1 Release 3.21.0 (?? Apr 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 * When GDB is used to debug a program running under valgrind using
14 the valgrind gdbserver, GDB will automatically load some
15 python code provided in valgrind defining GDB front end commands
16 corresponding to the valgrind monitor commands.
17 These GDB front end commands accept the same format as
18 the monitor commands directly sent to the Valgrind gdbserver.
19 These GDB front end commands provide a better integration
20 in the GDB command line interface, so as to use for example
21 GDB auto-completion, command specific help, searching for
22 a command or command help matching a regexp, ...
23 For relevant monitor commands, GDB will evaluate arguments
24 to make the use of monitor commands easier.
25 For example, instead of having to print the address of a variable
26 to pass it to a subsequent monitor command, the GDB front end
27 command will evaluate the address argument. It is for example
29 (gdb) memcheck who_point_at &some_struct sizeof(some_struct)
32 $2 = (some_struct_type *) 0x1130a0 <some_struct>
33 (gdb) p sizeof(some_struct)
35 (gdb) monitor who_point_at 0x1130a0 40
37 * The vgdb utility now supports extended-remote protocol when
38 invoked with --multi. In this mode the GDB run command is
39 supported. Which means you don't need to run gdb and valgrind
40 from different terminals. So for example to start you program
41 in gdb and run it under valgrind you can do:
43 (gdb) set remote exec-file prog
45 (gdb) target extended-remote | vgdb --multi
48 * The behaviour of realloc with a size of zero can now
49 be changed for tools that intercept malloc. Those
50 tools are memcheck, helgrind, drd, massif and dhat.
51 Realloc implementations generally do one of two things
52 - free the memory like free() and return NULL
53 (GNU libc and ptmalloc).
54 - either free the memory and then allocate a
55 minumum siized block or just return the
56 original pointer. Return NULL if the
57 allocation of the minimum sized block fails
58 (jemalloc, musl, snmalloc, Solaris, macOS).
59 When Valgrind is configured and built it will
60 try to match the OS and libc behaviour. However
61 if you are using a non-default library to replace
62 malloc and family (e.g., musl on a glibc Linux or
63 tcmalloc on FreeBSD) then you can use a command line
64 option to change the behaviour of Valgrind:
65 --realloc-zero-bytes-frees=yes|no [yes on Linux glibc, no otherwise]
67 * ================== PLATFORM CHANGES =================
69 * Make the address space limit on FreeBSD amd64 128Gbytes
70 (the same as Linux and Solaris, it was 32Gbytes)
72 * ==================== TOOL CHANGES ===================
75 - When doing a delta leak_search, it is now possible to only
76 output the new loss records compared to the previous leak search.
77 This is available in the memcheck monitor command 'leak_search'
78 by specifying the "new" keyword or in your program by using
79 the client request VALGRIND_DO_NEW_LEAK_CHECK.
80 Whenever a "delta" leak search is done (i.e. when specifying
81 "new" or "increased" or "changed" in the monitor command),
82 the new loss records have a "new" marker.
83 - Valgrind now contains python code that defines GDB memcheck
84 front end monitor commands. See CORE CHANGES.
85 - Performs checks for the use of realloc with a size of zero.
86 This is non-portable and a source of errors. If memcheck
87 detects such a usage it will generate an error
89 followed by the usual callstacks.
90 A switch has been added to allow this to be turned off:
91 --show-realloc-size-zero=yes|no [yes]
94 - The option ---history-backtrace-size=<number> allows to configure
95 the number of entries to record in the stack traces of "old"
96 accesses. Previously, this number was hardcoded to 8.
97 - Valgrind now contains python code that defines GDB helgrind
98 front end monitor commands. See CORE CHANGES.
101 - `cg_annotate` has been rewritten from Perl into Python. The new
102 version is twice as fast, has more flexible argument parsing, and
103 will make future improvements easier.
106 - Valgrind now contains python code that defines GDB callgrind
107 front end monitor commands. See CORE CHANGES.
110 - Valgrind now contains python code that defines GDB massif
111 front end monitor commands. See CORE CHANGES.
113 * ==================== FIXED BUGS ====================
115 The following bugs have been fixed or resolved. Note that "n-i-bz"
116 stands for "not in bugzilla" -- that is, a bug that was reported to us
117 but never got a bugzilla entry. We encourage you to file bugs in
118 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
119 than mailing the developers (or mailing lists) directly -- bugs that
120 are not entered into bugzilla tend to get forgotten about or ignored.
122 170510 Don't warn about ioctl of size 0 without direction hint
123 241072 List tools in --help output
124 327548 false positive while destroying mutex
125 382034 Testcases build fixes for musl
126 351857 confusing error message about valid command line option
127 374596 inconsistent RDTSCP support on x86_64
128 392331 Spurious lock not held error from inside pthread_cond_timedwait
129 400793 pthread_rwlock_timedwrlock false positive
130 419054 Unhandled syscall getcpu on arm32
131 433873 openat2 syscall unimplemented on Linux
132 434057 Add stdio mode to valgrind's gdbserver
133 435441 valgrind fails to interpose malloc on musl 1.2.2 due to weak symbol name and no libc soname
134 436413 Warn about realloc of size zero
135 439685 compiler warning in callgrind/main.c
136 444110 priv/guest_ppc_toIR.c:36198:31: warning: duplicated 'if' condition.
137 444487 hginfo test detects an extra lock inside data symbol "_rtld_local"
138 444488 Use glibc.pthread.stack_cache_size tunable
139 444568 drd/tests/pth_barrier_thr_cr fails on Fedora 38
140 445743 "The impossible happened: mutex is locked simultaneously by two threads"
141 while using mutexes with priority inheritance and signals
142 449309 Missing loopback device ioctl(s)
143 459476 vgdb: allow address reuse to avoid "address already in use" errorsuse" errors
144 460356 s390: Sqrt32Fx4 -- cannot reduce tree
145 462830 WARNING: unhandled amd64-freebsd syscall: 474
146 463027 broken check for MPX instruction support in assembler
147 464476 Firefox fails to start under Valgrind
148 464609 Valgrind memcheck should support Linux pidfd_open
149 464680 Show issues caused by memory policies like selinux deny_execmem
150 464859 Build failures with GCC-13 (drd tsan_unittest)
151 464969 D language demangling
152 465435 m_libcfile.c:66 (vgPlain_safe_fd): Assertion 'newfd >= VG_(fd_hard_limit)' failed.
153 466104 aligned_alloc problems, part 1
154 467036 Add time cost statistics for Regtest
155 467482 Build failure on aarch64 Alpine
156 467714 fdleak_* and rlimit tests fail when parent process has more than
157 64 descriptors opened
158 467839 Gdbserver: Improve compatibility of library directory name
159 468401 [PATCH] Add a style file for clang-format
160 468556 Build failure for vgdb
161 468606 build: remove "Valgrind relies on GCC" check/output
162 n-i-bz FreeBSD rfork syscall fail with EINVAL or ENOSYS rather than VG_(unimplemented)
164 To see details of a given bug, visit
165 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
166 where XXXXXX is the bug number as listed above.
168 (3.21.0.RC1: 14 Apr 2023)
170 Release 3.20.0 (24 Oct 2022)
171 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
173 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
174 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
175 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
176 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
177 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
178 AMD64/macOS 10.13 and nanoMIPS/Linux.
180 * ==================== CORE CHANGES ===================
182 * The option "--vgdb-stop-at=event1,event2,..." accepts the new value abexit.
183 This indicates to invoke gdbserver when your program exits abnormally
184 (i.e. with a non zero exit code).
185 * Fix Rust v0 name demangling.
186 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
187 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
188 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with
190 * The option --enable-debuginfod=<no|yes> [default: yes] has been added on
192 * More DWARF5 support as generated by clang14.
194 * ==================== FIXED BUGS ====================
196 The following bugs have been fixed or resolved. Note that "n-i-bz"
197 stands for "not in bugzilla" -- that is, a bug that was reported to us
198 but never got a bugzilla entry. We encourage you to file bugs in
199 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
200 than mailing the developers (or mailing lists) directly -- bugs that
201 are not entered into bugzilla tend to get forgotten about or ignored.
203 131186 writev reports error in (vector[...])
204 434764 iconv_open causes ld.so v2.28+ to use optimised strncmp
205 446754 Improve error codes from alloc functions under memcheck
206 452274 memcheck crashes with Assertion 'sci->status.what == SsIdle' failed
207 452779 Valgrind fails to build on FreeBSD 13.0 with llvm-devel (15.0.0)
208 453055 shared_timed_mutex drd test fails with "Lock shared failed" message
209 453602 Missing command line option to enable/disable debuginfod
210 452802 Handle lld 9+ split RW PT_LOAD segments correctly
211 454040 s390x: False-positive memcheck:cond in memmem on arch13 systems
212 456171 [PATCH] FreeBSD: Don't record address errors when accessing the 'kern.ps_strings' sysctl struct
213 n-i-bz Implement vgdb invoker on FreeBSD
214 458845 PowerPC: The L field for the dcbf and sync instruction should be
216 458915 Remove register cache to fix 458915 gdbserver causes wrong syscall return
217 459031 Documentation on --error-exitcode incomplete
218 459477 XERROR messages lacks ending '\n' in vgdb
219 462007 Implicit int in none/tests/faultstatus.c
221 To see details of a given bug, visit
222 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
223 where XXXXXX is the bug number as listed above.
225 (3.20.0.RC1: 20 Oct 2022)
228 Release 3.19.0 (11 Apr 2022)
229 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
231 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
232 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
233 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
234 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
235 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
236 AMD64/macOS 10.13 and nanoMIPS/Linux.
238 * ==================== CORE CHANGES ===================
240 * Fix Rust v0 name demangling.
241 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
242 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
243 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with HardenedBSD
245 * ================== PLATFORM CHANGES =================
248 - ignore the "v8.x" architecture levels, only look at actual CPU features
249 present. Fixes mismatch detected between RDMA and atomics features
250 preventing startup on some QEMU configurations.
251 - Implement LD{,A}XP and ST{,L}XP
252 - Fix incorrect code emitted for doubleword CAS.
255 - Fix sys_ipc semtimedop syscall
256 - Fix VFLRX and WFLRX instructions
257 - Fix EXRL instruction with negative offset
260 - Reimplement the vbpermq instruction support to generate less Iops and
261 avoid overflowing internal buffers.
262 - Fix checking for scv support to avoid "Facility 'SCV' unavailable (12),
263 exception" messages in dmsg.
264 - Fix setting condition code for Vector Compare quad word instructions.
265 - Fix fix lxsibzx, lxsihzx and lxsihzx instructions so they only load
266 their respective sized data.
267 - Fix the prefixed stq instruction in PC relative mode.
269 * ==================== TOOL CHANGES ===================
272 - Speed up --track-origins=yes for large (in the range of hundreds to
273 thousands of megabytes) mmap/munmaps.
275 - Several fixes for new versions of libstd++ using new posix try_lock
278 * ==================== FIXED BUGS ====================
280 The following bugs have been fixed or resolved. Note that "n-i-bz"
281 stands for "not in bugzilla" -- that is, a bug that was reported to us
282 but never got a bugzilla entry. We encourage you to file bugs in
283 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
284 than mailing the developers (or mailing lists) directly -- bugs that
285 are not entered into bugzilla tend to get forgotten about or ignored.
287 403802 leak_cpp_interior fails with some reachable blocks different than expected
288 435732 memcheck/tests/leak_cpp_interior fails with gcc11
289 444242 s390x: Valgrind crashes on EXRL with negative offset
290 444399 arm64: unhandled instruction 0xC87F2D89 (LD{,A}XP and ST{,L}XP).
292 444481 gdb_server test failures on s390x
293 444495 dhat/tests/copy fails on s390x
294 444552 memcheck/tests/sem fails on s390x with glibc 2.34
295 444571 PPC, fix the lxsibzx and lxsihzx so they only load their respective
297 444836 PPC, pstq instruction for R=1 is not storing to the correct address.
298 444925 fexecve syscall wrapper not properly implemented
299 445032 valgrind/memcheck crash with SIGSEGV when SIGVTALRM timer used and
301 445211 Fix out of tree builds
302 445300 [PATCH] Fix building tests with Musl
303 445011 SIGCHLD is sent when valgrind uses debuginfod-find
304 445354 arm64 backend: incorrect code emitted for doubleword CAS
305 445415 arm64 front end: alignment checks missing for atomic instructions
306 445504 Using C++ condition_variable results in bogus "mutex is locked simultaneously by two threads" warning
307 445607 Unhandled amd64-freebsd syscall: 247
308 445668 Inline stack frame generation is broken for Rust binaries
309 445916 Demangle Rust v0 symbols with .llvm suffix
310 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false positives
311 446138 DRD/Helgrind with std::timed_mutex::try_lock_until false positives
312 446281 Add a DRD suppression for fwrite
313 446103 Memcheck: `--track-origins=yes` causes extreme slowdowns for large mmap/munmap
314 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false
315 446251 TARGET_SIGNAL_THR added to enum target_signal
316 446823 FreeBSD - missing syscalls when using libzm4
317 447991 s390x: Valgrind indicates illegal instruction on wflrx
318 447995 Valgrind segfault on power10 due to hwcap checking code
319 449483 Powerpc: vcmpgtsq., vcmpgtuq,, vcmpequq. instructions not setting the
320 condition code correctly.
321 449672 ppc64 --track-origins=yes failures because of bad cmov addHRegUse
322 449838 sigsegv liburing the 'impossible' happened for io_uring_setup
323 450025 Powerc: ACC file not implemented as a logical overlay of the VSR
325 450437 Warn for execve syscall with argv or argv[0] being NULL
326 450536 Powerpc: valgrind throws 'facility scv unavailable exception'
327 451626 Syscall param bpf(attr->raw_tracepoint.name) points to unaddressable byte(s)
328 451827 [ppc64le] VEX temporary storage exhausted with several vbpermq instructions
329 451843 valgrind fails to start on a FreeBSD system which enforces W^X
331 To see details of a given bug, visit
332 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
333 where XXXXXX is the bug number as listed above.
335 (3.19.0.RC1: 02 Apr 2022)
336 (3.19.0.RC2: 08 Apr 2022)
339 Release 3.18.0 (15 Oct 2021)
340 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
342 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
343 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
344 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
345 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
346 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
347 AMD64/macOS 10.13 and nanoMIPS/Linux.
349 * ==================== CORE CHANGES ===================
351 * The libiberty demangler has been updated, which brings support for
352 Rust v0 name demangling. [Update: alas, due to a bug, this support
353 isn't working in 3.18.0.]
355 * __libc_freeres isn't called anymore after the program recieves a
356 fatal signal. Causing some internal glibc resources to hang around,
357 but preventing any crashes after the program has ended.
359 * The DWARF reader is now very much faster at startup when just
360 --read-inline-info=yes (the default in most cases) is given.
362 * glibc 2.34, which moved various functions from libpthread.so into
363 libc.so, is now supported.
365 * ================== PLATFORM CHANGES =================
369 - v8.2 scalar and vector FABD, FACGE, FACGT and FADD.
370 - v8.2 FP compare & conditional compare instructions.
371 - Zero variants of v8.2 FP compare instructions.
375 - Support the miscellaneous-instruction-extensions facility 3 and
376 the vector-enhancements facility 2. This enables programs
377 compiled with "-march=arch13" or "-march=z15" to be executed
382 - ISA 3.1 support is now complete
383 - ISA 3.0 support for the darn instruction added.
384 - ISA 3.0 support for the vector system call instruction scv added.
385 - ISA 3.0 support for the copy, paste and cpabort instructions added.
387 * Support for X86/FreeBSD and AMD64/FreeBSD has been added.
389 * ==================== OTHER CHANGES ====================
391 * Memcheck on amd64: minor fixes to remove some false positive
394 * ==================== FIXED BUGS ====================
396 The following bugs have been fixed or resolved. Note that "n-i-bz"
397 stands for "not in bugzilla" -- that is, a bug that was reported to us
398 but never got a bugzilla entry. We encourage you to file bugs in
399 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
400 than mailing the developers (or mailing lists) directly -- bugs that
401 are not entered into bugzilla tend to get forgotten about or ignored.
403 208531 [PATCH]: FreeBSD support for valgrind
404 368960 WARNING: unhandled amd64-linux syscall: 163 (acct)
405 407589 [Linux] Add support for C11 aligned_alloc() and GNU reallocarray()
406 423963 Error in child thread when CLONE_PIDFD is used
407 426148 crash with "impossible happened" when running BPF CO-RE programs
408 429375 PPC ISA 3.1 support is missing, part 9
409 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
410 431306 Update demangler to support Rust v0 name mangling
411 432387 s390x: z15 instructions support
412 433437 FreeBSD support, part 1
413 433438 FreeBSD support, part 2
414 433439 FreeBSD support, part 3
415 433469 FreeBSD support, part 4
416 433473 FreeBSD support, part 5
417 433477 FreeBSD support, part 6
418 433479 FreeBSD support, part 7
419 433504 FreeBSD support, part 8
420 433506 FreeBSD support, part 9
421 433507 FreeBSD support, part 10
422 433508 FreeBSD support, part 11
423 433510 FreeBSD support, part 12
424 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
425 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
426 434296 s390x: False-positive memcheck diagnostics from vector string
428 434840 PPC64 darn instruction not supported
429 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
430 435908 valgrind tries to fetch from deubginfod for files which already
431 have debug information
432 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2
433 439046 valgrind is unusably large when linked with lld
434 439090 Implement close_range(2)
435 439326 Valgrind 3.17.0 won't compile with Intel 2021 oneAPI compilers
436 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
437 440670 unhandled ppc64le-linux syscall: 252 statfs64 and 253 fstatfs64
438 440906 Fix impossible constraint issue in P10 testcase.
439 441512 Remove a unneeded / unnecessary prefix check.
440 441534 Update the expected output for test_isa_3_1_VRT.
441 442061 very slow execution under Fedora 34 (readdwarf3)
442 443031 Gcc -many change requires explicit .machine directives
443 443033 Add support for the ISA 3.0 mcrxrx instruction
444 443034 Sraw, srawi, srad, sradi, mfs
445 443178 Powerpc, test jm-mfspr expected output needs to be updated.
446 443179 Need new test for the lxvx and stxvx instructions on ISA 2.07 and
448 443180 The subnormal test and the ISA 3.0 test generate compiler warnings
449 443314 In the latest GIT version, Valgrind with "--trace-flags" crashes
451 443605 Don't call final_tidyup (__libc_freeres) on FatalSignal
453 To see details of a given bug, visit
454 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
455 where XXXXXX is the bug number as listed below.
