 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3716 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3717 281468 s390x: handle do_clone and gcc clones in call traces
3718 281488 ARM: VFP register corruption
3719 281828 == 275284 (false memmove warning: "Source and destination overlap")
3720 281883 s390x: Fix system call wrapper for "clone".
3721 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3722 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3723 282238 SLES10: make check fails
3724 282979 strcasestr needs replacement with recent(>=2.12) glibc
3725 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3726 283243 Regression in ppc64 memcheck tests
3727 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3728 283427 re-connect epoll_pwait syscall on ARM linux
3729 283600 gdbsrv: android: port vgdb.c
3730 283709 none/tests/faultstatus needs to account for page size
3731 284305 filter_gdb needs enhancement to work on ppc64
3732 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3733 284472 Thumb2 ROR.W encoding T2 not implemented
3734 284621 XML-escape process command line in XML output
3735 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3736 that have non-power-of-2 sizes (also AMDs)
3737 n-i-bz don't be spooked by libraries mashed by elfhack
3738 n-i-bz don't be spooked by libxul.so linked with gold
3739 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3741 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3742 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3743 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3747 Release 3.6.1 (16 February 2011)
3748 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3749 3.6.1 is a bug fix release. It adds support for some SSE4
3750 instructions that were omitted in 3.6.0 due to lack of time. Initial
3751 support for glibc-2.13 has been added. A number of bugs causing
3752 crashing or assertion failures have been fixed.
3754 The following bugs have been fixed or resolved. Note that "n-i-bz"
3755 stands for "not in bugzilla" -- that is, a bug that was reported to us
3756 but never got a bugzilla entry. We encourage you to file bugs in
3757 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3758 mailing the developers (or mailing lists) directly -- bugs that are
3759 not entered into bugzilla tend to get forgotten about or ignored.
3761 To see details of a given bug, visit
3762 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3763 where XXXXXX is the bug number as listed below.
3765 188572 Valgrind on Mac should suppress setenv() mem leak
3766 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3767 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3768 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3769 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3770 254420 memory pool tracking broken
3771 254957 Test code failing to compile due to changes in memcheck.h
3772 255009 helgrind/drd: crash on chmod with invalid parameter
3773 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3774 255355 helgrind/drd: crash on threaded programs doing fork
3776 255418 (SSE4.x) rint call compiled with ICC
3777 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3778 255888 closing valgrindoutput tag outputted to log-stream on error
3779 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3780 255966 Slowness when using mempool annotations
3781 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3782 256600 super-optimized strcasecmp() false positive
3783 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3784 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3785 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3786 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3787 257276 Missing case in memcheck --track-origins=yes
3788 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3789 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3790 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3791 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3792 263099 callgrind_annotate counts Ir improperly [...]
3793 263877 undefined coprocessor instruction on ARMv7
3794 265964 configure FAIL with glibc-2.13
3795 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3796 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3797 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3798 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3799 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3801 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3805 Release 3.6.0 (21 October 2010)
3806 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3807 3.6.0 is a feature release with many significant improvements and the
3808 usual collection of bug fixes.
3810 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3811 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3812 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3814 -------------------------
3816 Here are some highlights. Details are shown further down:
3818 * Support for ARM/Linux.
3820 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3822 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3824 * Support for the SSE4.2 instruction set.
3826 * Enhancements to the Callgrind profiler, including the ability to
3827 handle CPUs with three levels of cache.
3829 * A new experimental heap profiler, DHAT.
3831 * A huge number of bug fixes and small enhancements.
3833 -------------------------
3835 Here are details of the above changes, together with descriptions of
3836 many other changes, and a list of fixed bugs.
3838 * ================== PLATFORM CHANGES =================
3840 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3841 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3842 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3844 This requires a CPU capable of running the ARMv7-A instruction set
3845 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3846 of the user space instruction set, including ARM and Thumb integer
3847 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3848 Cachegrind and Massif tools work properly; other tools work to
3851 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3852 with support for recent releases of the underlying toolchain
3853 components, notably gcc-4.5 and glibc-2.12.
3855 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3856 support also works much better on OS X 10.5, and is as solid as
3859 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3860 64-bit mode. In 32-bit mode, support is only available up to and
3861 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3862 supported in 64-bit mode, and 32-bit mode does in fact support the
3863 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3864 10.6 on 32-bit targets.
3866 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3867 and including version 2.05 is supported.
3869 * ==================== TOOL CHANGES ====================
3871 * Cachegrind has a new processing script, cg_diff, which finds the
3872 difference between two profiles. It's very useful for evaluating
3873 the performance effects of a change in a program.
3875 Related to this change, the meaning of cg_annotate's (rarely-used)
3876 --threshold option has changed; this is unlikely to affect many
3877 people, if you do use it please see the user manual for details.
3879 * Callgrind now can do branch prediction simulation, similar to
3880 Cachegrind. In addition, it optionally can count the number of
3881 executed global bus events. Both can be used for a better
3882 approximation of a "Cycle Estimation" as derived event (you need to
3883 update the event formula in KCachegrind yourself).