457 (3.18.0.RC1: 12 Oct 2021)
458 (3.18.0: 15 Oct 2021)
462 Release 3.17.0 (19 Mar 2021)
463 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
465 3.17.0 fixes a number of bugs and adds some functional changes: support for
466 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
467 some new instructions for Arm64, S390 and POWER. There are also some tool
470 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
471 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
472 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
473 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
474 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
476 * ==================== CORE CHANGES ===================
478 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
481 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
482 debugging information. When a debuginfo file cannot be found locally,
483 Valgrind is able to query debuginfod servers for the file using its
484 build-id. See the user manual for more information about debuginfod support.
486 * ================== PLATFORM CHANGES =================
490 - Inaccuracies resulting from double-rounding in the simulation of
491 floating-point multiply-add/subtract instructions have been fixed. These
492 should now behave exactly as the hardware does.
494 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
495 ongoing. Support for the half-word variants of at least the following
496 instructions has been added:
498 FABS <Vd>.<T>, <Vn>.<T>
500 FNEG <Vd>.<T>, <Vn>.<T>
502 FSQRT <Vd>.<T>, <Vn>.<T>
507 - Implement the new instructions/features that were added to z/Architecture
508 with the vector-enhancements facility 1. Also cover the instructions from
509 the vector-packed-decimal facility that are defined outside the chapter
510 "Vector Decimal Instructions", but not the ones from that chapter itself.
512 For a detailed list of newly supported instructions see the updates to
513 `docs/internals/s390-opcodes.csv'.
515 Since the miscellaneous instruction extensions facility 2 was already
516 added in Valgrind 3.16.0, this completes the support necessary to run
517 general programs built with `--march=z14' under Valgrind. The
518 vector-packed-decimal facility is currently not exploited by the standard
519 toolchain and libraries.
523 - Various bug fixes. Fix for the sync field to limit setting just two of
524 the two bits in the L-field. Fix the write size for the stxsibx and
525 stxsihx instructions. Fix the modsw and modsd instructions.
527 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
528 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
529 conversion instructions are still missing.
531 * ==================== TOOL CHANGES ====================
533 * General tool changes
535 - All the tools and their vgpreload libraries are now installed under
536 libexec because they cannot be executed directly and should be run through
537 the valgrind executable. This should be an internal, not user visible,
538 change, but might impact valgrind packagers.
540 - The --track-fds option now respects -q, --quiet and won't output anything
541 if no file descriptors are leaked. It also won't report the standard stdin
542 (0), stdout (1) or stderr (2) descriptors as being leaked with
543 --trace-fds=yes anymore. To track whether the standard file descriptors
544 are still open at the end of the program run use --trace-fds=all.
548 - DHAT has been extended, with two new modes of operation. The new
549 --mode=copy flag triggers copy profiling, which records calls to memcpy,
550 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
551 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
552 the new dhat/dhat.h file. This is useful for learning more about hot code
553 paths. See the user manual for more information about the new modes.
555 - Because of these changes, DHAT's file format has changed. DHAT output
556 files produced with earlier versions of DHAT will not work with this
557 version of DHAT's viewer, and DHAT output files produced with this version
558 of DHAT will not work with earlier versions of DHAT's viewer.
560 * ==================== FIXED BUGS ====================
562 The following bugs have been fixed or resolved. Note that "n-i-bz"
563 stands for "not in bugzilla" -- that is, a bug that was reported to us
564 but never got a bugzilla entry. We encourage you to file bugs in
565 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
566 than mailing the developers (or mailing lists) directly -- bugs that
567 are not entered into bugzilla tend to get forgotten about or ignored.
569 To see details of a given bug, visit
570 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
571 where XXXXXX is the bug number as listed below.
573 140178 open("/proc/self/exe", ...); doesn't quite work
574 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
575 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
576 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
577 345077 linux syscall execveat support (linux 3.19)
578 361770 Missing F_ADD_SEALS
579 369029 handle linux syscalls sched_getattr and sched_setattr
580 384729 __libc_freeres inhibits cross-platform valgrind
581 388787 Support for C++17 new/delete
582 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
583 396656 Warnings while reading debug info
584 397605 ioctl FICLONE mishandled
585 401416 Compile failure with openmpi 4.0
586 408663 Suppression file for musl libc
587 404076 s390x: z14 vector instructions not implemented
588 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
589 (actually affected all x86 and nanomips regardless of host bitness)
590 413547 regression test does not check for Arm 64 features.
591 414268 Enable AArch64 feature detection and decoding for v8.x instructions
592 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
593 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
594 422261 platform selection fails for unqualified client name
595 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
596 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
597 423195 PPC ISA 3.1 support is missing, part 1
598 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
599 424012 crash with readv/writev having invalid but not NULL arg2 iovec
600 424298 amd64: Implement RDSEED
601 425232 PPC ISA 3.1 support is missing, part 2
602 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
603 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
604 426123 PPC ISA 3.1 support is missing, part 3
605 426144 Fix "condition variable has not been initialized" on Fedora 33.
606 427400 PPC ISA 3.1 support is missing, part 4
607 427401 PPC ISA 3.1 support is missing, part 5
608 427404 PPC ISA 3.1 support is missing, part 6
609 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
610 427787 Support new faccessat2 linux syscall (439)
611 427969 debuginfo section duplicates a section in the main ELF file
612 428035 drd: Unbreak the musl build
613 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
614 428716 cppcheck detects potential leak in VEX/useful/smchash.c
615 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
616 429352 PPC ISA 3.1 support is missing, part 7
617 429354 PPC ISA 3.1 support is missing, part 8
618 429692 unhandled ppc64le-linux syscall: 147 (getsid)
619 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
621 429952 Errors when building regtest with clang
622 430354 ppc stxsibx and stxsihx instructions write too much data
623 430429 valgrind.h doesn't compile on s390x with clang
624 430485 expr_is_guardable doesn't handle Iex_Qop
625 431556 Complete arm64 FADDP v8.2 instruction support
626 432102 Add support for DWARF5 as produced by GCC11
627 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
628 432381 drd: Process STACK_REGISTER client requests
629 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
630 432672 vg_regtest: test-specific environment variables not reset between tests
631 432809 VEX should support REX.W + POPF
632 432861 PPC modsw and modsd give incorrect results for 1 mod 12
633 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
634 432215 Add debuginfod functionality
635 433323 Use pkglibexecdir as vglibdir
636 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
637 433629 valgrind/README has type "abd" instead of "and"
638 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
639 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
640 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
641 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
642 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
644 (3.17.0.RC1: 13 Mar 2021)
645 (3.17.0.RC2: 17 Mar 2021)
646 (3.17.0: 19 Mar 2021)
650 Release 3.16.1 (22 June 2020)
651 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
653 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
654 fixes character encoding problems in the documentation HTML.
656 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
657 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
659 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
663 Release 3.16.0 (27 May 2020)
664 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
666 3.16.0 is a feature release with many improvements and the usual collection of
669 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
670 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
671 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
672 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
673 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
675 * ==================== CORE CHANGES ===================
677 * It is now possible to dynamically change the value of many command line
678 options while your program (or its children) are running under Valgrind.
680 To see the list of dynamically changeable options, run
681 "valgrind --help-dyn-options".
683 You can change the options from the shell by using vgdb to launch
684 the monitor command "v.clo <clo option>...".
685 The same monitor command can be used from a gdb connected
686 to the valgrind gdbserver.
687 Your program can also change the dynamically changeable options using
688 the client request VALGRIND_CLO_CHANGE(option).
690 * ================== PLATFORM CHANGES =================
692 * MIPS: preliminary support for nanoMIPS instruction set has been added.
694 * ==================== TOOL CHANGES ====================
698 - The implicit memcpy done by each call to realloc now counts towards the
699 read and write counts of resized heap blocks, making those counts higher
704 - cg_annotate's --auto and --show-percs options now default to 'yes', because
705 they are usually wanted.
709 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
710 because they are usually wanted.
712 - The command option --collect-systime has been enhanced to specify
713 the unit used to record the elapsed time spent during system calls.
714 The command option now accepts the values no|yes|msec|usec|nsec,
715 where yes is a synonym of msec. When giving the value nsec, the
716 system cpu time of system calls is also recorded.
720 - Several memcheck options are now dynamically changeable.
721 Use valgrind --help-dyn-options to list them.
723 - The release 3.15 introduced a backward incompatible change for
724 some suppression entries related to preadv and pwritev syscalls.
725 When reading a suppression entry using the unsupported 3.14 format,
726 valgrind will now produce a warning to say the suppression entry will not
727 work, and suggest the needed change.
729 - Significantly fewer false positive errors on optimised code generated by
730 Clang and GCC. In particular, Memcheck now deals better with the
731 situation where the compiler will transform C-level "A && B" into "B && A"
732 under certain circumstances (in which the transformation is valid).
733 Handling of integer equality/non-equality checks on partially defined
734 values is also improved on some architectures.
738 - The exprimental Stack and Global Array Checking tool has been removed.
739 It only ever worked on x86 and amd64, and even on those it had a
740 high false positive rate and was slow. An alternative for detecting
741 stack and global array overruns is using the AddressSanitizer (ASAN)
742 facility of the GCC and Clang compilers, which require you to rebuild
743 your code with -fsanitize=address.
745 * ==================== OTHER CHANGES ====================
747 * New and modified GDB server monitor features:
749 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
751 - The gdbserver monitor commands that require an address and an optional
752 length argument now accepts the alternate 'C like' syntax "address[length]".
753 For example, the memcheck command "monitor who_points_at 0x12345678 120"
754 can now also be given as "monitor who_points_at 0x12345678[120]".
756 * ==================== FIXED BUGS ====================
758 The following bugs have been fixed or resolved. Note that "n-i-bz"
759 stands for "not in bugzilla" -- that is, a bug that was reported to us
760 but never got a bugzilla entry. We encourage you to file bugs in
761 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
762 than mailing the developers (or mailing lists) directly -- bugs that
763 are not entered into bugzilla tend to get forgotten about or ignored.
765 To see details of a given bug, visit
766 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
767 where XXXXXX is the bug number as listed below.
769 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
770 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
771 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
772 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
773 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
774 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
775 400872 Add nanoMIPS support to Valgrind
776 403212 drd/tests/trylock hangs on FreeBSD
777 404406 s390x: z14 miscellaneous instructions not implemented
778 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
779 406561 mcinfcallWSRU gdbserver_test fails on ppc64
780 406824 Unsupported baseline
781 407218 Add support for the copy_file_range syscall
782 407307 Intercept stpcpy also in ld.so for arm64
783 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
785 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
786 408009 Expose rdrand and f16c even on avx if host cpu supports them
787 408091 Missing pkey syscalls
788 408414 Add support for missing for preadv2 and pwritev2 syscalls
789 409141 Valgrind hangs when SIGKILLed
790 409206 Support for Linux PPS and PTP ioctls
791 409367 exit_group() after signal to thread waiting in futex() causes hangs
792 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
793 409780 References to non-existent configure.in
794 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
795 410599 Non-deterministic behaviour of pth_self_kill_15_other test
796 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
797 411134 Allow the user to change a set of command line options during execution
798 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
799 412344 Problem setting mips flags with specific paths
800 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
801 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
802 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
803 413603 callgrind_annotate/cg_annotate truncate function names at '#'
804 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
805 415136 ARMv8.1 Compare-and-Swap instructions are not supported
806 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
807 416239 valgrind crashes when handling clock_adjtime
808 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
809 416286 DRD reports "conflicting load" error on std::mutex::lock()
810 416301 s390x: "compare and signal" not supported
811 416387 finit_module and bpf syscalls are unhandled on arm64
812 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
813 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
814 416753 new 32bit time syscalls for 2038+
815 417075 pwritev(vector[...]) suppression ignored
816 417075 is not fixed, but incompatible supp entries are detected
817 and a warning is produced for these.
818 417187 [MIPS] Conditional branch problem since 'grail' changes
819 417238 Test memcheck/tests/vbit-test fails on mips64 BE
820 417266 Make memcheck/tests/linux/sigqueue usable with musl
821 417281 s390x: /bin/true segfaults with "grail" enabled
822 417427 commit to fix vki_siginfo_t definition created numerous regression
824 417452 s390_insn_store_emit: dst->tag for HRcVec128
825 417578 Add suppressions for glibc DTV leaks
826 417906 clone with CLONE_VFORK and no CLONE_VM fails
827 418004 Grail code additions break ppc64.
828 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
829 418997 s390x: Support Iex_ITE for float and vector types
830 419503 s390x: Avoid modifying registers returned from isel functions
831 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
832 421570 std_mutex fails on Arm v8.1 h/w
833 434035 vgdb might crash if valgrind is killed
834 n-i-bz Fix minor one time leaks in dhat.
835 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
836 n-i-bz Add support for the Linux io_uring system calls
837 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
838 n-i-bz Fix non-glibc build of test suite with s390x_features
839 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
840 423195 PPC ISA 3.1 support is missing, part 1
842 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
843 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
844 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
848 Release 3.15.0 (12 April 2019)
849 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
851 3.15.0 is a feature release with many improvements and the usual collection of
854 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
855 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
856 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
857 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
858 support for X86/macOS 10.13 and AMD64/macOS 10.13.
860 * ==================== CORE CHANGES ===================
862 * The XTree Massif output format now makes use of the information obtained
863 when specifying --read-inline-info=yes.
865 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
867 * ==================== TOOL CHANGES ====================
871 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
872 it has been promoted from an experimental tool to a regular tool. Run it
873 with --tool=dhat instead of --tool=exp-dhat.
875 - DHAT now prints only minimal data when the program ends, instead writing
876 the bulk of the profiling data to a file. As a result, the --show-top-n
877 and --sort-by options have been removed.
879 - Profile results can be viewed with the new viewer, dh_view.html. When
880 a run ends, a short message is printed, explaining how to view the result.
882 - See the documentation for more details.
886 - cg_annotate has a new option, --show-percs, which prints percentages next
891 - callgrind_annotate has a new option, --show-percs, which prints percentages
892 next to all event counts.
894 - callgrind_annotate now inserts commas in call counts, and
895 sort the caller/callee lists in the call tree.
899 - The default value for --read-inline-info is now "yes" on
900 Linux/Android/Solaris. It is still "no" on other OS.
904 - The option --xtree-leak=yes (to output leak result in xtree format)
905 automatically activates the option --show-leak-kinds=all, as xtree
906 visualisation tools such as kcachegrind can in any case select what kind
907 of leak to visualise.
909 - There has been further work to avoid false positives. In particular,
910 integer equality on partially defined inputs (C == and !=) is now handled
913 * ==================== OTHER CHANGES ====================
915 * The new option --show-error-list=no|yes displays, at the end of the run, the
916 list of detected errors and the used suppressions. Prior to this change,
917 showing this information could only be done by specifying "-v -v", but that
918 also produced a lot of other possibly-non-useful messages. The option -s is
919 equivalent to --show-error-list=yes.
921 * ==================== FIXED BUGS ====================
923 The following bugs have been fixed or resolved. Note that "n-i-bz"
924 stands for "not in bugzilla" -- that is, a bug that was reported to us
925 but never got a bugzilla entry. We encourage you to file bugs in
926 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
927 than mailing the developers (or mailing lists) directly -- bugs that
928 are not entered into bugzilla tend to get forgotten about or ignored.
930 To see details of a given bug, visit
931 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
932 where XXXXXX is the bug number as listed below.
934 385411 s390x: z13 vector floating-point instructions not implemented
935 397187 z13 vector register support for vgdb gdbserver
936 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
937 398870 Please add support for instruction vcvtps2ph
938 399287 amd64 front end: Illegal Instruction vcmptrueps
939 399301 Use inlined frames in Massif XTree output.
940 399322 Improve callgrind_annotate output
941 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
942 400164 helgrind test encounters mips x-compiler warnings and assembler error
943 400490 s390x: VRs allocated as if separate from FPRs
944 400491 s390x: Operand of LOCH treated as unsigned integer
945 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
946 options, which specify a mips64 processor
947 401112 LLVM 5.0 generates comparison against partially initialized data
948 401277 More bugs in z13 support
949 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
950 401578 drd: crashes sometimes on fork()
951 401627 memcheck errors with glibc avx2 optimized wcsncmp
952 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
953 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
954 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
956 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
957 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
958 402123 invalid assembler opcodes for mips32r2
959 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
960 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
961 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
962 402351 mips64 libvexmultiarch_test fails on s390x
964 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
965 402480 Do not use %rsp in clobber list
966 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
967 402515 Implement new option --show-error-list=no|yes / -s
968 402519 POWER 3.0 addex instruction incorrectly implemented
969 402781 Redo the cache used to process indirect branch targets
970 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
971 403552 s390x: wrong facility bit checked for vector facility
972 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
973 404638 Add VG_(replaceIndexXA)
974 404843 s390x: backtrace sometimes ends prematurely
975 404888 autotools cleanup series
976 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
977 405182 Valgrind fails to build with Clang
978 405205 filter_libc: remove the line holding the futex syscall error entirely
979 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
980 the upper and lower 32-bits of the 64-bit result
981 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
982 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
983 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
985 405403 s390x disassembler cannot be used on x86
986 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
987 405458 MIPS mkFormVEC arguments swapped?
988 405716 drd: Fix an integer overflow in the stack margin calculation
989 405722 Support arm64 core dump
990 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
991 of the 64-bit destination field.
992 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
993 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
994 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
996 406256 PPC64, vector floating point instructions don't handle subnormal
997 according to VSCR[NJ] bit setting.
998 406352 cachegrind/callgrind fails ann tests because of missing a.c
999 406354 dhat is broken on x86 (32bit)
1000 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
1001 406357 gdbserver_tests fails because of gdb output change
1002 406360 memcheck/tests/libstdc++.supp needs more supression variants
1003 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
1004 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
1005 407340 PPC64, does not support the vlogefp, vexptefp instructions.
1006 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
1007 n-i-bz Fix callgrind_annotate non deterministic order for equal total
1008 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
1009 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
1010 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
1012 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
1013 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
1014 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
1018 Release 3.14.0 (9 October 2018)
1019 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1021 3.14.0 is a feature release with many improvements and the usual collection of
1024 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1025 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1026 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1027 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
1028 support for X86/macOS 10.13, AMD64/macOS 10.13.
1030 * ==================== CORE CHANGES ===================
1032 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
1033 debug info for unloaded code. This allows saved stack traces (e.g. for
1034 memory leaks) to include file/line info for code that has been dlclose'd (or
1035 similar). See the user manual for more information and known limitations.
1037 * Ability to specify suppressions based on source file name and line number.
1039 * Majorly overhauled register allocator. No end-user changes, but the JIT
1040 generates code a bit more quickly now.
1042 * ================== PLATFORM CHANGES =================
1044 * Preliminary support for macOS 10.13 has been added.
1046 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
1048 * mips: support for MIPS SIMD architecture (MSA) has been added.
1050 * mips: support for MIPS N32 ABI has been added.
1052 * s390: partial support for vector instructions (integer and string) has been
1055 * ==================== TOOL CHANGES ====================
1057 * Helgrind: Addition of a flag
1058 --delta-stacktrace=no|yes [yes on linux amd64/x86]
1059 which specifies how full history stack traces should be computed.
1060 Setting this to =yes can speed up Helgrind by 25% when using
1061 --history-level=full.
1063 * Memcheck: reduced false positive rate for optimised code created by Clang 6
1064 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
1065 blocks more carefully to determine where it can avoid expensive definedness
1066 checks without loss of precision. This is controlled by the flag
1067 --expensive-definedness-checks=no|auto|yes [auto].
1069 * ==================== OTHER CHANGES ====================
1071 * Valgrind is now buildable with link-time optimisation (LTO). A new
1072 configure option --enable-lto=yes allows building Valgrind with LTO. If the
1073 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
1074 Note that if you are doing Valgrind development, --enable-lto=yes massively
1075 slows down the build process.
1077 * ==================== FIXED BUGS ====================
1079 The following bugs have been fixed or resolved. Note that "n-i-bz"
1080 stands for "not in bugzilla" -- that is, a bug that was reported to us
1081 but never got a bugzilla entry. We encourage you to file bugs in
1082 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1083 than mailing the developers (or mailing lists) directly -- bugs that
1084 are not entered into bugzilla tend to get forgotten about or ignored.
1086 To see details of a given bug, visit
1087 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1088 where XXXXXX is the bug number as listed below.
1090 79362 Debug info is lost for .so files when they are dlclose'd
1091 208052 strlcpy error when n = 0
1092 255603 exp-sgcheck Assertion '!already_present' failed
1093 338252 building valgrind with -flto (link time optimisation) fails
1094 345763 MIPS N32 ABI support
1095 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
1096 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
1097 372347 Replacement problem of the additional c++14/c++17 new/delete operators
1098 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1099 376257 helgrind history full speed up using a cached stack
1100 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
1102 379748 Fix missing pselect syscall (OS X 10.11)
1103 379754 Fix missing syscall ulock_wait (OS X 10.12)
1104 380397 s390x: __GI_strcspn() replacemenet needed
1105 381162 possible array overrun in VEX register allocator
1106 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
1107 381274 powerpc too chatty even with --sigill-diagnostics=no
1108 381289 epoll_pwait can have a NULL sigmask
1109 381553 VEX register allocator v3
1110 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
1111 381769 Use ucontext_t instead of struct ucontext
1112 381805 arm32 needs ld.so index hardwire for new glibc security fixes
1113 382256 gz compiler flag test doesn't work for gold
1114 382407 vg_perf needs "--terse" command line option
1115 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
1116 382563 MIPS MSA ASE support
1117 382998 xml-socket doesn't work
1118 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
1119 383723 Fix missing kevent_qos syscall (macOS 10.11)
1120 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
1121 384096 Mention AddrCheck at Memcheck's command line option [..]
1122 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
1123 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
1124 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
1125 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
1126 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
1127 384337 performance improvements to VEX register allocator v2 and v3
1128 384526 reduce number of spill insns generated by VEX register allocator v3
1129 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
1130 384631 Sanitise client args as printed with -v
1131 384633 Add a simple progress-reporting facility
1132 384987 VEX regalloc: allocate caller-save registers for short lived vregs
1133 385055 PPC VEX temporary storage exhausted
1134 385182 PPC64 is missing support for the DSCR
1135 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
1136 385207 PPC64, generate_store_FPRF() generates too many Iops
1137 385208 PPC64, xxperm instruction exhausts temporary memory
1138 385210 PPC64, vpermr instruction could exhaust temporary memory
1139 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
1140 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
1141 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
1142 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
1143 385408 s390x: z13 vector "support" instructions not implemented
1144 385409 s390x: z13 vector integer instructions not implemented
1145 385410 s390x: z13 vector string instructions not implemented
1146 385412 s390x: new non-vector z13 instructions not implemented
1147 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
1148 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
1149 385939 Optionally exit on the first error
1150 386318 valgrind.org/info/tools.html is missing SGCheck
1151 386425 running valgrind + wine on armv7l gives illegal opcode
1152 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
1153 387410 MIPSr6 support
1154 387664 Memcheck: make expensive-definedness-checks be the default
1155 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
1156 387766 asm shifts cause false positive "Conditional jump or move depends
1157 on uninitialised value"
1158 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
1159 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
1160 388786 Support bpf syscall in amd64 Linux
1161 388862 Add replacements for wmemchr and wcsnlen on Linux
1162 389065 valgrind meets gcc flag -Wlogical-op
1163 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
1164 390471 suppression by specification of source-file line number
1165 390723 make xtree dump files world wide readable, similar to log files
1166 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
1167 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
1168 392118 unhandled amd64-linux syscall: 332 (statx)
1169 392449 callgrind not clearing the number of calls properly
1170 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
1171 lxssp, stxssp and stxvl instructions.
1172 393023 callgrind_control risks using the wrong vgdb
1173 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
1174 393099 posix_memalign() invalid write if alignment == 0
1175 393146 failing assert "is_DebugInfo_active(di)"
1176 395709 PPC64 is missing support for the xvnegsp instruction
1177 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
1179 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
1180 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
1181 396839 s390x: Trap instructions not implemented
1182 396887 arch_prctl should return EINVAL on unknown option
1183 == 397286 crash before launching binary (Unsupported arch_prctl option)
1184 == 397393 valgrind: the 'impossible' happened: (Archlinux)
1185 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
1186 396906 compile tests failure on mips32-linux: broken inline asm in tests on
1188 397012 glibc ld.so uses arch_prctl on i386
1189 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
1190 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
1191 397424 glibc 2.27 and gdb_server tests
1192 398028 Assertion `cfsi_fits` failing in simple C program
1193 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
1195 n-i-bz Fix missing workq_ops operations (macOS)
1196 n-i-bz fix bug in strspn replacement
1197 n-i-bz Add support for the Linux BLKFLSBUF ioctl
1198 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
1199 n-i-bz Fix possible stack trashing by semctl syscall wrapping
1200 n-i-bz Add support for the Linux membarrier() system call
1201 n-i-bz x86 front end: recognise and handle UD2 correctly
1202 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
1203 correctly aligned before entering the handler.
1205 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
1206 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
1207 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
1211 Release 3.13.0 (15 June 2017)
1212 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1214 3.13.0 is a feature release with many improvements and the usual collection of
1217 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1218 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1219 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1220 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
1222 * ==================== CORE CHANGES ===================
1224 * The translation cache size has been increased to keep up with the demands of
1225 large applications. The maximum number of sectors has increased from 24 to
1226 48. The default number of sectors has increased from 16 to 32 on all
1227 targets except Android, where the increase is from 6 to 12.
1229 * The amount of memory that Valgrind can use has been increased from 64GB to
1230 128GB. In particular this means your application can allocate up to about
1231 60GB when running on Memcheck.
1233 * Valgrind's default load address has been changed from 0x3800'0000 to
1234 0x5800'0000, so as to make it possible to load larger executables. This
1235 should make it possible to load executables of size at least 1200MB.
1237 * A massive spaceleak caused by reading compressed debuginfo files has been
1238 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
1241 * The C++ demangler has been updated.
1243 * Support for demangling Rust symbols has been added.
1245 * A new representation of stack traces, the "XTree", has been added. An XTree
1246 is a tree of stacktraces with data associated with the stacktraces. This is
1247 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
1248 consumption of your program. Reporting is controlled by the new options
1249 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
1251 A report can also be produced on demand using the gdbserver monitor command
1252 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
1253 format' and 'massif format. The existing visualisers for these formats (e.g.
1254 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
1255 analyse these reports.
1257 Memcheck can also produce XTree leak reports using the Callgrind file
1258 format. For more details, see the user manual.
1260 * ================== PLATFORM CHANGES =================
1262 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
1264 * amd64: fixes for JIT failure problems on long AVX2 code blocks
1266 * amd64 and x86: support for CET prefixes has been added
1268 * arm32: a few missing ARMv8 instructions have been implemented
1270 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
1271 Store-Conditional instructions has been added. This is to deal with
1272 processor implementations that implement the LL/SC specifications strictly
1273 and as a result cause Valgrind to hang in certain situations. The
1274 alternative implementation is automatically enabled at startup, as required.
1275 You can use the option --sim-hints=fallback-llsc to force-enable it if you
1278 * Support for OSX 10.12 has been improved.
1280 * On Linux, clone handling has been improved to honour CLONE_VFORK that
1281 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
1282 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
1283 CLONE_VM exact semantics will (still) not work.
1285 * The TileGX/Linux port has been removed because it appears to be both unused
1288 * ==================== TOOL CHANGES ====================
1292 - Memcheck should give fewer false positives when running optimised
1293 Clang/LLVM generated code.
1295 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1297 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
1298 to produce the end of execution leak report in a xtree callgrind format
1301 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
1302 the leak report in an xtree file.
1306 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1308 - For some workloads (typically, for big applications), Massif memory
1309 consumption and CPU consumption has decreased significantly.
1313 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1315 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
1316 for Ada gnat compiled applications.
1318 * ==================== OTHER CHANGES ====================
1320 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
1321 append the inner guest stacktrace to the inner host stacktrace. This helps
1322 to investigate the errors reported by the outer, when they are caused by the
1323 inner guest program (such as an inner regtest). See README_DEVELOPERS for
1326 * To allow fast detection of callgrind files by desktop environments and file
1327 managers, the format was extended to have an optional first line that
1328 uniquely identifies the format ("# callgrind format"). Callgrind creates
1329 this line now, as does the new xtree functionality.
1331 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
1332 have a new %n format letter that is replaced by a sequence number.
1334 * "--version -v" now shows the SVN revision numbers from which Valgrind was
1337 * ==================== FIXED BUGS ====================
1339 The following bugs have been fixed or resolved. Note that "n-i-bz"
1340 stands for "not in bugzilla" -- that is, a bug that was reported to us
1341 but never got a bugzilla entry. We encourage you to file bugs in
1342 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1343 than mailing the developers (or mailing lists) directly -- bugs that
1344 are not entered into bugzilla tend to get forgotten about or ignored.
1346 To see details of a given bug, visit
1347 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1348 where XXXXXX is the bug number as listed below.
1350 162848 --log-file output isn't split when a program forks
1351 340777 Illegal instruction on mips (ar71xx)
1352 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
1353 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
1354 to a different stack.
1355 344139 x86 stack-seg overrides, needed by the Wine people
1356 344524 store conditional of guest applications always fail - observed on
1358 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
1359 352395 Please provide SVN revision info in --version -v
1360 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
1361 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
1362 INVALID_POSIX_THREADID' failed
1363 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
1364 barrier implementation
1365 358697 valgrind.h: Some code remains even when defining NVALGRIND
1366 359202 Add musl libc configure/compile
1367 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
1368 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
1369 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
1370 362223 assertion failed when .valgrindrc is a directory instead of a file
1371 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
1372 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
1373 368507 can't malloc chunks larger than about 34GB
1374 368529 Android arm target link error, missing atexit and pthread_atfork
1375 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
1376 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
1377 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
1378 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
1379 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
1380 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
1381 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
1382 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
1383 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
1384 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
1385 370028 Reduce the number of compiler warnings on MIPS platforms
1386 370635 arm64 missing syscall getcpu
1387 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
1388 371227 Clean AArch64 syscall table
1389 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
1390 371471 Valgrind complains about non legit memory leaks on placement new (C++)
1391 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
1392 371503 disInstr(arm64): unhandled instruction 0xF89F0000
1393 371869 support '%' in symbol Z-encoding
1394 371916 execution tree xtree concept
1395 372120 c++ demangler demangles symbols which are not c++
1396 372185 Support of valgrind on ARMv8 with 32 bit executable
1397 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1398 372195 Power PC, xxsel instruction is not always recognized.
1399 372504 Hanging on exit_group
1400 372600 process loops forever when fatal signals are arriving quickly
1401 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1402 373046 Stacks registered by core are never deregistered
1403 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1404 373086 Implement additional Xen hypercalls
1405 373192 Calling posix_spawn in glibc 2.24 completely broken
1406 373488 Support for fanotify API on ARM64 architecture
1407 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1408 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1409 373938 const IRExpr arguments for matchIRExpr()
1410 374719 some spelling fixes
1411 374963 increase valgrind's load address to prevent mmap failure
1412 375514 valgrind_get_tls_addr() does not work in case of static TLS
1413 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1414 for ML_(find_rx_mapping)()
1415 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1416 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1417 == 377159 "vex: the `impossible' happened" still present
1418 == 375150 Assertion 'tres.status == VexTransOK' failed
1419 == 378068 valgrind crashes on AVX2 function in FFmpeg
1420 376142 Segfaults on MIPS Cavium Octeon boards
1421 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1422 376455 Solaris: unhandled syscall lgrpsys(180)
1423 376518 Solaris: unhandled fast trap getlgrp(6)
1424 376611 ppc64 and arm64 don't know about prlimit64 syscall
1425 376729 PPC64, remove R2 from the clobber list
1427 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1428 to be wrongly marked as addressable
1429 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1430 PIE enabled by default
1431 377376 memcheck/tests/linux/getregset fails with glibc2.24
1432 377427 PPC64, lxv instruction failing on odd destination register
1433 377478 PPC64: ISA 3.0 setup fixes
1434 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1435 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1436 and 2 args for FUTEX_TRYLOCK_PI
1437 377717 Fix massive space leak when reading compressed debuginfo sections
1438 377891 Update Xen 4.6 domctl wrappers
1439 377930 fcntl syscall wrapper is missing flock structure check
1440 378524 libvexmultiarch_test regression on s390x and ppc64
1441 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1442 378673 Update libiberty demangler
1443 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1444 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1445 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1446 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1447 (task_register_dyld_image_infos)
1448 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1449 (task_register_dyld_shared_cache_image_info)
1450 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1451 379473 MIPS: add support for rdhwr cycle counter register
1452 379504 remove TileGX/Linux port
1453 379525 Support more x86 nop opcodes
1454 379838 disAMode(x86): not an addr!
1455 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1456 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1457 379895 clock_gettime does not execute POST syscall wrapper
1458 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1459 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1460 380200 xtree generated callgrind files refer to files without directory name
1461 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1462 380397 s390x: __GI_strcspn() replacement needed
1463 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1465 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1466 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1467 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1471 Release 3.12.0 (20 October 2016)
1472 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1474 3.12.0 is a feature release with many improvements and the usual
1475 collection of bug fixes.
1477 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1478 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1479 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1480 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1481 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1482 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1484 * ================== PLATFORM CHANGES =================
1486 * POWER: Support for ISA 3.0 has been added
1488 * mips: support for O32 FPXX ABI has been added.
1489 * mips: improved recognition of different processors
1490 * mips: determination of page size now done at run time
1492 * amd64: Partial support for AMD FMA4 instructions.
1494 * arm, arm64: Support for v8 crypto and CRC instructions.
1496 * Improvements and robustification of the Solaris port.
1498 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1500 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1501 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1502 where possible. Valgrind's support for 32-bit x86 has stagnated in
1503 recent years and has fallen far behind that for 64-bit x86
1504 instructions. By contrast 64-bit x86 is well supported, up to and
1507 * ==================== TOOL CHANGES ====================
1511 - Added meta mempool support for describing a custom allocator which:
1512 - Auto-frees all chunks assuming that destroying a pool destroys all
1514 - Uses itself to allocate other memory blocks
1516 - New flag --ignore-range-below-sp to ignore memory accesses below
1517 the stack pointer, if you really have to. The related flag
1518 --workaround-gcc296-bugs=yes is now deprecated. Use
1519 --ignore-range-below-sp=1024-1 as a replacement.
1523 - Improved thread startup time significantly on non-Linux platforms.
1527 - Added collection of the metric "tot-blocks-allocd"
1529 * ==================== OTHER CHANGES ====================
1531 * Replacement/wrapping of malloc/new related functions is now done not just
1532 for system libraries by default, but for any globally defined malloc/new
1533 related function (both in shared libraries and statically linked alternative
1534 malloc implementations). The dynamic (runtime) linker is excluded, though.
1535 To only intercept malloc/new related functions in
1536 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1537 "nouserintercepts" can be any non-existing library name).
1538 This new functionality is not implemented for MacOS X.
1540 * The maximum number of callers in a suppression entry is now equal to
1541 the maximum size for --num-callers (500).
1542 Note that --gen-suppressions=yes|all similarly generates suppressions
1543 containing up to --num-callers frames.
1545 * New and modified GDB server monitor features:
1547 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1548 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1551 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1552 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1555 * Valgrind is able to read compressed debuginfo sections in two formats:
1556 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1557 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1559 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1560 for the most common use case (x86_64-linux, Memcheck) has been
1563 * Improved performance for programs that do a lot of discarding of
1564 instruction address ranges of 8KB or less.
1566 * The C++ symbol demangler has been updated.
1568 * More robustness against invalid syscall parameters on Linux.
1570 * ==================== FIXED BUGS ====================
1572 The following bugs have been fixed or resolved. Note that "n-i-bz"
1573 stands for "not in bugzilla" -- that is, a bug that was reported to us
1574 but never got a bugzilla entry. We encourage you to file bugs in
1575 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1576 than mailing the developers (or mailing lists) directly -- bugs that
1577 are not entered into bugzilla tend to get forgotten about or ignored.
1579 To see details of a given bug, visit
1580 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1581 where XXXXXX is the bug number as listed below.
1583 191069 Exiting due to signal not reported in XML output
1584 199468 Suppressions: stack size limited to 25
1585 while --num-callers allows more frames
1586 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1587 278744 cvtps2pd with redundant RexW
1588 303877 valgrind doesn't support compressed debuginfo sections.