3885 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3886 rather than the L2 cache. This is to accommodate machines with
3887 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3888 cache configuration of such a machine it will run the simulation as
3889 if the L2 cache isn't present. This means the results are less
3890 likely to match the true result for the machine, but
3891 Cachegrind/Callgrind's results are already only approximate, and
3892 should not be considered authoritative. The results are still
3893 useful for giving a general idea about a program's locality.
3895 * Massif has a new option, --pages-as-heap, which is disabled by
3896 default. When enabled, instead of tracking allocations at the level
3897 of heap blocks (as allocated with malloc/new/new[]), it instead
3898 tracks memory allocations at the level of memory pages (as mapped by
3899 mmap, brk, etc). Each mapped page is treated as its own block.
3900 Interpreting the page-level output is harder than the heap-level
3901 output, but this option is useful if you want to account for every
3902 byte of memory used by a program.
3904 * DRD has two new command-line options: --free-is-write and
3905 --trace-alloc. The former allows to detect reading from already freed
3906 memory, and the latter allows tracing of all memory allocations and
3909 * DRD has several new annotations. Custom barrier implementations can
3910 now be annotated, as well as benign races on static variables.
3912 * DRD's happens before / happens after annotations have been made more
3913 powerful, so that they can now also be used to annotate e.g. a smart
3914 pointer implementation.
3916 * Helgrind's annotation set has also been drastically improved, so as
3917 to provide to users a general set of annotations to describe locks,
3918 semaphores, barriers and condition variables. Annotations to
3919 describe thread-safe reference counted heap objects have also been
3922 * Memcheck has a new command-line option, --show-possibly-lost, which
3923 is enabled by default. When disabled, the leak detector will not
3924 show possibly-lost blocks.
3926 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3927 has been added. DHAT keeps track of allocated heap blocks, and also
3928 inspects every memory reference to see which block (if any) is being
3929 accessed. This gives a lot of insight into block lifetimes,
3930 utilisation, turnover, liveness, and the location of hot and cold
3931 fields. You can use DHAT to do hot-field profiling.
3933 * ==================== OTHER CHANGES ====================
3935 * Improved support for unfriendly self-modifying code: the extra
3936 overhead incurred by --smc-check=all has been reduced by
3937 approximately a factor of 5 as compared with 3.5.0.
3939 * Ability to show directory names for source files in error messages.
3940 This is combined with a flexible mechanism for specifying which
3941 parts of the paths should be shown. This is enabled by the new flag
3944 * A new flag, --require-text-symbol, which will stop the run if a
3945 specified symbol is not found it a given shared object when it is
3946 loaded into the process. This makes advanced working with function
3947 intercepting and wrapping safer and more reliable.
3949 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3950 and control of Valgrind is now available for the tools Memcheck and
3951 Helgrind. XML output from Valgrind is available for Memcheck,
3952 Helgrind and exp-Ptrcheck.
3954 * More reliable stack unwinding on amd64-linux, particularly in the
3955 presence of function wrappers, and with gcc-4.5 compiled code.
3957 * Modest scalability (performance improvements) for massive
3958 long-running applications, particularly for those with huge amounts
3961 * Support for analyzing programs running under Wine with has been
3962 improved. The header files <valgrind/valgrind.h>,
3963 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3964 Windows-programs compiled with MinGW or one of the Microsoft Visual
3967 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3968 The 32-bit simulator was not affected. This did not occur often,
3969 but when it did would usually crash the program under test.
3972 * A large number of bugs were fixed. These are shown below.
3974 * A number of bugs were investigated, and were candidates for fixing,
3975 but are not fixed in 3.6.0, due to lack of developer time. They may
3976 get fixed in later releases. They are:
3978 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3979 212419 false positive "lock order violated" (A+B vs A)
3980 213685 Undefined value propagates past dependency breaking instruction
3981 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3982 237920 valgrind segfault on fork failure
3983 242137 support for code compiled by LLVM-2.8
3984 242423 Another unknown Intel cache config value
3985 243232 Inconsistent Lock Orderings report with trylock
3986 243483 ppc: callgrind triggers VEX assertion failure
3987 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3988 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3990 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3991 249435 Analyzing wine programs with callgrind triggers a crash
3992 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3993 250065 Handling large allocations
3994 250101 huge "free" memory usage due to m_mallocfree.c
3995 "superblocks fragmentation"
3996 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3997 252091 Callgrind on ARM does not detect function returns correctly
3998 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3999 254420 memory pool tracking broken
4000 n-i-bz support for adding symbols for JIT generated code
4003 The following bugs have been fixed or resolved. Note that "n-i-bz"
4004 stands for "not in bugzilla" -- that is, a bug that was reported to us
4005 but never got a bugzilla entry. We encourage you to file bugs in
4006 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4007 mailing the developers (or mailing lists) directly -- bugs that are
4008 not entered into bugzilla tend to get forgotten about or ignored.
4010 To see details of a given bug, visit
4011 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4012 where XXXXXX is the bug number as listed below.
4014 135264 dcbzl instruction missing
4016 153699 Valgrind should report unaligned reads with movdqa
4018 190429 Valgrind reports lost of errors in ld.so
4019 with x86_64 2.9.90 glibc
4020 197266 valgrind appears to choke on the xmms instruction
4022 197988 Crash when demangling very large symbol names
4023 202315 unhandled syscall: 332 (inotify_init1)
4024 203256 Add page-level profiling to Massif
4025 205093 dsymutil=yes needs quotes, locking (partial fix)
4026 205241 Snow Leopard 10.6 support (partial fix)
4027 206600 Leak checker fails to upgrade indirect blocks when their
4028 parent becomes reachable
4029 210935 port valgrind.h (not valgrind) to win32 so apps run under
4030 wine can make client requests
4031 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
4032 within Linux ip-stack checksum functions
4033 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
4035 213685 Undefined value propagates past dependency breaking instruction
4037 215914 Valgrind inserts bogus empty environment variable
4039 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
4040 222545 shmat fails under valgind on some arm targets
4041 222560 ARM NEON support
4044 232509 Docs build fails with formatting inside <title></title> elements
4046 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
4047 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
4048 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
4049 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
4050 237485 symlink (syscall 57) is not supported on Mac OS
4051 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
4053 238208 is_just_below_ESP doesn't take into account red-zone
4054 238345 valgrind passes wrong $0 when executing a shell script
4055 238679 mq_timedreceive syscall doesn't flag the reception buffer
4057 238696 fcntl command F_DUPFD_CLOEXEC not supported
4058 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4059 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
4060 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
4062 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
4068 242606 unhandled syscall: setegid (in Ptrcheck)
4069 242814 Helgrind "Impossible has happened" during
4070 QApplication::initInstance();
4071 243064 Valgrind attempting to read debug information from iso
4072 243270 Make stack unwinding in Valgrind wrappers more reliable
4073 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
4074 sysno = 277 (mq_open)
4075 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
4076 244493 ARM VFP d16-d31 registers support
4077 244670 add support for audit_session_self syscall on Mac OS 10.6
4078 244921 The xml report of helgrind tool is not well format
4079 244923 In the xml report file, the <preamble> not escape the
4080 xml char, eg '<','&','>'
4081 245535 print full path names in plain text reports
4082 245925 x86-64 red zone handling problem
4083 246258 Valgrind not catching integer underruns + new [] s
4084 246311 reg/reg cmpxchg doesn't work on amd64
4085 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
4086 246888 Improve Makefile.vex.am
4087 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
4088 to [f]chmod_extended
4089 247526 IBM POWER6 (ISA 2.05) support is incomplete
4090 247561 Some leak testcases fails due to reachable addresses in
4092 247875 sizeofIRType to handle Ity_I128
4093 247894 [PATCH] unhandled syscall sys_readahead
4094 247980 Doesn't honor CFLAGS passed to configure
4095 248373 darwin10.supp is empty in the trunk
4096 248822 Linux FIBMAP ioctl has int parameter instead of long
4097 248893 [PATCH] make readdwarf.c big endianess safe to enable
4098 unwinding on big endian systems
4099 249224 Syscall 336 not supported (SYS_proc_info)
4101 249775 Incorrect scheme for detecting NEON capabilities of host CPU
4102 249943 jni JVM init fails when using valgrind
4103 249991 Valgrind incorrectly declares AESKEYGENASSIST support
4105 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
4106 250799 frexp$fenv_access_off function generates SIGILL
4107 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
4108 251251 support pclmulqdq insn
4109 251362 valgrind: ARM: attach to debugger either fails or provokes
4111 251674 Unhandled syscall 294
4114 254257 Add support for debugfiles found by build-id
4115 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
4116 254646 Wrapped functions cause stack misalignment on OS X
4117 (and possibly Linux)
4118 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
4120 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
4124 Release 3.5.0 (19 August 2009)
4125 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4126 3.5.0 is a feature release with many significant improvements and the
4127 usual collection of bug fixes. The main improvement is that Valgrind
4128 now works on Mac OS X.
4130 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
4131 and X86/Darwin. Support for recent distros and toolchain components
4132 (glibc 2.10, gcc 4.5) has been added.
4134 -------------------------
4136 Here is a short summary of the changes. Details are shown further
4139 * Support for Mac OS X (10.5.x).
4141 * Improvements and simplifications to Memcheck's leak checker.
4143 * Clarification and simplifications in various aspects of Valgrind's
4146 * XML output for Helgrind and Ptrcheck.
4148 * Performance and stability improvements for Helgrind and DRD.
4150 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
4152 * A new experimental tool, BBV, useful for computer architecture
4155 * Improved Wine support, including ability to read Windows PDB
4158 -------------------------
4160 Here are details of the above changes, followed by descriptions of
4161 many other minor changes, and a list of fixed bugs.
4164 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
4165 called "Darwin" because that is the name of the OS core, which is the
4166 level that Valgrind works at.)
4170 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
4171 because it would require work and 10.4 is only becoming less common.
4173 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
4174 fairly well. For 10.5.x, 32-bit programs are the default even on
4175 64-bit machines, so it handles most current programs.
4177 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
4178 officially supported, but simple programs at least will probably work.
4179 However, start-up is slow.
4181 - PowerPC machines are not supported.
4183 Things that don't work:
4185 - The Ptrcheck tool.
4187 - Objective-C garbage collection.
4191 - If you have Rogue Amoeba's "Instant Hijack" program installed,
4192 Valgrind will fail with a SIGTRAP at start-up. See
4193 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
4198 - You will likely find --dsymutil=yes a useful option, as error
4199 messages may be imprecise without it.