1589 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1590 348345 Assertion fails for negative lineno
1591 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1592 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1593 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1594 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1595 352197 mips: mmap2() not wrapped correctly for page size > 4096
1596 353083 arm64 doesn't implement various xattr system calls
1597 353084 arm64 doesn't support sigpending system call
1598 353137 www: update info for Supported Platforms
1599 353138 www: update "The Valgrind Developers" page
1600 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1603 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1604 353398 WARNING: unhandled amd64-solaris syscall: 207
1605 353660 XML in auxwhat tag not escaping reserved symbols properly
1606 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1607 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1608 353802 ELF debug info reader confused with multiple .rodata sections
1609 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1610 353917 unhandled amd64-solaris syscall fchdir(120)
1611 353920 unhandled amd64-solaris syscall: 170
1612 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1613 354392 unhandled amd64-solaris syscall: 171
1614 354797 Vbit test does not include Iops for Power 8 instruction support
1615 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1619 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1620 355188 valgrind should intercept all malloc related global functions
1621 355454 do not intercept malloc related symbols from the runtime linker
1622 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1623 356044 Dwarf line info reader misinterprets is_stmt register
1624 356112 mips: replace addi with addiu
1625 356393 valgrind (vex) crashes because isZeroU happened
1628 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1629 356678 arm64-linux: unhandled syscall 232 (mincore)
1630 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1631 356823 Unsupported ARM instruction: stlex
1632 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1633 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1634 357673 crash if I try to run valgrind with a binary link with libcurl
1635 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1636 357871 pthread_spin_destroy not properly wrapped
1637 357887 Calls to VG_(fclose) do not close the file descriptor
1638 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1639 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1640 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1641 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1642 359181 Buffer Overflow during Demangling
1643 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1644 359289 s390x: popcnt (B9E1) not implemented
1645 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1646 359503 Add missing syscalls for aarch64 (arm64)
1647 359645 "You need libc6-dbg" help message could be more helpful
1648 359703 s390: wire up separate socketcalls system calls
1649 359724 getsockname might crash - deref_UInt should call safe_to_deref
1650 359733 amd64 implement ld.so strchr/index override like x86
1651 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1652 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1654 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1655 359871 Incorrect mask handling in ppoll
1656 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1657 360008 Contents of Power vr registers contents is not printed correctly when
1658 the --vgdb-shadow-registers=yes option is used
1659 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1660 non-zero shadow bits
1661 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1662 360425 arm64 unsupported instruction ldpsw
1664 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1665 360571 Error about the Android Runtime reading below the stack pointer on ARM
1666 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1667 360749 kludge for multiple .rodata sections on Solaris no longer needed
1668 360752 raise the number of reserved fds in m_main.c from 10 to 12
1669 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1670 361226 s390x: risbgn (EC59) not implemented
1671 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1672 361354 ppc64[le]: wire up separate socketcalls system calls
1673 361615 Inconsistent termination for multithreaded process terminated by signal
1674 361926 Unhandled Solaris syscall: sysfs(84)
1675 362009 V dumps core on unimplemented functionality before threads are created
1676 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1677 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1678 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1679 362953 Request for an update to the Valgrind Developers page
1680 363680 add renameat2() support
1681 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1682 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1683 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1684 364058 clarify in manual limitations of array overruns detections
1685 364413 pselect sycallwrapper mishandles NULL sigmask
1686 364728 Power PC, missing support for several HW registers in
1687 get_otrack_shadow_offset_wrk()
1688 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1689 365273 Invalid write to stack location reported after signal handler runs
1690 365912 ppc64BE segfault during jm-insns test (RELRO)
1691 366079 FPXX Support for MIPS32 Valgrind
1692 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1693 366344 Multiple unhandled instruction for Aarch64
1694 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1695 367995 Integration of memcheck with custom memory allocator
1696 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1697 368412 False positive result for altivec capability check
1698 368416 Add tc06_two_races_xml.exp output for ppc64
1699 368419 Perf Events ioctls not implemented
1700 368461 mmapunmap test fails on ppc64
1701 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1702 369000 AMD64 fma4 instructions unsupported.
1703 369169 ppc64 fails jm_int_isa_2_07 test
1704 369175 jm_vec_isa_2_07 test crashes on ppc64
1705 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1706 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1707 369359 msghdr_foreachfield can crash when handling bad iovec
1708 369360 Bad sigprocmask old or new sets can crash valgrind
1709 369361 vmsplice syscall wrapper crashes on bad iovec
1710 369362 Bad sigaction arguments crash valgrind
1711 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1712 369402 Bad set/get_thread_area pointer crashes valgrind
1713 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1714 369446 valgrind crashes on unknown fcntl command
1715 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1716 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1717 370265 ISA 3.0 HW cap stuff needs updating
1718 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1719 372195 Power PC, xxsel instruction is not always recognized
1721 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1722 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1723 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1724 n-i-bz Fix ppoll_alarm exclusion on OS X
1725 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1726 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1727 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1728 n-i-bz amd64: memcheck false positive with shr %edx
1729 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1730 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1731 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1732 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1734 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1735 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1736 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1740 Release 3.11.0 (22 September 2015)
1741 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1743 3.11.0 is a feature release with many improvements and the usual
1744 collection of bug fixes.
1746 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1747 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1748 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1749 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1750 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1751 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1753 * ================== PLATFORM CHANGES =================
1755 * Support for Solaris/x86 and Solaris/amd64 has been added.
1757 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1759 * Preliminary support for the Tilera TileGX architecture has been added.
1761 * s390x: It is now required for the host to have the "long displacement"
1762 facility. The oldest supported machine model is z990.
1764 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1765 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1767 * The JIT's register allocator is significantly faster, making the JIT
1768 as a whole somewhat faster, so JIT-intensive activities, for example
1769 program startup, are modestly faster, around 5%.
1771 * There have been changes to the default settings of several command
1772 line flags, as detailed below.
1774 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1775 capable hosts, the simulated CPUID will now indicate AVX2 support.
1777 * ==================== TOOL CHANGES ====================
1781 - The default value for --leak-check-heuristics has been changed from
1782 "none" to "all". This helps to reduce the number of possibly
1783 lost blocks, in particular for C++ applications.
1785 - The default value for --keep-stacktraces has been changed from
1786 "malloc-then-free" to "malloc-and-free". This has a small cost in
1787 memory (one word per malloc-ed block) but allows Memcheck to show the
1788 3 stacktraces of a dangling reference: where the block was allocated,
1789 where it was freed, and where it is acccessed after being freed.
1791 - The default value for --partial-loads-ok has been changed from "no" to
1792 "yes", so as to avoid false positive errors resulting from some kinds
1793 of vectorised loops.
1795 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1796 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1797 than get_vbits when you need to associate byte data value with
1798 their corresponding validity bits.
1800 - The 'block_list' monitor command has been enhanced:
1801 o it can print a range of loss records
1802 o it now accepts an optional argument 'limited <max_blocks>'
1803 to control the number of blocks printed.
1804 o if a block has been found using a heuristic, then
1805 'block_list' now shows the heuristic after the block size.
1806 o the loss records/blocks to print can be limited to the blocks
1807 found via specified heuristics.
1809 - The C helper functions used to instrument loads on
1810 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1811 replaced by handwritten assembly sequences. This gives speedups
1812 in the region of 0% to 7% for those targets only.
1814 - A new command line option, --expensive-definedness-checks=yes|no,
1815 has been added. This is useful for avoiding occasional invalid
1816 uninitialised-value errors in optimised code. Watch out for
1817 runtime degradation, as this can be up to 25%. As always, though,
1818 the slowdown is highly application specific. The default setting
1823 - A new monitor command 'all_snapshots <filename>' dumps all
1824 snapshots taken so far.
1828 - Significant memory reduction and moderate speedups for
1829 --history-level=full for applications accessing a lot of memory
1830 with many different stacktraces.
1832 - The default value for --conflict-cache-size=N has been doubled to
1833 2000000. Users that were not using the default value should
1834 preferably also double the value they give.
1836 The default was changed due to the changes in the "full history"
1837 implementation. Doubling the value gives on average a slightly more
1838 complete history and uses similar memory (or significantly less memory
1839 in the worst case) than the previous implementation.
1841 - The Helgrind monitor command 'info locks' now accepts an optional
1842 argument 'lock_addr', which shows information about the lock at the
1845 - When using --history-level=full, the new Helgrind monitor command
1846 'accesshistory <addr> [<len>]' will show the recorded accesses for
1847 <len> (or 1) bytes at <addr>.
1849 * ==================== OTHER CHANGES ====================
1851 * The default value for the --smc-check option has been changed from
1852 "stack" to "all-non-file" on targets that provide automatic D-I
1853 cache coherence (x86, amd64 and s390x). The result is to provide,
1854 by default, transparent support for JIT generated and self-modifying
1855 code on all targets.
1857 * Mac OS X only: the default value for the --dsymutil option has been
1858 changed from "no" to "yes", since any serious usage on Mac OS X
1859 always required it to be "yes".
1861 * The command line options --db-attach and --db-command have been removed.
1862 They were deprecated in 3.10.0.
1864 * When a process dies due to a signal, Valgrind now shows the signal
1865 and the stacktrace at default verbosity (i.e. verbosity 1).
1867 * The address description logic used by Memcheck and Helgrind now
1868 describes addresses in anonymous segments, file mmap-ed segments,
1869 shared memory segments and the brk data segment.
1871 * The new option --error-markers=<begin>,<end> can be used to mark the
1872 begin/end of errors in textual output mode, to facilitate
1873 searching/extracting errors in output files that mix valgrind errors
1874 with program output.
1876 * The new option --max-threads=<number> can be used to change the number
1877 of threads valgrind can handle. The default is 500 threads which
1878 should be more than enough for most applications.
1880 * The new option --valgrind-stacksize=<number> can be used to change the
1881 size of the private thread stacks used by Valgrind. This is useful
1882 for reducing memory use or increasing the stack size if Valgrind
1883 segfaults due to stack overflow.
1885 * The new option --avg-transtab-entry-size=<number> can be used to specify
1886 the expected instrumented block size, either to reduce memory use or
1887 to avoid excessive retranslation.
1889 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1891 * New and modified GDB server monitor features:
1893 - When a signal is reported in GDB, you can now use the GDB convenience
1894 variable $_siginfo to examine detailed signal information.
1896 - Valgrind's gdbserver now allows the user to change the signal
1897 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1898 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1899 continue without passing the signal to the process.
1901 - With GDB >= 7.10, the command 'target remote'
1902 will automatically load the executable file of the process running
1903 under Valgrind. This means you do not need to specify the executable
1904 file yourself, GDB will discover it itself. See GDB documentation about
1905 'qXfer:exec-file:read' packet for more info.
1907 * ==================== FIXED BUGS ====================
1909 The following bugs have been fixed or resolved. Note that "n-i-bz"
1910 stands for "not in bugzilla" -- that is, a bug that was reported to us
1911 but never got a bugzilla entry. We encourage you to file bugs in
1912 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1913 than mailing the developers (or mailing lists) directly -- bugs that
1914 are not entered into bugzilla tend to get forgotten about or ignored.
1916 To see details of a given bug, visit
1917 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1918 where XXXXXX is the bug number as listed below.
1920 116002 VG_(printf): Problems with justification of strings and integers
1921 155125 avoid cutting away file:lineno after long function name
1922 197259 Unsupported arch_prtctl PR_SET_GS option
1923 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
1924 201216 Fix Valgrind does not support pthread_sigmask() on OS X
1925 201435 Fix Darwin: -v does not show kernel version
1926 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
1927 211256 Fixed an outdated comment regarding the default platform.
1928 211529 Incomplete call stacks for code compiled by newer versions of MSVC
1929 211926 Avoid compilation warnings in valgrind.h with -pedantic
1930 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
1932 226609 Crediting upstream authors in man page
1933 231257 Valgrind omits path when executing script from shebang line
1934 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
1935 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
1936 269360 s390x: Fix addressing mode selection for compare-and-swap
1937 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
1939 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
1940 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
1941 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
1942 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
1943 330147 libmpiwrap PMPI_Get_count returns undefined value
1944 333051 mmap of huge pages fails due to incorrect alignment
1946 334802 valgrind does not always explain why a given option is bad
1947 335618 mov.w rN, pc/sp (ARM32)
1948 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
1953 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
1954 338602 AVX2 bit in CPUID missing
1955 338606 Strange message for scripts with invalid interpreter
1956 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
1957 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
1958 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
1960 339156 gdbsrv not called for fatal signal
1961 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
1962 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
1963 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
1964 339442 Fix testsuite build failure on OS X 10.9
1965 339542 Enable compilation with Intel's ICC compiler
1966 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
1967 339688 Mac-specific ASM does not support .version directive (cpuid,
1968 tronical and pushfpopf tests)
1969 339745 Valgrind crash when check Marmalade app (partial fix)
1970 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
1971 339778 Linux/TileGx platform support for Valgrind
1972 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
1973 339789 Fix none/tests/execve test on Mac OS X 10.9
1974 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
1975 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
1976 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
1977 340392 Allow user to select more accurate definedness checking in memcheck
1978 to avoid invalid complaints on optimised code
1979 340430 Fix some grammatical weirdness in the manual.
1980 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
1981 341419 Signal handler ucontext_t not filled out correctly on OS X
1982 341539 VG_(describe_addr) should not describe address as belonging to client
1983 segment if it is past the heap end
1984 341613 Enable building of manythreads and thread-exits tests on Mac OS X
1985 341615 Fix none/tests/darwin/access_extended test on Mac OS X
1986 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
1987 341789 aarch64: shmat fails with valgrind on ARMv8
1988 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
1989 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
1990 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
1991 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
1992 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
1993 342221 socket connect false positive uninit memory for unknown af family
1994 342353 Allow dumping full massif output while valgrind is still running
1995 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
1999 342603 Add I2C_SMBUS ioctl support
2000 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
2001 342683 Mark memory past the initial brk limit as unaddressable
2002 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
2003 342795 Internal glibc __GI_mempcpy call should be intercepted
2004 342841 s390x: Support instructions fiebr(a) and fidbr(a)
2005 343012 Unhandled syscall 319 (memfd_create)
2006 343069 Patch updating v4l2 API support
2007 343173 helgrind crash during stack unwind
2008 343219 fix GET_STARTREGS for arm
2009 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
2010 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
2011 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
2012 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
2013 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
2014 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
2015 343597 ppc64le: incorrect use of offseof macro
2016 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
2017 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
2018 343732 Unhandled syscall 144 (setgid) on aarch64
2019 343733 Unhandled syscall 187 (msgctl and related) on aarch64
2020 343802 s390x: False positive "conditional jump or move depends on [..]
2021 343902 --vgdb=yes doesn't break when --xml=yes is used
2022 343967 Don't warn about setuid/setgid/setcap executable for directories
2023 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
2024 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
2025 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
2026 344054 www - update info for Solaris/illumos
2027 344416 'make regtest' does not work cleanly on OS X
2028 344235 Remove duplicate include of pub_core_aspacemgr.h
2029 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
2030 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
2031 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
2032 344314 callgrind_annotate ... warnings about commands containing newlines
2033 344318 socketcall should wrap recvmmsg and sendmmsg
2034 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
2035 344416 Fix 'make regtest' does not work cleanly on OS X
2036 344499 Fix compilation for Linux kernel >= 4.0.0
2037 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
2038 unix:349 (__pthread_fchdir)
2039 344559 Garbage collection of unused segment names in address space manager
2040 344560 Fix stack traces missing penultimate frame on OS X
2041 344621 Fix memcheck/tests/err_disable4 test on OS X
2042 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
2043 344702 Fix missing libobjc suppressions on OS X 10.10
2045 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
2046 344939 Fix memcheck/tests/xml1 on OS X 10.10
2047 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
2048 345079 Fix build problems in VEX/useful/test_main.c
2049 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
2050 345177 arm64: prfm (reg) not implemented
2051 345215 Performance improvements for the register allocator
2052 345248 add support for Solaris OS in valgrind
2053 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
2054 345394 Fix memcheck/tests/strchr on OS X
2055 345637 Fix memcheck/tests/sendmsg on OS X
2056 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
2057 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
2058 345887 Fix an assertion in the address space manager
2059 345928 amd64: callstack only contains current function for small stacks
2060 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
2061 345987 MIPS64: Implement cavium LHX instruction
2062 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
2063 346185 Fix typo saving altivec register v24
2064 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
2065 and LibVEX_GuestPPC64_get_CR()
2066 346270 Regression tests none/tests/jm_vec/isa_2_07 and
2067 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
2068 346307 fuse filesystem syscall deadlocks
2069 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
2070 346411 MIPS: SysRes::_valEx handling is incorrect
2071 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
2072 346474 PPC64 Power 8, spr TEXASRU register not supported
2073 346487 Compiler generates "note" about a future ABI change for PPC64
2074 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
2075 and causing spurious "invalid read of size 8" warnings
2076 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
2077 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
2078 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
2079 347322 Power PC regression test cleanup
2080 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
2082 347389 unhandled syscall: 373 (Linux ARM syncfs)
2083 347686 Patch set to cleanup PPC64 regtests
2084 347978 Remove bash dependencies where not needed
2085 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
2086 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
2088 348102 Patch updating v4l2 API support
2089 348247 amd64 front end: jno jumps wrongly when overflow is not set
2090 348269 Improve mmap MAP_HUGETLB support.
2091 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
2092 348345 Assertion fails for negative lineno
2093 348377 Unsupported ARM instruction: yield
2094 348565 Fix detection of command line option availability for clang
2095 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
2096 348728 Fix broken check for VIDIOC_G_ENC_INDEX
2097 348748 Fix redundant condition
2098 348890 Fix clang warning about unsupported --param inline-unit-growth=900
2099 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
2100 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
2101 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
2102 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
2103 349626 Implemented additional Xen hypercalls
2104 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
2105 349790 Clean up of the hardware capability checking utilities.
2106 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
2107 349874 Fix typos in source code
2108 349879 memcheck: add handwritten assembly for helperc_LOADV*
2109 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
2110 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
2111 350202 Add limited param to 'monitor block_list'
2112 350290 s390x: Support instructions fixbr(a)
2113 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
2114 350809 Fix none/tests/async-sigs for Solaris
2115 350811 Remove reference to --db-attach which has been removed.
2116 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
2117 350854 hard-to-understand code in VG_(load_ELF)()
2118 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2119 351386 Solaris: Cannot run ld.so.1 under Valgrind
2120 351474 Fix VG_(iseqsigset) as obvious
2121 351531 Typo in /include/vki/vki-xen-physdev.h header guard
2122 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
2123 351858 ldsoexec support on Solaris
2124 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
2125 352130 helgrind reports false races for printfs using mempcpy on FILE* state
2126 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
2127 352320 arm64 crash on none/tests/nestedfs
2128 352765 Vbit test fails on Power 6
2129 352768 The mbar instruction is missing from the Power PC support
2130 352769 Power PC program priority register (PPR) is not supported
2131 n-i-bz Provide implementations of certain compiler builtins to support
2132 compilers that may not provide those
2133 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
2134 n-i-bz Fix compilation on distros with glibc < 2.5
2135 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
2136 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
2137 n-i-bz mremap did not work properly on shared memory
2138 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
2139 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
2141 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
2142 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
2143 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
2147 Release 3.10.1 (25 November 2014)
2148 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2149 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
2150 and backports fixes for all reported missing AArch64 ARMv8 instructions
2151 and syscalls from the trunk. If you package or deliver 3.10.0 for others
2152 to use, you might want to consider upgrading to 3.10.1 instead.