4201 - Mac OS X support is new and therefore will be less robust than the
4202 Linux support. Please report any bugs you find.
4204 - Threaded programs may run more slowly than on Linux.
4206 Many thanks to Greg Parker for developing this port over several years.
4209 * Memcheck's leak checker has been improved.
4211 - The results for --leak-check=summary now match the summary results
4212 for --leak-check=full. Previously they could differ because
4213 --leak-check=summary counted "indirectly lost" blocks and
4214 "suppressed" blocks as "definitely lost".
4216 - Blocks that are only reachable via at least one interior-pointer,
4217 but are directly pointed to by a start-pointer, were previously
4218 marked as "still reachable". They are now correctly marked as
4221 - The default value for the --leak-resolution option has been
4222 changed from "low" to "high". In general, this means that more
4223 leak reports will be produced, but each leak report will describe
4224 fewer leaked blocks.
4226 - With --leak-check=full, "definitely lost" and "possibly lost"
4227 leaks are now considered as proper errors, ie. they are counted
4228 for the "ERROR SUMMARY" and affect the behaviour of
4229 --error-exitcode. These leaks are not counted as errors if
4230 --leak-check=summary is specified, however.
4232 - Documentation for the leak checker has been improved.
4235 * Various aspects of Valgrind's text output have changed.
4237 - Valgrind's start-up message has changed. It is shorter but also
4238 includes the command being run, which makes it easier to use
4239 --trace-children=yes. An example:
4241 - Valgrind's shut-down messages have also changed. This is most
4242 noticeable with Memcheck, where the leak summary now occurs before
4243 the error summary. This change was necessary to allow leaks to be
4244 counted as proper errors (see the description of the leak checker
4245 changes above for more details). This was also necessary to fix a
4246 longstanding bug in which uses of suppressions against leaks were
4247 not "counted", leading to difficulties in maintaining suppression
4248 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
4250 - Behavior of -v has changed. In previous versions, -v printed out
4251 a mixture of marginally-user-useful information, and tool/core
4252 statistics. The statistics printing has now been moved to its own
4253 flag, --stats=yes. This means -v is less verbose and more likely
4254 to convey useful end-user information.
4256 - The format of some (non-XML) stack trace entries has changed a
4257 little. Previously there were six possible forms:
4259 0x80483BF: really (a.c:20)
4260 0x80483BF: really (in /foo/a.out)
4262 0x80483BF: (within /foo/a.out)
4263 0x80483BF: ??? (a.c:20)
4266 The third and fourth of these forms have been made more consistent
4267 with the others. The six possible forms are now:
4269 0x80483BF: really (a.c:20)
4270 0x80483BF: really (in /foo/a.out)
4271 0x80483BF: really (in ???)
4272 0x80483BF: ??? (in /foo/a.out)
4273 0x80483BF: ??? (a.c:20)
4276 Stack traces produced when --xml=yes is specified are different
4280 * Helgrind and Ptrcheck now support XML output, so they can be used
4281 from GUI tools. Also, the XML output mechanism has been
4284 - The XML format has been overhauled and generalised, so it is more
4285 suitable for error reporting tools in general. The Memcheck
4286 specific aspects of it have been removed. The new format, which
4287 is an evolution of the old format, is described in
4288 docs/internals/xml-output-protocol4.txt.
4290 - Memcheck has been updated to use the new format.
4292 - Helgrind and Ptrcheck are now able to emit output in this format.
4294 - The XML output mechanism has been overhauled. XML is now output
4295 to its own file descriptor, which means that:
4297 * Valgrind can output text and XML independently.
4299 * The longstanding problem of XML output being corrupted by
4300 unexpected un-tagged text messages is solved.
4302 As before, the destination for text output is specified using
4303 --log-file=, --log-fd= or --log-socket=.
4305 As before, XML output for a tool is enabled using --xml=yes.
4307 Because there's a new XML output channel, the XML output
4308 destination is now specified by --xml-file=, --xml-fd= or
4311 Initial feedback has shown this causes some confusion. To
4312 clarify, the two envisaged usage scenarios are:
4314 (1) Normal text output. In this case, do not specify --xml=yes
4315 nor any of --xml-file=, --xml-fd= or --xml-socket=.
4317 (2) XML output. In this case, specify --xml=yes, and one of
4318 --xml-file=, --xml-fd= or --xml-socket= to select the XML
4319 destination, one of --log-file=, --log-fd= or --log-socket=
4320 to select the destination for any remaining text messages,
4321 and, importantly, -q.
4323 -q makes Valgrind completely silent on the text channel,
4324 except in the case of critical failures, such as Valgrind
4325 itself segfaulting, or failing to read debugging information.
4326 Hence, in this scenario, it suffices to check whether or not
4327 any output appeared on the text channel. If yes, then it is
4328 likely to be a critical error which should be brought to the
4329 attention of the user. If no (the text channel produced no
4330 output) then it can be assumed that the run was successful.
4332 This allows GUIs to make the critical distinction they need to
4333 make (did the run fail or not?) without having to search or
4334 filter the text output channel in any way.
4336 It is also recommended to use --child-silent-after-fork=yes in
4340 * Improvements and changes in Helgrind:
4342 - XML output, as described above
4344 - Checks for consistent association between pthread condition
4345 variables and their associated mutexes are now performed.
4347 - pthread_spinlock functions are supported.
4349 - Modest performance improvements.
4351 - Initial (skeletal) support for describing the behaviour of
4352 non-POSIX synchronisation objects through ThreadSanitizer
4353 compatible ANNOTATE_* macros.
4355 - More controllable tradeoffs between performance and the level of
4356 detail of "previous" accesses in a race. There are now three
4359 * --history-level=full. This is the default, and was also the
4360 default in 3.4.x. It shows both stacks involved in a race, but
4361 requires a lot of memory and can be very slow in programs that
4362 do many inter-thread synchronisation events.
4364 * --history-level=none. This only shows the later stack involved
4365 in a race. This can be much faster than --history-level=full,
4366 but makes it much more difficult to find the other access
4367 involved in the race.
4369 The new intermediate setting is
4371 * --history-level=approx
4373 For the earlier (other) access, two stacks are presented. The
4374 earlier access is guaranteed to be somewhere in between the two
4375 program points denoted by those stacks. This is not as useful
4376 as showing the exact stack for the previous access (as per
4377 --history-level=full), but it is better than nothing, and it's
4378 almost as fast as --history-level=none.
4381 * New features and improvements in DRD:
4383 - The error messages printed by DRD are now easier to interpret.
4384 Instead of using two different numbers to identify each thread
4385 (Valgrind thread ID and DRD thread ID), DRD does now identify
4386 threads via a single number (the DRD thread ID). Furthermore
4387 "first observed at" information is now printed for all error
4388 messages related to synchronization objects.
4390 - Added support for named semaphores (sem_open() and sem_close()).
4392 - Race conditions between pthread_barrier_wait() and
4393 pthread_barrier_destroy() calls are now reported.
4395 - Added support for custom allocators through the macros
4396 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
4397 in <valgrind/valgrind.h>). An alternative for these two macros is
4398 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
4401 - Added support for annotating non-POSIX synchronization objects
4402 through several new ANNOTATE_*() macros.
4404 - OpenMP: added support for the OpenMP runtime (libgomp) included
4405 with gcc versions 4.3.0 and 4.4.0.
4409 - Added two new command-line options (--first-race-only and
4410 --segment-merging-interval).
4413 * Genuinely atomic support for x86/amd64/ppc atomic instructions
4415 Valgrind will now preserve (memory-access) atomicity of LOCK-
4416 prefixed x86/amd64 instructions, and any others implying a global
4417 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
4419 This means that Valgrinded processes will "play nicely" in
4420 situations where communication with other processes, or the kernel,
4421 is done through shared memory and coordinated with such atomic
4422 instructions. Prior to this change, such arrangements usually
4423 resulted in hangs, races or other synchronisation failures, because
4424 Valgrind did not honour atomicity of such instructions.
4427 * A new experimental tool, BBV, has been added. BBV generates basic
4428 block vectors for use with the SimPoint analysis tool, which allows
4429 a program's overall behaviour to be approximated by running only a
4430 fraction of it. This is useful for computer architecture
4431 researchers. You can run BBV by specifying --tool=exp-bbv (the
4432 "exp-" prefix is short for "experimental"). BBV was written by
4436 * Modestly improved support for running Windows applications under
4437 Wine. In particular, initial support for reading Windows .PDB debug
4438 information has been added.
4441 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
4442 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
4446 * The Valgrind client requests VALGRIND_PRINTF and
4447 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
4448 the string was always printed immediately on its own line. Now, the
4449 string will be added to a buffer but not printed until a newline is
4450 encountered, or other Valgrind output is printed (note that for
4451 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
4452 "other Valgrind output"). This allows you to use multiple
4453 VALGRIND_PRINTF calls to build up a single output line, and also to
4454 print multiple output lines with a single request (by embedding
4455 multiple newlines in the string).
4458 * The graphs drawn by Massif's ms_print program have changed slightly:
4460 - The half-height chars '.' and ',' are no longer drawn, because
4461 they are confusing. The --y option can be used if the default
4462 y-resolution is not high enough.
4464 - Horizontal lines are now drawn after the top of a snapshot if
4465 there is a gap until the next snapshot. This makes it clear that
4466 the memory usage has not dropped to zero between snapshots.
4469 * Something that happened in 3.4.0, but wasn't clearly announced: the
4470 option --read-var-info=yes can be used by some tools (Memcheck,
4471 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
4472 variable type and location information. This makes those tools
4473 start up more slowly and increases memory consumption, but
4474 descriptions of data addresses in error messages become more
4478 * exp-Omega, an experimental instantaneous leak-detecting tool, was
4479 disabled in 3.4.0 due to a lack of interest and maintenance,
4480 although the source code was still in the distribution. The source
4481 code has now been removed from the distribution. For anyone
4482 interested, the removal occurred in SVN revision r10247.
4485 * Some changes have been made to the build system.
4487 - VEX/ is now integrated properly into the build system. This means
4488 that dependency tracking within VEX/ now works properly, "make
4489 install" will work without requiring "make" before it, and
4490 parallel builds (ie. 'make -j') now work (previously a
4491 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
4492 was effectively ignored).