2154 The following bugs have been fixed or resolved. Note that "n-i-bz"
2155 stands for "not in bugzilla" -- that is, a bug that was reported to us
2156 but never got a bugzilla entry. We encourage you to file bugs in
2157 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2158 than mailing the developers (or mailing lists) directly -- bugs that
2159 are not entered into bugzilla tend to get forgotten about or ignored.
2161 To see details of a given bug, visit
2162 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2163 where XXXXXX is the bug number as listed below.
2165 335440 arm64: ld1 (single structure) is not implemented
2166 335713 arm64: unhanded instruction: prfm (immediate)
2167 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
2168 339182 ppc64: AvSplat ought to load destination vector register with [..]
2169 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
2170 339433 ppc64 lxvw4x instruction uses four 32-byte loads
2171 339645 Use correct tag names in sys_getdents/64 wrappers
2172 339706 Fix false positive for ioctl(TIOCSIG) on linux
2173 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
2174 339853 arm64 times syscall unknown
2175 339855 arm64 unhandled getsid/setsid syscalls
2176 339858 arm64 dmb sy not implemented
2177 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
2178 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
2179 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
2181 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
2182 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
2183 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
2184 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
2185 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
2186 340509 arm64: unhandled instruction fcvtas
2187 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
2188 340632 arm64: unhandled instruction fcvtas
2189 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
2190 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
2191 340788 warning: unhandled syscall: 318 (getrandom)
2192 340807 disInstr(arm): unhandled instruction: 0xEE989B20
2193 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
2194 340922 arm64: unhandled getgroups/setgroups syscalls
2195 350251 Fix typo in VEX utility program (test_main.c).
2196 350407 arm64: unhandled instruction ucvtf (vector, integer)
2197 350809 none/tests/async-sigs breaks when run under cron on Solaris
2198 350811 update README.solaris after r15445
2199 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
2200 350854 strange code in VG_(load_ELF)()
2201 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2202 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
2203 n-i-bz Add missing ]] to terminate CDATA.
2204 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
2205 n-i-bz Enable sys_fadvise64_64 on arm32.
2206 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
2207 n-i-bz Add test cases for all known arm64 load/store instructions.
2208 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
2209 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
2210 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
2211 n-i-bz arm64: Implement "fcvtpu w, s".
2212 n-i-bz arm64: implement ADDP and various others
2213 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
2214 n-i-bz arm64: enable FCVT{A,N}S X,S.
2216 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
2220 Release 3.10.0 (10 September 2014)
2221 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2223 3.10.0 is a feature release with many improvements and the usual
2224 collection of bug fixes.
2226 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
2227 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
2228 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
2229 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
2230 significantly improved relative to the 3.9.0 release.
2232 * ================== PLATFORM CHANGES =================
2234 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
2235 is mostly complete, and is usable, but some SIMD instructions are as
2238 * Support for little-endian variant of the 64-bit POWER architecture.
2240 * Support for Android on MIPS32.
2242 * Support for 64bit FPU on MIPS32 platforms.
2244 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
2246 * Configuration for and running on Android targets has changed.
2247 See README.android in the source tree for details.
2249 * ================== DEPRECATED FEATURES =================
2251 * --db-attach is now deprecated and will be removed in the next
2252 valgrind feature release. The built-in GDB server capabilities are
2253 superior and should be used instead. Learn more here:
2254 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
2256 * ==================== TOOL CHANGES ====================
2260 - Client code can now selectively disable and re-enable reporting of
2261 invalid address errors in specific ranges using the new client
2262 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
2263 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
2265 - Leak checker: there is a new leak check heuristic called
2266 "length64". This is used to detect interior pointers pointing 8
2267 bytes inside a block, on the assumption that the first 8 bytes
2268 holds the value "block size - 8". This is used by
2269 sqlite3MemMalloc, for example.
2271 - Checking of system call parameters: if a syscall parameter
2272 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
2273 several fields not initialised, an error is now reported for each
2274 field. Previously, an error was reported only for the first
2275 uninitialised field.
2277 - Mismatched alloc/free checking: a new flag
2278 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
2279 such checks if necessary.
2283 - Improvements to error messages:
2285 o Race condition error message involving heap allocated blocks also
2286 show the thread number that allocated the raced-on block.
2288 o All locks referenced by an error message are now announced.
2289 Previously, some error messages only showed the lock addresses.
2291 o The message indicating where a lock was first observed now also
2292 describes the address/location of the lock.
2294 - Helgrind now understands the Ada task termination rules and
2295 creates a happens-before relationship between a terminated task
2296 and its master. This avoids some false positives and avoids a big
2297 memory leak when a lot of Ada tasks are created and terminated.
2298 The interceptions are only activated with forthcoming releases of
2299 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
2301 - A new GDB server monitor command "info locks" giving the list of
2302 locks, their location, and their status.
2306 - callgrind_control now supports the --vgdb-prefix argument,
2307 which is needed if valgrind was started with this same argument.
2309 * ==================== OTHER CHANGES ====================
2311 * Unwinding through inlined function calls. Stack unwinding can now
2312 make use of Dwarf3 inlined-unwind information if it is available.
2313 The practical effect is that inlined calls become visible in stack
2314 traces. The suppression matching machinery has been adjusted
2315 accordingly. This is controlled by the new option
2316 --read-inline-info=yes|no. Currently this is enabled by default
2317 only on Linux and Android targets and only for the tools Memcheck,
2320 * Valgrind can now read EXIDX unwind information on 32-bit ARM
2321 targets. If an object contains both CFI and EXIDX unwind
2322 information, Valgrind will prefer the CFI over the EXIDX. This
2323 facilitates unwinding through system libraries on arm-android
2326 * Address description logic has been improved and is now common
2327 between Memcheck and Helgrind, resulting in better address
2328 descriptions for some kinds of error messages.
2330 * Error messages about dubious arguments (eg, to malloc or calloc) are
2331 output like other errors. This means that they can be suppressed
2332 and they have a stack trace.
2334 * The C++ demangler has been updated for better C++11 support.
2336 * New and modified GDB server monitor features:
2338 - Thread local variables/storage (__thread) can now be displayed.
2340 - The GDB server monitor command "v.info location <address>"
2341 displays information about an address. The information produced
2342 depends on the tool and on the options given to valgrind.
2343 Possibly, the following are described: global variables, local
2344 (stack) variables, allocated or freed blocks, ...
2346 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
2347 ask the GDB server to stop at the start of program execution, at
2348 the end of the program execution and on Valgrind internal errors.
2350 - A new monitor command "v.info stats" shows various Valgrind core
2351 and tool statistics.
2353 - A new monitor command "v.set hostvisibility" allows the GDB server
2354 to provide access to Valgrind internal host status/memory.
2356 * A new option "--aspace-minaddr=<address>" can in some situations
2357 allow the use of more memory by decreasing the address above which
2358 Valgrind maps memory. It can also be used to solve address
2359 conflicts with system libraries by increasing the default value.
2360 See user manual for details.
2362 * The amount of memory used by Valgrind to store debug info (unwind
2363 info, line number information and symbol data) has been
2364 significantly reduced, even though Valgrind now reads more
2365 information in order to support unwinding of inlined function calls.
2367 * Dwarf3 handling with --read-var-info=yes has been improved:
2369 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
2371 - Code compiled with
2372 -ffunction-sections -fdata-sections -Wl,--gc-sections
2373 no longer causes assertion failures.
2375 * Improved checking for the --sim-hints= and --kernel-variant=
2376 options. Unknown strings are now detected and reported to the user
2379 * The semantics of stack start/end boundaries in the valgrind.h
2380 VALGRIND_STACK_REGISTER client request has been clarified and
2381 documented. The convention is that start and end are respectively
2382 the lowest and highest addressable bytes of the stack.
2384 * ==================== FIXED BUGS ====================
2386 The following bugs have been fixed or resolved. Note that "n-i-bz"
2387 stands for "not in bugzilla" -- that is, a bug that was reported to us
2388 but never got a bugzilla entry. We encourage you to file bugs in
2389 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2390 than mailing the developers (or mailing lists) directly -- bugs that
2391 are not entered into bugzilla tend to get forgotten about or ignored.
2393 To see details of a given bug, visit
2394 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2395 where XXXXXX is the bug number as listed below.
2397 175819 Support for ipv6 socket reporting with --track-fds
2398 232510 make distcheck fails
2399 249435 Analyzing wine programs with callgrind triggers a crash
2400 278972 support for inlined function calls in stacktraces and suppression
2402 291310 FXSAVE instruction marks memory as undefined on amd64
2403 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2404 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2405 315199 vgcore file for threaded app does not show which thread crashed
2406 315952 tun/tap ioctls are not supported
2407 323178 Unhandled instruction: PLDW register (ARM)
2408 323179 Unhandled instruction: PLDW immediate (ARM)
2409 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2410 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2411 325124 [MIPSEL] Compilation error
2412 325477 Phase 4 support for IBM Power ISA 2.07
2413 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2414 325628 Phase 5 support for IBM Power ISA 2.07
2415 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2416 325751 Missing the two privileged Power PC Transactional Memory Instructions
2417 325816 Phase 6 support for IBM Power ISA 2.07
2418 325856 Make SGCheck fail gracefully on unsupported platforms
2419 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2420 326436 DRD: False positive in libstdc++ std::list::push_back
2421 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2422 326462 Refactor vgdb to isolate invoker stuff into separate module
2423 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2424 326623 DRD: false positive conflict report in a field assignment
2425 326724 Valgrind does not compile on OSX 1.9 Mavericks
2426 326816 Intercept for __strncpy_sse2_unaligned missing?
2427 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2428 326983 Clear direction flag after tests on amd64.
2429 327212 Do not prepend the current directory to absolute path names.
2430 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2431 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2432 327284 s390x: Fix translation of the risbg instruction
2433 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2434 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2435 327916 DW_TAG_typedef may have no name
2436 327943 s390x: add a redirection for the 'index' function
2437 328100 XABORT not implemented
2438 328205 Implement additional Xen hypercalls
2439 328454 add support Backtraces with ARM unwind tables (EXIDX)
2440 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2441 328711 valgrind.1 manpage "memcheck options" section is badly generated
2442 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2443 329612 Incorrect handling of AT_BASE for image execution
2444 329694 clang warns about using uninitialized variable
2445 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2446 330228 mmap must align to VKI_SHMLBA on mips32
2447 330257 LLVM does not support `-mno-dynamic-no-pic` option
2448 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2449 330459 --track-fds=yes doesn't track eventfds
2450 330469 Add clock_adjtime syscall support
2451 330594 Missing sysalls on PowerPC / uClibc
2452 330622 Add test to regression suite for POWER instruction: dcbzl
2453 330939 Support for AMD's syscall instruction on x86
2455 330941 Typo in PRE(poll) syscall wrapper
2456 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2457 331254 Fix expected output for memcheck/tests/dw4
2458 331255 Fix race condition in test none/tests/coolo_sigaction
2459 331257 Fix type of jump buffer in test none/tests/faultstatus
2460 331305 configure uses bash specific syntax
2461 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2462 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2463 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2464 331829 Unexpected ioctl opcode sign extension
2465 331830 ppc64: WARNING: unhandled syscall: 96/97
2466 331839 drd/tests/sem_open specifies invalid semaphore name
2467 331847 outcome of drd/tests/thread_name is nondeterministic
2468 332037 Valgrind cannot handle Thumb "add pc, reg"
2469 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2470 consistency checks enabled
2471 332263 intercepts for pthread_rwlock_timedrdlock and
2472 pthread_rwlock_timedwrlock are incorrect
2473 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2475 332276 Implement additional Xen hypercalls
2476 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2477 332765 Fix ms_print to create temporary files in a proper directory
2478 333072 drd: Add semaphore annotations
2479 333145 Tests for missaligned PC+#imm access for arm
2480 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2481 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2482 333248 WARNING: unhandled syscall: unix:443
2483 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2484 333501 cachegrind: assertion: Cache set count is not a power of two.
2487 333666 Recognize MPX instructions and bnd prefix.
2488 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2489 333817 Valgrind reports the memory areas written to by the SG_IO
2491 334049 lzcnt fails silently (x86_32)
2492 334384 Valgrind does not have support Little Endian support for
2494 334585 recvmmsg unhandled (+patch) (arm)
2495 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2496 334727 Build fails with -Werror=format-security
2497 334788 clarify doc about --log-file initial program directory
2498 334834 PPC64 Little Endian support, patch 2
2499 334836 PPC64 Little Endian support, patch 3 testcase fixes
2500 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2501 335034 Unhandled ioctl: HCIGETDEVLIST
2502 335155 vgdb, fix error print statement.
2503 335262 arm64: movi 8bit version is not supported
2504 335263 arm64: dmb instruction is not implemented
2505 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2506 335496 arm64: sbc/abc instructions are not implemented
2507 335554 arm64: unhandled instruction: abs
2508 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2509 335735 arm64: unhandled instruction: cnt
2510 335736 arm64: unhandled instruction: uaddlv
2511 335848 arm64: unhandled instruction: {s,u}cvtf
2512 335902 arm64: unhandled instruction: sli
2513 335903 arm64: unhandled instruction: umull (vector)
2514 336055 arm64: unhandled instruction: mov (element)
2515 336062 arm64: unhandled instruction: shrn{,2}
2516 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2517 336189 arm64: unhandled Instruction: mvn
2518 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2519 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2520 336772 Make moans about unknown ioctls more informative
2521 336957 Add a section about the Solaris/illumos port on the webpage
2522 337094 ifunc wrapper is broken on ppc64
2523 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2524 337528 leak check heuristic for block prefixed by length as 64bit number
2525 337740 Implement additional Xen hypercalls
2526 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2527 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2528 337871 deprecate --db-attach
2529 338023 Add support for all V4L2/media ioctls
2530 338024 inlined functions are not shown if DW_AT_ranges is used
2531 338106 Add support for 'kcmp' syscall
2532 338115 DRD: computed conflict set differs from actual after fork
2533 338160 implement display of thread local storage in gdbsrv
2534 338205 configure.ac and check for -Wno-tautological-compare
2535 338300 coredumps are missing one byte of every segment
2536 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2537 338499 --sim-hints parsing broken due to wrong order in tokens
2538 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2539 338681 Unable to unwind through clone thread created on i386-linux
2540 338698 race condition between gdbsrv and vgdb on startup
2541 338703 helgrind on arm-linux gets false positives in dynamic loader
2542 338791 alt dwz files can be relative of debug/main file
2543 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2544 338932 build V-trunk with gcc-trunk
2545 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2546 345079 Fix build problems in VEX/useful/test_main.c
2547 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2548 n-i-bz s390x: Fix memory corruption for multithreaded applications
2549 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2550 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2551 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2552 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2553 n-i-bz Add command-line option --default-suppressions
2554 n-i-bz Add support for BLKDISCARDZEROES ioctl
2555 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2556 n-i-bz Add support for sys_pivot_root and sys_unshare
2558 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2559 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2560 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2564 Release 3.9.0 (31 October 2013)
2565 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2566 3.9.0 is a feature release with many improvements and the usual
2567 collection of bug fixes.
2569 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2570 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2571 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2572 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2574 * ================== PLATFORM CHANGES =================
2576 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2577 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2579 * Support for MIPS DSP ASE on MIPS32 platforms.
2581 * Support for s390x Decimal Floating Point instructions on hosts that
2582 have the DFP facility installed.
2584 * Support for POWER8 (Power ISA 2.07) instructions
2586 * Support for Intel AVX2 instructions. This is available only on 64
2589 * Initial support for Intel Transactional Synchronization Extensions,
2592 * Initial support for Hardware Transactional Memory on POWER.
2594 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2595 run large GUI apps tolerably well.
2597 * ==================== TOOL CHANGES ====================
2601 - Improvements in handling of vectorised code, leading to
2602 significantly fewer false error reports. You need to use the flag
2603 --partial-loads-ok=yes to get the benefits of these changes.
2605 - Better control over the leak checker. It is now possible to
2606 specify which leak kinds (definite/indirect/possible/reachable)
2607 should be displayed, which should be regarded as errors, and which
2608 should be suppressed by a given leak suppression. This is done
2609 using the options --show-leak-kinds=kind1,kind2,..,
2610 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2611 "match-leak-kinds:" line in suppression entries, respectively.
2613 Note that generated leak suppressions contain this new line and
2614 are therefore more specific than in previous releases. To get the
2615 same behaviour as previous releases, remove the "match-leak-kinds:"
2616 line from generated suppressions before using them.
2618 - Reduced "possible leak" reports from the leak checker by the use
2619 of better heuristics. The available heuristics provide detection
2620 of valid interior pointers to std::stdstring, to new[] allocated
2621 arrays with elements having destructors and to interior pointers
2622 pointing to an inner part of a C++ object using multiple
2623 inheritance. They can be selected individually using the
2624 option --leak-check-heuristics=heur1,heur2,...
2626 - Better control of stacktrace acquisition for heap-allocated
2627 blocks. Using the --keep-stacktraces option, it is possible to
2628 control independently whether a stack trace is acquired for each
2629 allocation and deallocation. This can be used to create better
2630 "use after free" errors or to decrease Valgrind's resource
2631 consumption by recording less information.
2633 - Better reporting of leak suppression usage. The list of used
2634 suppressions (shown when the -v option is given) now shows, for
2635 each leak suppressions, how many blocks and bytes it suppressed
2636 during the last leak search.
2640 - False errors resulting from the use of statically initialised
2641 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2644 - False errors resulting from the use of pthread_cond_waits that
2645 timeout, have been removed.
2647 * ==================== OTHER CHANGES ====================
2649 * Some attempt to tune Valgrind's space requirements to the expected
2650 capabilities of the target:
2652 - The default size of the translation cache has been reduced from 8
2653 sectors to 6 on Android platforms, since each sector occupies
2654 about 40MB when using Memcheck.