4494 - The --with-vex configure option has been removed. It was of
4495 little use and removing it simplified the build system.
4497 - The location of some install files has changed. This should not
4498 affect most users. Those who might be affected:
4500 * For people who use Valgrind with MPI programs, the installed
4501 libmpiwrap.so library has moved from
4502 $(INSTALL)/<platform>/libmpiwrap.so to
4503 $(INSTALL)/libmpiwrap-<platform>.so.
4505 * For people who distribute standalone Valgrind tools, the
4506 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
4507 have moved to $(INSTALL)/libcoregrind-<platform>.a.
4509 These changes simplify the build system.
4511 - Previously, all the distributed suppression (*.supp) files were
4512 installed. Now, only default.supp is installed. This should not
4513 affect users as the other installed suppression files were not
4514 read; the fact that they were installed was a mistake.
4517 * KNOWN LIMITATIONS:
4519 - Memcheck is unusable with the Intel compiler suite version 11.1,
4520 when it generates code for SSE2-and-above capable targets. This
4521 is because of icc's use of highly optimised inlined strlen
4522 implementations. It causes Memcheck to report huge numbers of
4523 false errors even in simple programs. Helgrind and DRD may also
4526 Versions 11.0 and earlier may be OK, but this has not been
4530 The following bugs have been fixed or resolved. Note that "n-i-bz"
4531 stands for "not in bugzilla" -- that is, a bug that was reported to us
4532 but never got a bugzilla entry. We encourage you to file bugs in
4533 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4534 mailing the developers (or mailing lists) directly -- bugs that are
4535 not entered into bugzilla tend to get forgotten about or ignored.
4537 To see details of a given bug, visit
4538 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4539 where XXXXXX is the bug number as listed below.
4541 84303 How about a LockCheck tool?
4542 91633 dereference of null ptr in vgPlain_st_basetype
4543 97452 Valgrind doesn't report any pthreads problems
4544 100628 leak-check gets assertion failure when using
4545 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4546 108528 NPTL pthread cleanup handlers not called
4547 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4548 110128 mallinfo is not implemented...
4549 110770 VEX: Generated files not always updated when making valgrind
4550 111102 Memcheck: problems with large (memory footprint) applications
4551 115673 Vex's decoder should never assert
4552 117564 False positive: Syscall param clone(child_tidptr) contains
4553 uninitialised byte(s)
4554 119404 executing ssh from inside valgrind fails
4555 133679 Callgrind does not write path names to sources with dwarf debug
4557 135847 configure.in problem with non gnu compilers (and possible fix)
4558 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4559 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4560 136230 memcheck reports "possibly lost", should be "still reachable"
4561 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4562 137904 Valgrind reports a memory leak when using POSIX threads,
4564 139076 valgrind VT_GETSTATE error
4565 142228 complaint of elf_dynamic_do_rela in trivial usage
4566 145347 spurious warning with USBDEVFS_REAPURB
4567 148441 (wine) can't find memory leak in Wine, win32 binary
4569 148742 Leak-check fails assert on exit
4570 149878 add (proper) check for calloc integer overflow
4571 150606 Call graph is broken when using callgrind control
4572 152393 leak errors produce an exit code of 0. I need some way to
4573 cause leak errors to result in a nonzero exit code.
4574 157154 documentation (leak-resolution doc speaks about num-callers
4575 def=4) + what is a loss record
4576 159501 incorrect handling of ALSA ioctls
4577 162020 Valgrinding an empty/zero-byte file crashes valgrind
4578 162482 ppc: Valgrind crashes while reading stabs information
4579 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4580 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4581 163560 VEX/test_main.c is missing from valgrind-3.3.1
4582 164353 malloc_usable_size() doesn't return a usable size
4583 165468 Inconsistent formatting in memcheck manual -- please fix
4584 169505 main.c:286 (endOfInstr):
4585 Assertion 'ii->cost_offset == *cost_offset' failed
4586 177206 Generate default.supp during compile instead of configure
4587 177209 Configure valt_load_address based on arch+os
4588 177305 eventfd / syscall 323 patch lost
4589 179731 Tests fail to build because of inlining of non-local asm labels
4590 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4591 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4592 181594 Bogus warning for empty text segment
4593 181707 dwarf doesn't require enumerations to have name
4594 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4595 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4596 Assertion '!already_present' failed.