2656 - The default size of the translation cache has been increased to 16
2657 sectors on all other platforms, reflecting the fact that large
2658 applications require instrumentation and storage of huge amounts
2659 of code. For similar reasons, the number of memory mapped
2660 segments that can be tracked has been increased by a factor of 6.
2662 - In all cases, the maximum number of sectors in the translation
2663 cache can be controlled by the new flag --num-transtab-sectors.
2665 * Changes in how debug info (line numbers, etc) is read:
2667 - Valgrind no longer temporarily mmaps the entire object to read
2668 from it. Instead, reading is done through a small fixed sized
2669 buffer. This avoids virtual memory usage spikes when Valgrind
2670 reads debuginfo from large shared objects.
2672 - A new experimental remote debug info server. Valgrind can read
2673 debug info from a different machine (typically, a build host)
2674 where debuginfo objects are stored. This can save a lot of time
2675 and hassle when running Valgrind on resource-constrained targets
2676 (phones, tablets) when the full debuginfo objects are stored
2677 somewhere else. This is enabled by the --debuginfo-server=
2680 - Consistency checking between main and debug objects can be
2681 disabled using the --allow-mismatched-debuginfo option.
2683 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2684 scanning can recover stack traces in some cases when the normal
2685 unwind mechanisms fail. Stack scanning is best described as "a
2686 nasty, dangerous and misleading hack" and so is disabled by default.
2687 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2690 * Detection and merging of recursive stack frame cycles. When your
2691 program has recursive algorithms, this limits the memory used by
2692 Valgrind for recorded stack traces and avoids recording
2693 uninteresting repeated calls. This is controlled by the command
2694 line option --merge-recursive-frame and by the monitor command
2695 "v.set merge-recursive-frames".
2697 * File name and line numbers for used suppressions. The list of used
2698 suppressions (shown when the -v option is given) now shows, for each
2699 used suppression, the file name and line number where the suppression
2702 * New and modified GDB server monitor features:
2704 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2705 that can be used to execute gdbserver monitor commands from the
2708 - A new monitor command, "v.info open_fds", that gives the list of
2709 open file descriptors and additional details.
2711 - An optional message in the "v.info n_errs_found" monitor command,
2712 for example "v.info n_errs_found test 1234 finished", allowing a
2713 comment string to be added to the process output, perhaps for the
2714 purpose of separating errors of different tests or test phases.
2716 - A new monitor command "v.info execontext" that shows information
2717 about the stack traces recorded by Valgrind.
2719 - A new monitor command "v.do expensive_sanity_check_general" to run
2720 some internal consistency checks.
2722 * New flag --sigill-diagnostics to control whether a diagnostic
2723 message is printed when the JIT encounters an instruction it can't
2724 translate. The actual behavior -- delivery of SIGILL to the
2725 application -- is unchanged.
2727 * The maximum amount of memory that Valgrind can use on 64 bit targets
2728 has been increased from 32GB to 64GB. This should make it possible
2729 to run applications on Memcheck that natively require up to about 35GB.
2731 * ==================== FIXED BUGS ====================
2733 The following bugs have been fixed or resolved. Note that "n-i-bz"
2734 stands for "not in bugzilla" -- that is, a bug that was reported to us
2735 but never got a bugzilla entry. We encourage you to file bugs in
2736 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2737 than mailing the developers (or mailing lists) directly -- bugs that
2738 are not entered into bugzilla tend to get forgotten about or ignored.
2740 To see details of a given bug, visit
2741 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2742 where XXXXXX is the bug number as listed below.
2744 123837 system call: 4th argument is optional, depending on cmd
2745 135425 memcheck should tell you where Freed blocks were Mallocd
2746 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2747 207815 Adds some of the drm ioctls to syswrap-linux.c
2748 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2749 252955 Impossible to compile with ccache
2750 253519 Memcheck reports auxv pointer accesses as invalid reads.
2751 263034 Crash when loading some PPC64 binaries
2752 269599 Increase deepest backtrace
2753 274695 s390x: Support "compare to/from logical" instructions (z196)
2754 275800 s390x: Autodetect cache info (part 2)
2755 280271 Valgrind reports possible memory leaks on still-reachable std::string
2756 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2757 289578 Backtraces with ARM unwind tables (stack scan flags)
2758 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2759 304832 ppc32: build failure
2760 305431 Use find_buildid shdr fallback for separate .debug files
2761 305728 Add support for AVX2 instructions
2762 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2763 306035 s390x: Fix IR generation for LAAG and friends
2764 306054 s390x: Condition code computation for convert-to-int/logical
2765 306098 s390x: alternate opcode form for convert to/from fixed
2766 306587 Fix cache line detection from auxiliary vector for PPC.
2767 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2768 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2769 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2770 307101 sys_capget second argument can be NULL
2771 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2772 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2773 307113 s390x: DFP support
2774 307141 valgrind does't work in mips-linux system
2775 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2776 307285 x86_amd64 feature test for avx in test suite is wrong
2777 307290 memcheck overlap testcase needs memcpy version filter
2778 307463 Please add "&limit=0" to the "all open bugs" link
2779 307465 --show-possibly-lost=no should reduce the error count / exit code
2780 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2781 307729 pkgconfig support broken valgrind.pc
2782 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2783 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2784 308089 Unhandled syscall on ppc64: prctl
2785 308135 PPC32 MPC8xx has 16 bytes cache size
2786 308321 testsuite memcheck filter interferes with gdb_filter
2788 308341 vgdb should report process exit (or fatal signal)
2789 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2790 308495 Remove build dependency on installed Xen headers
2791 308573 Internal error on 64-bit instruction executed in 32-bit mode
2793 308627 pmovmskb validity bit propagation is imprecise
2794 308644 vgdb command for having the info for the track-fds option
2795 308711 give more info about aspacemgr and arenas in out_of_memory
2796 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2797 308718 ARM implement SMLALBB family of instructions
2798 308886 Missing support for PTRACE_SET/GETREGSET
2799 308930 syscall name_to_handle_at (303 on amd64) not handled
2800 309229 V-bit tester does not report number of tests generated
2801 309323 print unrecognized instuction on MIPS
2802 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2803 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2804 309430 Self hosting ppc64 encounters a vassert error on operand type
2805 309600 valgrind is a bit confused about 0-sized sections
2806 309823 Generate errors for still reachable blocks
2807 309921 PCMPISTRI validity bit propagation is imprecise
2808 309922 none/tests/ppc64/test_dfp5 sometimes fails
2809 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2810 310424 --read-var-info does not properly describe static variables
2811 310792 search additional path for debug symbols
2812 310931 s390x: Message-security assist (MSA) instruction extension [..]
2813 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2814 311318 ARM: "128-bit constant is not implemented" error message
2815 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2816 311690 V crashes because it redirects branches inside of a redirected function
2817 311880 x86_64: make regtest hangs at shell_valid1
2818 311922 WARNING: unhandled syscall: 170
2820 312171 ppc: insn selection for DFP
2821 312571 Rounding mode call wrong for the DFP Iops [..]
2822 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2823 312913 Dangling pointers error should also report the alloc stack trace
2824 312980 Building on Mountain Lion generates some compiler warnings
2825 313267 Adding MIPS64/Linux port to Valgrind
2828 313811 Buffer overflow in assert_fail
2829 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2830 314269 ppc: dead code in insn selection
2831 314718 ARM: implement integer divide instruction (sdiv and udiv)
2832 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2833 315441 sendmsg syscall should ignore unset msghdr msg_flags
2834 315534 msgrcv inside a thread causes valgrind to hang (block)
2835 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2836 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2837 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2838 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2839 316144 valgrind.1 manpage contains unknown ??? strings [..]
2840 316145 callgrind command line options in manpage reference (unknown) [..]
2841 316145 callgrind command line options in manpage reference [..]
2842 316181 drd: Fixed a 4x slowdown for certain applications
2843 316503 Valgrind does not support SSE4 "movntdqa" instruction
2844 316535 Use of |signed int| instead of |size_t| in valgrind messages
2845 316696 fluidanimate program of parsec 2.1 stuck
2846 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2847 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2848 317186 "Impossible happens" when occurs VCVT instruction on ARM
2849 317318 Support for Threading Building Blocks "scalable_malloc"
2850 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2851 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2852 317463 bmi testcase IR SANITY CHECK FAILURE
2853 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2854 318050 libmpiwrap fails to compile with out-of-source build
2855 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2856 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2857 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2858 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2859 318932 Add missing PPC64 and PPC32 system call support
2860 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2861 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2862 319494 VEX Makefile-gcc standalone build update after r2702
2863 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2864 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2865 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2866 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2867 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2868 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2869 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2870 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2871 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2872 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2873 320895 add fanotify support (patch included)
2874 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2875 321065 Valgrind updates for Xen 4.3
2876 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2877 321363 Unhandled instruction: SSAX (ARM + Thumb)
2878 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2879 321466 Unhandled instruction: SHASX (ARM + Thumb)
2880 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2881 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2882 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2883 321620 Unhandled instruction: UASX (ARM + Thumb)
2884 321621 Unhandled instruction: USAX (ARM + Thumb)
2885 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2886 321693 Unhandled instruction: LDRSBT (Thumb)
2887 321694 Unhandled instruction: UQASX (ARM + Thumb)
2888 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2889 321697 Unhandled instruction: UHASX (ARM + Thumb)
2890 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2891 321704 Unhandled instruction: REVSH (ARM + Thumb)
2892 321730 Add cg_diff and cg_merge man pages
2893 321738 Add vgdb and valgrind-listener man pages
2895 321891 Unhandled instruction: LDRHT (Thumb)
2896 321960 pthread_create() then alloca() causing invalid stack write errors
2897 321969 ppc32 and ppc64 don't support [lf]setxattr
2898 322254 Show threadname together with tid if set by application
2899 322294 Add initial support for IBM Power ISA 2.07
2900 322368 Assertion failure in wqthread_hijack under OS X 10.8
2901 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2902 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2903 322851 0bXXX binary literal syntax is not standard
2904 323035 Unhandled instruction: LDRSHT(Thumb)
2905 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2906 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2907 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2908 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2909 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2910 323437 Phase 2 support for IBM Power ISA 2.07
2911 323713 Support mmxext (integer sse) subset on i386 (athlon)
2912 323803 Transactional memory instructions are not supported for Power
2913 323893 SSE3 not available on amd cpus in valgrind
2914 323905 Probable false positive from Valgrind/drd on close()
2915 323912 valgrind.h header isn't compatible for mingw64
2916 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
2917 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
2918 324181 mmap does not handle MAP_32BIT
2919 324227 memcheck false positive leak when a thread calls exit+block [..]
2920 324421 Support for fanotify API on ARM architecture
2921 324514 gdbserver monitor cmd output behaviour consistency [..]
2922 324518 ppc64: Emulation of dcbt instructions does not handle [..]
2923 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
2924 324582 When access is made to freed memory, report both allocation [..]
2925 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
2926 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
2927 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
2928 324834 Unhandled instructions in Microsoft C run-time for x86_64
2929 324894 Phase 3 support for IBM Power ISA 2.07
2930 326091 drd: Avoid false race reports from optimized strlen() impls
2931 326113 valgrind libvex hwcaps error on AMD64
2932 n-i-bz Some wrong command line options could be ignored
2933 n-i-bz patch to allow fair-sched on android
2934 n-i-bz report error for vgdb snapshot requested before execution
2935 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
2937 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
2941 Release 3.8.1 (19 September 2012)
2942 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2943 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
2944 that occur moderately frequently in real use cases, adds support for
2945 some missing instructions on ARM, and fixes a deadlock condition on
2946 MacOSX. If you package or deliver 3.8.0 for others to use, you might
2947 want to consider upgrading to 3.8.1 instead.
2949 The following bugs have been fixed or resolved. Note that "n-i-bz"
2950 stands for "not in bugzilla" -- that is, a bug that was reported to us
2951 but never got a bugzilla entry. We encourage you to file bugs in
2952 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2953 than mailing the developers (or mailing lists) directly -- bugs that
2954 are not entered into bugzilla tend to get forgotten about or ignored.
2956 To see details of a given bug, visit
2957 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2958 where XXXXXX is the bug number as listed below.
2961 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
2962 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
2963 298281 wcslen causes false(?) uninitialised value warnings
2964 301281 valgrind hangs on OS X when the process calls system()
2965 304035 disInstr(arm): unhandled instruction 0xE1023053
2966 304867 implement MOVBE instruction in x86 mode
2967 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
2968 305042 amd64: implement 0F 7F encoding of movq between two registers
2969 305199 ARM: implement QDADD and QDSUB
2970 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
2971 305513 killed by fatal signal: SIGSEGV
2972 305690 DRD reporting invalid semaphore when sem_trywait fails
2973 305926 Invalid alignment checks for some AVX instructions
2974 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
2975 306310 3.8.0 release tarball missing some files
2976 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
2977 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
2978 n-i-bz shmat of a segment > 4Gb does not work
2979 n-i-bz simulate_control_c script wrong USR1 signal number on mips
2980 n-i-bz vgdb ptrace calls wrong on mips [...]
2981 n-i-bz Fixes for more MPI false positives
2982 n-i-bz exp-sgcheck's memcpy causes programs to segfault
2983 n-i-bz OSX build w/ clang: asserts at startup
2984 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
2985 n-i-bz fix a couple of union tag-vs-field mixups
2986 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
2988 The following bugs were fixed in 3.8.0 but not listed in this NEWS
2991 254088 Valgrind should know about UD2 instruction
2994 304754 NEWS blows TeX's little mind
2996 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
3000 Release 3.8.0 (10 August 2012)
3001 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3002 3.8.0 is a feature release with many improvements and the usual
3003 collection of bug fixes.
3005 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3006 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
3007 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
3008 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
3009 There is initial support for MacOSX 10.8, but it is not usable for
3010 serious work at present.
3012 * ================== PLATFORM CHANGES =================
3014 * Support for MIPS32 platforms running Linux. Valgrind has been
3015 tested on MIPS32 and MIPS32r2 platforms running different Debian
3016 Squeeze and MeeGo distributions. Both little-endian and big-endian
3017 cores are supported. The tools Memcheck, Massif and Lackey have
3018 been tested and are known to work. See README.mips for more details.
3020 * Preliminary support for Android running on x86.
3022 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
3024 * Support for Intel AVX instructions and for AES instructions. This
3025 support is available only for 64 bit code.
3027 * Support for POWER Decimal Floating Point instructions.
3029 * ==================== TOOL CHANGES ====================
3031 * Non-libc malloc implementations are now supported. This is useful
3032 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
3033 Using the new option --soname-synonyms, such tools can be informed
3034 that the malloc implementation is either linked statically into the
3035 executable, or is present in some other shared library different
3036 from libc.so. This makes it possible to process statically linked
3037 programs, and programs using other malloc libraries, for example
3038 TCMalloc or JEMalloc.
3040 * For tools that provide their own replacement for malloc et al, the
3041 option --redzone-size=<number> allows users to specify the size of
3042 the padding blocks (redzones) added before and after each client
3043 allocated block. Smaller redzones decrease the memory needed by
3044 Valgrind. Bigger redzones increase the chance to detect blocks
3045 overrun or underrun. Prior to this change, the redzone size was
3046 hardwired to 16 bytes in Memcheck.
3050 - The leak_check GDB server monitor command now can
3051 control the maximum nr of loss records to output.
3053 - Reduction of memory use for applications allocating
3054 many blocks and/or having many partially defined bytes.
3056 - Addition of GDB server monitor command 'block_list' that lists
3057 the addresses/sizes of the blocks of a leak search loss record.
3059 - Addition of GDB server monitor command 'who_points_at' that lists
3060 the locations pointing at a block.
3062 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
3063 detect an invalid access of these redzones, by marking them
3064 noaccess. Similarly, if a redzone size is given for a memory
3065 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
3066 This still allows to find some bugs if the user has forgotten to
3067 mark the pool superblock noaccess.
3069 - Performance of memory leak check has been improved, especially in
3070 cases where there are many leaked blocks and/or many suppression
3071 rules used to suppress leak reports.
3073 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
3074 more precise analysis, which is important for LLVM/Clang
3075 generated code. This is at the cost of somewhat reduced
3076 performance. Note there is no change to analysis precision or
3077 costs on Linux targets.
3081 - Added even more facilities that can help finding the cause of a data
3082 race, namely the command-line option --ptrace-addr and the macro
3083 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
3085 - Fixed a subtle bug that could cause false positive data race reports.
3087 * ==================== OTHER CHANGES ====================
3089 * The C++ demangler has been updated so as to work well with C++
3090 compiled by up to at least g++ 4.6.
3092 * Tool developers can make replacement/wrapping more flexible thanks
3093 to the new option --soname-synonyms. This was reported above, but
3094 in fact is very general and applies to all function
3095 replacement/wrapping, not just to malloc-family functions.
3097 * Round-robin scheduling of threads can be selected, using the new
3098 option --fair-sched= yes. Prior to this change, the pipe-based
3099 thread serialisation mechanism (which is still the default) could
3100 give very unfair scheduling. --fair-sched=yes improves
3101 responsiveness of interactive multithreaded applications, and
3102 improves repeatability of results from the thread checkers Helgrind
3105 * For tool developers: support to run Valgrind on Valgrind has been
3106 improved. We can now routinely Valgrind on Helgrind or Memcheck.
3108 * gdbserver now shows the float shadow registers as integer
3109 rather than float values, as the shadow values are mostly
3110 used as bit patterns.
3112 * Increased limit for the --num-callers command line flag to 500.
3114 * Performance improvements for error matching when there are many
3115 suppression records in use.
3117 * Improved support for DWARF4 debugging information (bug 284184).
3119 * Initial support for DWZ compressed Dwarf debug info.
3121 * Improved control over the IR optimiser's handling of the tradeoff
3122 between performance and precision of exceptions. Specifically,
3123 --vex-iropt-precise-memory-exns has been removed and replaced by
3124 --vex-iropt-register-updates, with extended functionality. This
3125 allows the Valgrind gdbserver to always show up to date register
3128 * Modest performance gains through the use of translation chaining for
3131 * ==================== FIXED BUGS ====================
3133 The following bugs have been fixed or resolved. Note that "n-i-bz"
3134 stands for "not in bugzilla" -- that is, a bug that was reported to us
3135 but never got a bugzilla entry. We encourage you to file bugs in
3136 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3137 than mailing the developers (or mailing lists) directly -- bugs that
3138 are not entered into bugzilla tend to get forgotten about or ignored.
3140 To see details of a given bug, visit
3141 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3142 where XXXXXX is the bug number as listed below.