4597 185359 exp-ptrcheck: unhandled syscall getresuid()
4598 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4599 185816 Valgrind is unable to handle debug info for files with split
4600 debug info that are prelinked afterwards
4601 185980 [darwin] unhandled syscall: sem_open
4602 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4603 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4604 186790 Suppression pattern used for leaks are not reported
4605 186796 Symbols with length>200 in suppression files are ignored
4606 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4607 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4608 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4609 188046 bashisms in the configure script
4610 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4611 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4612 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4613 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4614 assertion fail "!lock->heldBy"
4615 188427 Add support for epoll_create1 (with patch)
4616 188530 Support for SIOCGSTAMPNS
4617 188560 Include valgrind.spec in the tarball
4618 188572 Valgrind on Mac should suppress setenv() mem leak
4619 189054 Valgrind fails to build because of duplicate non-local asm labels
4620 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4621 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4622 189763 drd assertion failure: s_threadinfo[tid].is_recording
4623 190219 unhandled syscall: 328 (x86-linux)
4624 190391 dup of 181394; see above
4625 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4626 190820 No debug information on powerpc-linux
4627 191095 PATCH: Improve usbdevfs ioctl handling
4628 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4630 191189 --xml=yes should obey --gen-suppressions=all
4631 191192 syslog() needs a suppression on macosx
4632 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4633 191761 getrlimit on MacOSX
4634 191992 multiple --fn-skip only works sometimes; dependent on order
4635 192634 V. reports "aspacem sync_check_mapping_callback:
4636 segment mismatch" on Darwin
4637 192954 __extension__ missing on 2 client requests
4638 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4639 194474 "INSTALL" file has different build instructions than "README"
4640 194671 Unhandled syscall (sem_wait?) from mac valgrind
4641 195069 memcheck: reports leak (memory still reachable) for
4643 195169 drd: (vgDrd_barrier_post_wait):
4644 Assertion 'r->sg[p->post_iteration]' failed.
4645 195268 valgrind --log-file doesn't accept ~/...
4646 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4647 195860 WARNING: unhandled syscall: unix:223
4648 196528 need a error suppression for pthread_rwlock_init under os x?
4649 197227 Support aio_* syscalls on Darwin
4650 197456 valgrind should reject --suppressions=(directory)
4651 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4652 197591 unhandled syscall 27 (mincore)
4653 197793 Merge DCAS branch to the trunk == 85756, 142103
4654 197794 Avoid duplicate filenames in Vex
4655 197898 make check fails on current SVN
4656 197901 make check fails also under exp-ptrcheck in current SVN
4657 197929 Make --leak-resolution=high the default
4658 197930 Reduce spacing between leak reports
4659 197933 Print command line of client at start-up, and shorten preamble
4660 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4661 198395 add BBV to the distribution as an experimental tool
4662 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4663 198649 callgrind_annotate doesn't cumulate counters
4664 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4665 199977 Valgrind complains about an unrecognized instruction in the
4666 atomic_incs test program
4667 200029 valgrind isn't able to read Fedora 12 debuginfo
4668 200760 darwin unhandled syscall: unix:284
4669 200827 DRD doesn't work on Mac OS X
4670 200990 VG_(read_millisecond_timer)() does not work correctly
4671 201016 Valgrind does not support pthread_kill() on Mac OS
4672 201169 Document --read-var-info
4673 201323 Pre-3.5.0 performance sanity checking
4674 201384 Review user manual for the 3.5.0 release
4675 201585 mfpvr not implemented on ppc
4676 201708 tests failing because x86 direction flag is left set
4677 201757 Valgrind doesn't handle any recent sys_futex additions
4678 204377 64-bit valgrind can not start a shell script
4679 (with #!/path/to/shell) if the shell is a 32-bit executable
4680 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4681 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4682 about memory allocation events with memory access tracing enabled
4683 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4685 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4689 Release 3.4.1 (28 February 2009)
4690 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4691 3.4.1 is a bug-fix release that fixes some regressions and assertion
4692 failures in debug info reading in 3.4.0, most notably incorrect stack
4693 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4694 other debug info problems are also fixed. A number of bugs in the
4695 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4697 In view of the fact that 3.4.0 contains user-visible regressions
4698 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4699 encouraged to ship 3.4.1 in preference to 3.4.0.
4701 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4702 bugzilla" -- that is, a bug that was reported to us but never got a
4703 bugzilla entry. We encourage you to file bugs in bugzilla
4704 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4705 developers (or mailing lists) directly -- bugs that are not entered
4706 into bugzilla tend to get forgotten about or ignored.
4708 n-i-bz Fix various bugs reading icc-11 generated debug info
4709 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4710 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4711 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4712 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4713 179618 exp-ptrcheck crashed / exit prematurely
4714 179624 helgrind: false positive races with pthread_create and
4715 recv/open/close/read
4716 134207 pkg-config output contains @VG_PLATFORM@
4717 176926 floating point exception at valgrind startup with PPC 440EPX
4718 181594 Bogus warning for empty text segment
4719 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4720 181707 Dwarf3 doesn't require enumerations to have name
4721 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4722 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4723 Assertion '!already_present' failed.
4724 185359 exp-ptrcheck unhandled syscall getresuid()
4726 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4727 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4731 Release 3.4.0 (2 January 2009)
4732 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4733 3.4.0 is a feature release with many significant improvements and the
4734 usual collection of bug fixes. This release supports X86/Linux,
4735 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4736 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4738 3.4.0 brings some significant tool improvements. Memcheck can now
4739 report the origin of uninitialised values, the thread checkers
4740 Helgrind and DRD are much improved, and we have a new experimental
4741 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4742 global arrays. In detail:
4744 * Memcheck is now able to track the origin of uninitialised values.
4745 When it reports an uninitialised value error, it will try to show
4746 the origin of the value, as either a heap or stack allocation.
4747 Origin tracking is expensive and so is not enabled by default. To
4748 use it, specify --track-origins=yes. Memcheck's speed will be
4749 essentially halved, and memory usage will be significantly
4750 increased. Nevertheless it can drastically reduce the effort
4751 required to identify the root cause of uninitialised value errors,
4752 and so is often a programmer productivity win, despite running more
4755 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4756 3.4.0, will be released shortly.
4758 * Helgrind's race detection algorithm has been completely redesigned
4759 and reimplemented, to address usability and scalability concerns:
4761 - The new algorithm has a lower false-error rate: it is much less
4762 likely to report races that do not really exist.
4764 - Helgrind will display full call stacks for both accesses involved
4765 in a race. This makes it easier to identify the root causes of
4768 - Limitations on the size of program that can run have been removed.
4770 - Performance has been modestly improved, although that is very
4773 - Direct support for Qt4 threading has been added.
4775 - pthread_barriers are now directly supported.
4777 - Helgrind works well on all supported Linux targets.
4779 * The DRD thread debugging tool has seen major improvements:
4781 - Greatly improved performance and significantly reduced memory
4784 - Support for several major threading libraries (Boost.Thread, Qt4,
4785 glib, OpenMP) has been added.
4787 - Support for atomic instructions, POSIX semaphores, barriers and
4788 reader-writer locks has been added.
4790 - Works now on PowerPC CPUs too.
4792 - Added support for printing thread stack usage at thread exit time.
4794 - Added support for debugging lock contention.
4796 - Added a manual for Drd.
4798 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4799 checks for misuses of pointers. In that sense it is a bit like
4800 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4801 detect overruns of stack and global arrays, it can detect
4802 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4803 detect accesses heap blocks that have been freed a very long time
4804 ago (millions of blocks in the past).
4806 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4807 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4808 of the main Valgrind documentation. As this is an experimental
4809 tool, we would be particularly interested in hearing about your
4810 experiences with it.
4812 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4813 longer built by default, although the code remains in the repository
4814 and the tarball. This is due to three factors: a perceived lack of
4815 users, a lack of maintenance, and concerns that it may not be
4816 possible to achieve reliable operation using the existing design.
4818 * As usual, support for the latest Linux distros and toolchain
4819 components has been added. It should work well on Fedora Core 10,
4820 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4821 state) is supported, as is glibc-2.9. The C++ demangler has been
4822 updated so as to work well with C++ compiled by even the most recent
4825 * You can now use frame-level wildcards in suppressions. This was a
4826 frequently-requested enhancement. A line "..." in a suppression now
4827 matches zero or more frames. This makes it easier to write
4828 suppressions which are precise yet insensitive to changes in
4831 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4833 * Very basic support for IBM Power6 has been added (64-bit processes only).
4835 * Valgrind is now cross-compilable. For example, it is possible to
4836 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4837 on a ppc32/64-linux target.
4839 * You can set the main thread's stack size at startup using the
4840 new --main-stacksize= flag (subject of course to ulimit settings).
4841 This is useful for running apps that need a lot of stack space.
4843 * The limitation that you can't use --trace-children=yes together
4844 with --db-attach=yes has been removed.
4846 * The following bugs have been fixed. Note that "n-i-bz" stands for
4847 "not in bugzilla" -- that is, a bug that was reported to us but
4848 never got a bugzilla entry. We encourage you to file bugs in
4849 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4850 mailing the developers (or mailing lists) directly.
4852 n-i-bz Make return types for some client requests 64-bit clean
4853 n-i-bz glibc 2.9 support
4854 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4855 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4856 n-i-bz Building in an env without gdb gives bogus gdb attach
4857 92456 Tracing the origin of uninitialised memory
4858 106497 Valgrind does not demangle some C++ template symbols
4860 151612 Suppression with "..." (frame-level wildcards in .supp files)
4861 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4862 159285 unhandled syscall:25 (stime, on x86-linux)
4863 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4864 160954 ppc build of valgrind crashes with illegal instruction (isel)
4865 160956 mallinfo implementation, w/ patch
4866 162092 Valgrind fails to start gnome-system-monitor
4867 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4868 163794 assertion failure with "--track-origins=yes"
4869 163933 sigcontext.err and .trapno must be set together
4870 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4871 164476 Missing kernel module loading system calls
4872 164669 SVN regression: mmap() drops posix file locks
4873 166581 Callgrind output corruption when program forks
4874 167288 Patch file for missing system calls on Cell BE
4875 168943 unsupported scas instruction pentium
4876 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4877 172417 x86->IR: 0x82 ...
4878 172563 amd64->IR: 0xD9 0xF5 - fprem1
4879 173099 .lds linker script generation error
4880 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4881 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4883 174908 --log-file value not expanded correctly for core file
4884 175044 Add lookup_dcookie for amd64
4885 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4887 Developer-visible changes:
4889 * Valgrind's debug-info reading machinery has been majorly overhauled.
4890 It can now correctly establish the addresses for ELF data symbols,
4891 which is something that has never worked properly before now.
4893 Also, Valgrind can now read DWARF3 type and location information for
4894 stack and global variables. This makes it possible to use the
4895 framework to build tools that rely on knowing the type and locations
4896 of stack and global variables, for example exp-Ptrcheck.
4898 Reading of such information is disabled by default, because most
4899 tools don't need it, and because it is expensive in space and time.
4900 However, you can force Valgrind to read it, using the
4901 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4902 make use of such information, if present, to provide source-level
4903 descriptions of data addresses in the error messages they create.
4905 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4906 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).