3144 197914 Building valgrind from svn now requires automake-1.10
3145 203877 increase to 16Mb maximum allowed alignment for memalign et al
3146 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
3147 247386 make perf does not run all performance tests
3148 270006 Valgrind scheduler unfair
3149 270777 Adding MIPS/Linux port to Valgrind
3150 270796 s390x: Removed broken support for the TS insn
3151 271438 Fix configure for proper SSE4.2 detection
3152 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
3153 273475 Add support for AVX instructions
3154 274078 improved configure logic for mpicc
3155 276993 fix mremap 'no thrash checks'
3156 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
3157 281482 memcheck incorrect byte allocation count in realloc() for silly argument
3158 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
3159 283413 Fix wrong sanity check
3160 283671 Robustize alignment computation in LibVEX_Alloc
3161 283961 Adding support for some HCI IOCTLs
3162 284124 parse_type_DIE: confused by: DWARF 4
3163 284864 == 273475 (Add support for AVX instructions)
3164 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
3165 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
3166 285725 == 273475 (Add support for AVX instructions)
3167 286261 add wrapper for linux I2C_RDWR ioctl
3168 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
3169 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
3170 286384 configure fails "checking for a supported version of gcc"
3171 286497 == 273475 (Add support for AVX instructions)
3172 286596 == 273475 (Add support for AVX instructions)
3173 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
3174 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
3175 287260 Incorrect conditional jump or move depends on uninitialised value(s)
3176 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
3177 287307 == 273475 (Add support for AVX instructions)
3178 287858 VG_(strerror): unknown error
3179 288298 (MacOSX) unhandled syscall shm_unlink
3180 288995 == 273475 (Add support for AVX instructions)
3181 289470 Loading of large Mach-O thin binaries fails.
3182 289656 == 273475 (Add support for AVX instructions)
3183 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
3184 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
3185 289839 s390x: Provide support for unicode conversion instructions
3186 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
3187 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
3188 290655 Add support for AESKEYGENASSIST instruction
3189 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
3190 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
3191 291253 ES register not initialised in valgrind simulation
3192 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
3193 291865 s390x: Support the "Compare Double and Swap" family of instructions
3194 292300 == 273475 (Add support for AVX instructions)
3195 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
3196 292493 == 273475 (Add support for AVX instructions)
3197 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
3198 292627 Missing support for some SCSI ioctls
3199 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
3200 292841 == 273475 (Add support for AVX instructions)
3201 292993 implement the getcpu syscall on amd64-linux
3202 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
3203 293088 Add some VEX sanity checks for ppc64 unhandled instructions
3204 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
3205 293754 PCMPxSTRx not implemented for 16-bit characters
3206 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
3207 293808 CLFLUSH not supported by latest VEX for amd64
3208 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
3209 294048 MPSADBW instruction not implemented
3210 294055 regtest none/tests/shell fails when locale is not set to C
3211 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
3212 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
3213 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
3214 294260 disInstr_AMD64: disInstr miscalculated next %rip
3215 294523 --partial-loads-ok=yes causes false negatives
3216 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
3217 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
3218 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
3219 295089 can not annotate source for both helgrind and drd
3220 295221 POWER Processor decimal floating point instruction support missing
3221 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
3222 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
3223 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
3224 295617 ARM - Add some missing syscalls
3225 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
3226 296229 Linux user input device ioctls missing wrappers
3227 296318 ELF Debug info improvements (more than one rx/rw mapping)
3228 296422 Add translation chaining support
3229 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
3230 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
3231 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
3232 297078 gdbserver signal handling problems [..]
3233 297147 drd false positives on newly allocated memory
3234 297329 disallow decoding of IBM Power DFP insns on some machines
3235 297497 POWER Processor decimal floating point instruction support missing
3236 297701 Another alias for strncasecmp_l in libc-2.13.so
3237 297911 'invalid write' not reported when using APIs for custom mem allocators.
3238 297976 s390x: revisit EX implementation
3239 297991 Valgrind interferes with mmap()+ftell()
3240 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
3241 297993 Fix compilation of valgrind with gcc -g3.
3242 298080 POWER Processor DFP support missing, part 3
3243 298227 == 273475 (Add support for AVX instructions)
3244 298335 == 273475 (Add support for AVX instructions)
3245 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
3246 298394 s390x: Don't bail out on an unknown machine model. [..]
3247 298421 accept4() syscall (366) support is missing for ARM
3248 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
3249 298732 valgrind installation problem in ubuntu with kernel version 3.x
3250 298862 POWER Processor DFP instruction support missing, part 4
3251 298864 DWARF reader mis-parses DW_FORM_ref_addr
3252 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
3253 299053 Support DWARF4 DW_AT_high_pc constant form
3254 299104 == 273475 (Add support for AVX instructions)
3255 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
3256 299629 dup3() syscall (358) support is missing for ARM
3257 299694 POWER Processor DFP instruction support missing, part 5
3258 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
3259 299803 == 273475 (Add support for AVX instructions)
3260 299804 == 273475 (Add support for AVX instructions)
3261 299805 == 273475 (Add support for AVX instructions)
3262 300140 ARM - Missing (T1) SMMUL
3263 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
3264 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
3265 300414 FCOM and FCOMP unimplemented for amd64 guest
3266 301204 infinite loop in canonicaliseSymtab with ifunc symbol
3267 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
3268 301265 add x86 support to Android build
3269 301984 configure script doesn't detect certain versions of clang
3270 302205 Fix compiler warnings for POWER VEX code and POWER test cases
3271 302287 Unhandled movbe instruction on Atom processors
3272 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
3273 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
3274 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
3275 302656 == 273475 (Add support for AVX instructions)
3276 302709 valgrind for ARM needs extra tls support for android emulator [..]
3277 302827 add wrapper for CDROM_GET_CAPABILITY
3278 302901 Valgrind crashes with dwz optimized debuginfo
3279 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
3280 303116 Add support for the POWER instruction popcntb
3281 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
3282 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
3283 303466 == 273475 (Add support for AVX instructions)
3284 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
3285 303963 strstr() function produces wrong results under valgrind callgrind
3286 304054 CALL_FN_xx macros need to enforce stack alignment
3287 304561 tee system call not supported
3288 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
3289 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
3290 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
3291 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
3292 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
3293 n-i-bz s390x: Shadow registers can now be examined using vgdb
3295 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
3296 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
3300 Release 3.7.0 (5 November 2011)
3301 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3302 3.7.0 is a feature release with many significant improvements and the
3303 usual collection of bug fixes.
3305 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3306 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
3307 Support for recent distros and toolchain components (glibc 2.14, gcc
3308 4.6, MacOSX 10.7) has been added.
3310 * ================== PLATFORM CHANGES =================
3312 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
3313 analyse 64-bit programs running on z/Architecture. Most user space
3314 instructions up to and including z10 are supported. Valgrind has
3315 been tested extensively on z9, z10, and z196 machines running SLES
3316 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
3317 known to work well. Callgrind, Helgrind, and DRD work reasonably
3318 well on z9 and later models. See README.s390 for more details.
3320 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
3321 64-bit processes are supported. Some complex threaded applications
3322 (Firefox) are observed to hang when run as 32 bit applications,
3323 whereas 64-bit versions run OK. The cause is unknown. Memcheck
3324 will likely report some false errors. In general, expect some rough
3325 spots. This release also supports MacOSX 10.6, but drops support
3328 * Preliminary support for Android (on ARM). Valgrind can now run
3329 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
3330 README.android for more details, plus instructions on how to get
3333 * Support for the IBM Power ISA 2.06 (Power7 instructions)
3335 * General correctness and performance improvements for ARM/Linux, and,
3336 by extension, ARM/Android.
3338 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
3339 instruction set support is under development but is not available in
3342 * Support for AIX5 has been removed.
3344 * ==================== TOOL CHANGES ====================
3346 * Memcheck: some incremental changes:
3348 - reduction of memory use in some circumstances
3350 - improved handling of freed memory, which in some circumstances
3351 can cause detection of use-after-free that would previously have
3354 - fix of a longstanding bug that could cause false negatives (missed
3355 errors) in programs doing vector saturated narrowing instructions.
3357 * Helgrind: performance improvements and major memory use reductions,
3358 particularly for large, long running applications which perform many
3359 synchronisation (lock, unlock, etc) events. Plus many smaller
3362 - display of locksets for both threads involved in a race
3364 - general improvements in formatting/clarity of error messages
3366 - addition of facilities and documentation regarding annotation
3367 of thread safe reference counted C++ classes
3369 - new flag --check-stack-refs=no|yes [yes], to disable race checking
3370 on thread stacks (a performance hack)
3372 - new flag --free-is-write=no|yes [no], to enable detection of races
3373 where one thread accesses heap memory but another one frees it,
3374 without any coordinating synchronisation event
3376 * DRD: enabled XML output; added support for delayed thread deletion
3377 in order to detect races that occur close to the end of a thread
3378 (--join-list-vol); fixed a memory leak triggered by repeated client
3379 memory allocatation and deallocation; improved Darwin support.
3381 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
3383 * exp-sgcheck: this tool has been reduced in scope so as to improve
3384 performance and remove checking that Memcheck does better.
3385 Specifically, the ability to check for overruns for stack and global
3386 arrays is unchanged, but the ability to check for overruns of heap
3387 blocks has been removed. The tool has accordingly been renamed to
3388 exp-sgcheck ("Stack and Global Array Checking").
3390 * ==================== OTHER CHANGES ====================
3392 * GDB server: Valgrind now has an embedded GDB server. That means it
3393 is possible to control a Valgrind run from GDB, doing all the usual
3394 things that GDB can do (single stepping, breakpoints, examining
3395 data, etc). Tool-specific functionality is also available. For
3396 example, it is possible to query the definedness state of variables
3397 or memory from within GDB when running Memcheck; arbitrarily large
3398 memory watchpoints are supported, etc. To use the GDB server, start
3399 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3402 * Improved support for unfriendly self-modifying code: a new option
3403 --smc-check=all-non-file is available. This adds the relevant
3404 consistency checks only to code that originates in non-file-backed
3405 mappings. In effect this confines the consistency checking only to
3406 code that is or might be JIT generated, and avoids checks on code
3407 that must have been compiled ahead of time. This significantly
3408 improves performance on applications that generate code at run time.
3410 * It is now possible to build a working Valgrind using Clang-2.9 on
3413 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3414 These enable and disable error reporting on a per-thread, and
3415 nestable, basis. This is useful for hiding errors in particularly
3416 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3417 now uses this facility.
3419 * Added the --mod-funcname option to cg_diff.
3421 * ==================== FIXED BUGS ====================
3423 The following bugs have been fixed or resolved. Note that "n-i-bz"
3424 stands for "not in bugzilla" -- that is, a bug that was reported to us
3425 but never got a bugzilla entry. We encourage you to file bugs in
3426 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3427 mailing the developers (or mailing lists) directly -- bugs that are
3428 not entered into bugzilla tend to get forgotten about or ignored.
3430 To see details of a given bug, visit
3431 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3432 where XXXXXX is the bug number as listed below.
3434 79311 malloc silly arg warning does not give stack trace
3435 210935 port valgrind.h (not valgrind) to win32 to support client requests
3436 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3437 243404 Port to zSeries
3438 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3439 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3440 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3441 253206 Some fixes for the faultstatus testcase
3442 255223 capget testcase fails when running as root
3443 256703 xlc_dbl_u32.c testcase broken
3444 256726 Helgrind tests have broken inline asm
3445 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3446 264800 testcase compile failure on zseries
3447 265762 make public VEX headers compilable by G++ 3.x
3448 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3449 266753 configure script does not give the user the option to not use QtCore
3450 266931 gen_insn_test.pl is broken
3451 266961 ld-linux.so.2 i?86-linux strlen issues
3452 266990 setns instruction causes false positive
3453 267020 Make directory for temporary files configurable at run-time.
3454 267342 == 267997 (segmentation fault on Mac OS 10.6)
3455 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3456 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3457 267488 regtest: darwin support for 64-bit build
3458 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3459 267630 Add support for IBM Power ISA 2.06 -- stage 1
3460 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3461 267819 Add client request for informing the core about reallocation
3462 267925 laog data structure quadratic for a single sequence of lock
3463 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3464 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3465 268513 missed optimizations in fold_Expr
3466 268619 s390x: fpr - gpr transfer facility
3467 268620 s390x: reconsider "long displacement" requirement
3468 268621 s390x: improve IR generation for XC
3469 268715 s390x: FLOGR is not universally available
3470 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3471 268930 s390x: MHY is not universally available
3472 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3473 269079 Support ptrace system call on ARM
3474 269144 missing "Bad option" error message
3475 269209 conditional load and store facility (z196)
3476 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3477 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3478 269736 s390x: minor code generation tweaks
3479 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3480 269863 s390x: remove unused function parameters
3481 269864 s390x: tweak s390_emit_load_cc
3482 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3483 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3484 270115 s390x: rewrite some testcases
3485 270309 == 267997 (valgrind crash on startup)
3486 270320 add support for Linux FIOQSIZE ioctl() call
3487 270326 segfault while trying to sanitize the environment passed to execle
3488 270794 IBM POWER7 support patch causes regression in none/tests
3489 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3490 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3491 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3492 270959 s390x: invalid use of R0 as base register
3493 271042 VSX configure check fails when it should not
3494 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3495 271259 s390x: fix code confusion
3496 271337 == 267997 (Valgrind segfaults on MacOS X)
3497 271385 s390x: Implement Ist_MBE
3498 271501 s390x: misc cleanups
3499 271504 s390x: promote likely and unlikely
3500 271579 ppc: using wrong enum type
3501 271615 unhandled instruction "popcnt" (arch=amd10h)
3502 271730 Fix bug when checking ioctls: duplicate check
3503 271776 s390x: provide STFLE instruction support
3504 271779 s390x: provide clock instructions like STCK
3505 271799 Darwin: ioctls without an arg report a memory error
3506 271820 arm: fix type confusion
3507 271917 pthread_cond_timedwait failure leads to not-locked false positive
3508 272067 s390x: fix DISP20 macro
3509 272615 A typo in debug output in mc_leakcheck.c
3510 272661 callgrind_annotate chokes when run from paths containing regex chars
3511 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3512 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3513 272967 make documentation build-system more robust
3514 272986 Fix gcc-4.6 warnings with valgrind.h
3515 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3516 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3517 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3518 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3519 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3520 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3521 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3522 273778 exp-ptrcheck: unhandled sysno == 259
3523 274089 exp-ptrcheck: unhandled sysno == 208
3524 274378 s390x: Various dispatcher tweaks
3525 274447 WARNING: unhandled syscall: 340
3526 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3527 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3528 274926 valgrind does not build against linux-3
3529 275148 configure FAIL with glibc-2.14
3530 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3531 275168 Make Valgrind work for MacOSX 10.7 Lion
3532 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3533 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3534 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3535 275308 Fix implementation for ppc64 fres instruc
3536 275339 s390x: fix testcase compile warnings
3537 275517 s390x: Provide support for CKSM instruction
3538 275710 s390x: get rid of redundant address mode calculation
3539 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3540 275852 == 250101 (valgrind uses all swap space and is killed)
3541 276784 Add support for IBM Power ISA 2.06 -- stage 3
3542 276987 gdbsrv: fix tests following recent commits
3543 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3544 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3545 277471 Unhandled syscall: 340
3546 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3547 277653 ARM: support Thumb2 PLD instruction
3548 277663 ARM: NEON float VMUL by scalar incorrect
3549 277689 ARM: tests for VSTn with register post-index are broken
3550 277694 ARM: BLX LR instruction broken in ARM mode
3551 277780 ARM: VMOV.F32 (immediate) instruction is broken
3552 278057 fuse filesystem syscall deadlocks
3553 278078 Unimplemented syscall 280 on ppc32
3554 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3555 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3556 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3557 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3558 279027 Support for MVCL and CLCL instruction
3559 279027 s390x: Provide support for CLCL and MVCL instructions
3560 279062 Remove a redundant check in the insn selector for ppc.
3561 279071 JDK creates PTEST with redundant REX.W prefix
3562 279212 gdbsrv: add monitor cmd v.info scheduler.
3563 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3564 279698 memcheck discards valid-bits for packuswb
3565 279795 memcheck reports uninitialised values for mincore on amd64
3566 279994 Add support for IBM Power ISA 2.06 -- stage 3
3567 280083 mempolicy syscall check errors
3568 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3569 280710 s390x: config files for nightly builds
3570 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3571 280965 Valgrind breaks fcntl locks when program does mmap
3572 281138 WARNING: unhandled syscall: 340
3573 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3574 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3575 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3576 281468 s390x: handle do_clone and gcc clones in call traces
3577 281488 ARM: VFP register corruption
3578 281828 == 275284 (false memmove warning: "Source and destination overlap")
3579 281883 s390x: Fix system call wrapper for "clone".
3580 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3581 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3582 282238 SLES10: make check fails
3583 282979 strcasestr needs replacement with recent(>=2.12) glibc
3584 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3585 283243 Regression in ppc64 memcheck tests
3586 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3587 283427 re-connect epoll_pwait syscall on ARM linux
3588 283600 gdbsrv: android: port vgdb.c
3589 283709 none/tests/faultstatus needs to account for page size
3590 284305 filter_gdb needs enhancement to work on ppc64
3591 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3592 284472 Thumb2 ROR.W encoding T2 not implemented
3593 284621 XML-escape process command line in XML output
3594 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3595 that have non-power-of-2 sizes (also AMDs)
3596 n-i-bz don't be spooked by libraries mashed by elfhack
3597 n-i-bz don't be spooked by libxul.so linked with gold
3598 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3600 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3601 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3602 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3606 Release 3.6.1 (16 February 2011)
3607 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3608 3.6.1 is a bug fix release. It adds support for some SSE4
3609 instructions that were omitted in 3.6.0 due to lack of time. Initial
3610 support for glibc-2.13 has been added. A number of bugs causing
3611 crashing or assertion failures have been fixed.
3613 The following bugs have been fixed or resolved. Note that "n-i-bz"
3614 stands for "not in bugzilla" -- that is, a bug that was reported to us
3615 but never got a bugzilla entry. We encourage you to file bugs in
3616 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3617 mailing the developers (or mailing lists) directly -- bugs that are
3618 not entered into bugzilla tend to get forgotten about or ignored.
3620 To see details of a given bug, visit
3621 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3622 where XXXXXX is the bug number as listed below.
3624 188572 Valgrind on Mac should suppress setenv() mem leak
3625 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3626 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3627 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3628 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3629 254420 memory pool tracking broken
3630 254957 Test code failing to compile due to changes in memcheck.h
3631 255009 helgrind/drd: crash on chmod with invalid parameter
3632 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3633 255355 helgrind/drd: crash on threaded programs doing fork
3635 255418 (SSE4.x) rint call compiled with ICC
3636 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3637 255888 closing valgrindoutput tag outputted to log-stream on error
3638 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3639 255966 Slowness when using mempool annotations
3640 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3641 256600 super-optimized strcasecmp() false positive
3642 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3643 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3644 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3645 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3646 257276 Missing case in memcheck --track-origins=yes
3647 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3648 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3649 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3650 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3651 263099 callgrind_annotate counts Ir improperly [...]
3652 263877 undefined coprocessor instruction on ARMv7
3653 265964 configure FAIL with glibc-2.13
3654 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3655 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3656 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3657 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3658 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3660 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3664 Release 3.6.0 (21 October 2010)
3665 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3666 3.6.0 is a feature release with many significant improvements and the
3667 usual collection of bug fixes.
3669 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3670 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3671 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3673 -------------------------
3675 Here are some highlights. Details are shown further down:
3677 * Support for ARM/Linux.
3679 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3681 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3683 * Support for the SSE4.2 instruction set.
3685 * Enhancements to the Callgrind profiler, including the ability to
3686 handle CPUs with three levels of cache.
3688 * A new experimental heap profiler, DHAT.
3690 * A huge number of bug fixes and small enhancements.
3692 -------------------------
3694 Here are details of the above changes, together with descriptions of
3695 many other changes, and a list of fixed bugs.
3697 * ================== PLATFORM CHANGES =================
3699 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3700 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3701 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3703 This requires a CPU capable of running the ARMv7-A instruction set
3704 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3705 of the user space instruction set, including ARM and Thumb integer
3706 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3707 Cachegrind and Massif tools work properly; other tools work to
3710 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3711 with support for recent releases of the underlying toolchain
3712 components, notably gcc-4.5 and glibc-2.12.
3714 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3715 support also works much better on OS X 10.5, and is as solid as
3718 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3719 64-bit mode. In 32-bit mode, support is only available up to and
3720 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3721 supported in 64-bit mode, and 32-bit mode does in fact support the
3722 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3723 10.6 on 32-bit targets.
3725 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3726 and including version 2.05 is supported.
3728 * ==================== TOOL CHANGES ====================
3730 * Cachegrind has a new processing script, cg_diff, which finds the
3731 difference between two profiles. It's very useful for evaluating
3732 the performance effects of a change in a program.
3734 Related to this change, the meaning of cg_annotate's (rarely-used)
3735 --threshold option has changed; this is unlikely to affect many
3736 people, if you do use it please see the user manual for details.
3738 * Callgrind now can do branch prediction simulation, similar to
3739 Cachegrind. In addition, it optionally can count the number of
3740 executed global bus events. Both can be used for a better
3741 approximation of a "Cycle Estimation" as derived event (you need to
3742 update the event formula in KCachegrind yourself).
3744 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3745 rather than the L2 cache. This is to accommodate machines with
3746 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3747 cache configuration of such a machine it will run the simulation as
3748 if the L2 cache isn't present. This means the results are less
3749 likely to match the true result for the machine, but
3750 Cachegrind/Callgrind's results are already only approximate, and
3751 should not be considered authoritative. The results are still
3752 useful for giving a general idea about a program's locality.
3754 * Massif has a new option, --pages-as-heap, which is disabled by
3755 default. When enabled, instead of tracking allocations at the level
3756 of heap blocks (as allocated with malloc/new/new[]), it instead
3757 tracks memory allocations at the level of memory pages (as mapped by
3758 mmap, brk, etc). Each mapped page is treated as its own block.
3759 Interpreting the page-level output is harder than the heap-level
3760 output, but this option is useful if you want to account for every
3761 byte of memory used by a program.
3763 * DRD has two new command-line options: --free-is-write and
3764 --trace-alloc. The former allows to detect reading from already freed
3765 memory, and the latter allows tracing of all memory allocations and
3768 * DRD has several new annotations. Custom barrier implementations can
3769 now be annotated, as well as benign races on static variables.
3771 * DRD's happens before / happens after annotations have been made more
3772 powerful, so that they can now also be used to annotate e.g. a smart
3773 pointer implementation.
3775 * Helgrind's annotation set has also been drastically improved, so as
3776 to provide to users a general set of annotations to describe locks,
3777 semaphores, barriers and condition variables. Annotations to
3778 describe thread-safe reference counted heap objects have also been
3781 * Memcheck has a new command-line option, --show-possibly-lost, which
3782 is enabled by default. When disabled, the leak detector will not
3783 show possibly-lost blocks.
3785 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3786 has been added. DHAT keeps track of allocated heap blocks, and also
3787 inspects every memory reference to see which block (if any) is being
3788 accessed. This gives a lot of insight into block lifetimes,
3789 utilisation, turnover, liveness, and the location of hot and cold
3790 fields. You can use DHAT to do hot-field profiling.
3792 * ==================== OTHER CHANGES ====================
3794 * Improved support for unfriendly self-modifying code: the extra
3795 overhead incurred by --smc-check=all has been reduced by
3796 approximately a factor of 5 as compared with 3.5.0.
3798 * Ability to show directory names for source files in error messages.
3799 This is combined with a flexible mechanism for specifying which
3800 parts of the paths should be shown. This is enabled by the new flag
3803 * A new flag, --require-text-symbol, which will stop the run if a
3804 specified symbol is not found it a given shared object when it is
3805 loaded into the process. This makes advanced working with function
3806 intercepting and wrapping safer and more reliable.
3808 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3809 and control of Valgrind is now available for the tools Memcheck and
3810 Helgrind. XML output from Valgrind is available for Memcheck,
3811 Helgrind and exp-Ptrcheck.
3813 * More reliable stack unwinding on amd64-linux, particularly in the
3814 presence of function wrappers, and with gcc-4.5 compiled code.
3816 * Modest scalability (performance improvements) for massive
3817 long-running applications, particularly for those with huge amounts
3820 * Support for analyzing programs running under Wine with has been
3821 improved. The header files <valgrind/valgrind.h>,
3822 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3823 Windows-programs compiled with MinGW or one of the Microsoft Visual
3826 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3827 The 32-bit simulator was not affected. This did not occur often,
3828 but when it did would usually crash the program under test.
3831 * A large number of bugs were fixed. These are shown below.
3833 * A number of bugs were investigated, and were candidates for fixing,
3834 but are not fixed in 3.6.0, due to lack of developer time. They may
3835 get fixed in later releases. They are:
3837 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3838 212419 false positive "lock order violated" (A+B vs A)
3839 213685 Undefined value propagates past dependency breaking instruction
3840 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3841 237920 valgrind segfault on fork failure
3842 242137 support for code compiled by LLVM-2.8
3843 242423 Another unknown Intel cache config value
3844 243232 Inconsistent Lock Orderings report with trylock
3845 243483 ppc: callgrind triggers VEX assertion failure
3846 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3847 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3849 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3850 249435 Analyzing wine programs with callgrind triggers a crash
3851 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3852 250065 Handling large allocations
3853 250101 huge "free" memory usage due to m_mallocfree.c
3854 "superblocks fragmentation"
3855 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3856 252091 Callgrind on ARM does not detect function returns correctly
3857 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3858 254420 memory pool tracking broken
3859 n-i-bz support for adding symbols for JIT generated code
3862 The following bugs have been fixed or resolved. Note that "n-i-bz"
3863 stands for "not in bugzilla" -- that is, a bug that was reported to us
3864 but never got a bugzilla entry. We encourage you to file bugs in
3865 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3866 mailing the developers (or mailing lists) directly -- bugs that are
3867 not entered into bugzilla tend to get forgotten about or ignored.
3869 To see details of a given bug, visit
3870 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3871 where XXXXXX is the bug number as listed below.
3873 135264 dcbzl instruction missing
3875 153699 Valgrind should report unaligned reads with movdqa
3877 190429 Valgrind reports lost of errors in ld.so
3878 with x86_64 2.9.90 glibc
3879 197266 valgrind appears to choke on the xmms instruction
3881 197988 Crash when demangling very large symbol names
3882 202315 unhandled syscall: 332 (inotify_init1)
3883 203256 Add page-level profiling to Massif
3884 205093 dsymutil=yes needs quotes, locking (partial fix)
3885 205241 Snow Leopard 10.6 support (partial fix)
3886 206600 Leak checker fails to upgrade indirect blocks when their
3887 parent becomes reachable
3888 210935 port valgrind.h (not valgrind) to win32 so apps run under
3889 wine can make client requests
3890 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3891 within Linux ip-stack checksum functions
3892 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3894 213685 Undefined value propagates past dependency breaking instruction
3896 215914 Valgrind inserts bogus empty environment variable
3898 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3899 222545 shmat fails under valgind on some arm targets
3900 222560 ARM NEON support
3903 232509 Docs build fails with formatting inside <title></title> elements
3905 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3906 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3907 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3908 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3909 237485 symlink (syscall 57) is not supported on Mac OS
3910 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3912 238208 is_just_below_ESP doesn't take into account red-zone
3913 238345 valgrind passes wrong $0 when executing a shell script
3914 238679 mq_timedreceive syscall doesn't flag the reception buffer
3916 238696 fcntl command F_DUPFD_CLOEXEC not supported
3917 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3918 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3919 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
3921 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
3927 242606 unhandled syscall: setegid (in Ptrcheck)
3928 242814 Helgrind "Impossible has happened" during
3929 QApplication::initInstance();
3930 243064 Valgrind attempting to read debug information from iso
3931 243270 Make stack unwinding in Valgrind wrappers more reliable
3932 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
3933 sysno = 277 (mq_open)
3934 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
3935 244493 ARM VFP d16-d31 registers support
3936 244670 add support for audit_session_self syscall on Mac OS 10.6
3937 244921 The xml report of helgrind tool is not well format
3938 244923 In the xml report file, the <preamble> not escape the
3939 xml char, eg '<','&','>'
3940 245535 print full path names in plain text reports
3941 245925 x86-64 red zone handling problem
3942 246258 Valgrind not catching integer underruns + new [] s
3943 246311 reg/reg cmpxchg doesn't work on amd64
3944 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
3945 246888 Improve Makefile.vex.am
3946 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
3947 to [f]chmod_extended
3948 247526 IBM POWER6 (ISA 2.05) support is incomplete
3949 247561 Some leak testcases fails due to reachable addresses in
3951 247875 sizeofIRType to handle Ity_I128
3952 247894 [PATCH] unhandled syscall sys_readahead
3953 247980 Doesn't honor CFLAGS passed to configure
3954 248373 darwin10.supp is empty in the trunk
3955 248822 Linux FIBMAP ioctl has int parameter instead of long
3956 248893 [PATCH] make readdwarf.c big endianess safe to enable
3957 unwinding on big endian systems
3958 249224 Syscall 336 not supported (SYS_proc_info)
3960 249775 Incorrect scheme for detecting NEON capabilities of host CPU
3961 249943 jni JVM init fails when using valgrind
3962 249991 Valgrind incorrectly declares AESKEYGENASSIST support
3964 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
3965 250799 frexp$fenv_access_off function generates SIGILL
3966 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
3967 251251 support pclmulqdq insn
3968 251362 valgrind: ARM: attach to debugger either fails or provokes
3970 251674 Unhandled syscall 294
3973 254257 Add support for debugfiles found by build-id
3974 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
3975 254646 Wrapped functions cause stack misalignment on OS X
3976 (and possibly Linux)
3977 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
3979 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
3983 Release 3.5.0 (19 August 2009)
3984 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3985 3.5.0 is a feature release with many significant improvements and the
3986 usual collection of bug fixes. The main improvement is that Valgrind
3987 now works on Mac OS X.
3989 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
3990 and X86/Darwin. Support for recent distros and toolchain components
3991 (glibc 2.10, gcc 4.5) has been added.
3993 -------------------------
3995 Here is a short summary of the changes. Details are shown further
3998 * Support for Mac OS X (10.5.x).
4000 * Improvements and simplifications to Memcheck's leak checker.
4002 * Clarification and simplifications in various aspects of Valgrind's
4005 * XML output for Helgrind and Ptrcheck.
4007 * Performance and stability improvements for Helgrind and DRD.
4009 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
4011 * A new experimental tool, BBV, useful for computer architecture
4014 * Improved Wine support, including ability to read Windows PDB
4017 -------------------------
4019 Here are details of the above changes, followed by descriptions of
4020 many other minor changes, and a list of fixed bugs.
4023 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
4024 called "Darwin" because that is the name of the OS core, which is the
4025 level that Valgrind works at.)
4029 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
4030 because it would require work and 10.4 is only becoming less common.
4032 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
4033 fairly well. For 10.5.x, 32-bit programs are the default even on
4034 64-bit machines, so it handles most current programs.
4036 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
4037 officially supported, but simple programs at least will probably work.
4038 However, start-up is slow.
4040 - PowerPC machines are not supported.
4042 Things that don't work:
4044 - The Ptrcheck tool.
4046 - Objective-C garbage collection.
4050 - If you have Rogue Amoeba's "Instant Hijack" program installed,
4051 Valgrind will fail with a SIGTRAP at start-up. See
4052 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
4057 - You will likely find --dsymutil=yes a useful option, as error
4058 messages may be imprecise without it.
4060 - Mac OS X support is new and therefore will be less robust than the
4061 Linux support. Please report any bugs you find.
4063 - Threaded programs may run more slowly than on Linux.
4065 Many thanks to Greg Parker for developing this port over several years.
4068 * Memcheck's leak checker has been improved.
4070 - The results for --leak-check=summary now match the summary results
4071 for --leak-check=full. Previously they could differ because
4072 --leak-check=summary counted "indirectly lost" blocks and
4073 "suppressed" blocks as "definitely lost".
4075 - Blocks that are only reachable via at least one interior-pointer,
4076 but are directly pointed to by a start-pointer, were previously
4077 marked as "still reachable". They are now correctly marked as
4080 - The default value for the --leak-resolution option has been
4081 changed from "low" to "high". In general, this means that more
4082 leak reports will be produced, but each leak report will describe
4083 fewer leaked blocks.
4085 - With --leak-check=full, "definitely lost" and "possibly lost"
4086 leaks are now considered as proper errors, ie. they are counted
4087 for the "ERROR SUMMARY" and affect the behaviour of
4088 --error-exitcode. These leaks are not counted as errors if
4089 --leak-check=summary is specified, however.
4091 - Documentation for the leak checker has been improved.
4094 * Various aspects of Valgrind's text output have changed.
4096 - Valgrind's start-up message has changed. It is shorter but also
4097 includes the command being run, which makes it easier to use
4098 --trace-children=yes. An example:
4100 - Valgrind's shut-down messages have also changed. This is most
4101 noticeable with Memcheck, where the leak summary now occurs before
4102 the error summary. This change was necessary to allow leaks to be
4103 counted as proper errors (see the description of the leak checker
4104 changes above for more details). This was also necessary to fix a
4105 longstanding bug in which uses of suppressions against leaks were
4106 not "counted", leading to difficulties in maintaining suppression
4107 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
4109 - Behavior of -v has changed. In previous versions, -v printed out
4110 a mixture of marginally-user-useful information, and tool/core
4111 statistics. The statistics printing has now been moved to its own
4112 flag, --stats=yes. This means -v is less verbose and more likely
4113 to convey useful end-user information.
4115 - The format of some (non-XML) stack trace entries has changed a
4116 little. Previously there were six possible forms:
4118 0x80483BF: really (a.c:20)
4119 0x80483BF: really (in /foo/a.out)
4121 0x80483BF: (within /foo/a.out)
4122 0x80483BF: ??? (a.c:20)
4125 The third and fourth of these forms have been made more consistent
4126 with the others. The six possible forms are now:
4128 0x80483BF: really (a.c:20)
4129 0x80483BF: really (in /foo/a.out)
4130 0x80483BF: really (in ???)
4131 0x80483BF: ??? (in /foo/a.out)
4132 0x80483BF: ??? (a.c:20)
4135 Stack traces produced when --xml=yes is specified are different
4139 * Helgrind and Ptrcheck now support XML output, so they can be used
4140 from GUI tools. Also, the XML output mechanism has been
4143 - The XML format has been overhauled and generalised, so it is more
4144 suitable for error reporting tools in general. The Memcheck
4145 specific aspects of it have been removed. The new format, which
4146 is an evolution of the old format, is described in
4147 docs/internals/xml-output-protocol4.txt.
4149 - Memcheck has been updated to use the new format.
4151 - Helgrind and Ptrcheck are now able to emit output in this format.
4153 - The XML output mechanism has been overhauled. XML is now output
4154 to its own file descriptor, which means that:
4156 * Valgrind can output text and XML independently.
4158 * The longstanding problem of XML output being corrupted by
4159 unexpected un-tagged text messages is solved.
4161 As before, the destination for text output is specified using
4162 --log-file=, --log-fd= or --log-socket=.
4164 As before, XML output for a tool is enabled using --xml=yes.
4166 Because there's a new XML output channel, the XML output
4167 destination is now specified by --xml-file=, --xml-fd= or
4170 Initial feedback has shown this causes some confusion. To
4171 clarify, the two envisaged usage scenarios are:
4173 (1) Normal text output. In this case, do not specify --xml=yes
4174 nor any of --xml-file=, --xml-fd= or --xml-socket=.
4176 (2) XML output. In this case, specify --xml=yes, and one of
4177 --xml-file=, --xml-fd= or --xml-socket= to select the XML
4178 destination, one of --log-file=, --log-fd= or --log-socket=
4179 to select the destination for any remaining text messages,
4180 and, importantly, -q.
4182 -q makes Valgrind completely silent on the text channel,
4183 except in the case of critical failures, such as Valgrind
4184 itself segfaulting, or failing to read debugging information.
4185 Hence, in this scenario, it suffices to check whether or not
4186 any output appeared on the text channel. If yes, then it is
4187 likely to be a critical error which should be brought to the
4188 attention of the user. If no (the text channel produced no
4189 output) then it can be assumed that the run was successful.
4191 This allows GUIs to make the critical distinction they need to
4192 make (did the run fail or not?) without having to search or
4193 filter the text output channel in any way.
4195 It is also recommended to use --child-silent-after-fork=yes in
4199 * Improvements and changes in Helgrind:
4201 - XML output, as described above
4203 - Checks for consistent association between pthread condition
4204 variables and their associated mutexes are now performed.
4206 - pthread_spinlock functions are supported.
4208 - Modest performance improvements.
4210 - Initial (skeletal) support for describing the behaviour of
4211 non-POSIX synchronisation objects through ThreadSanitizer
4212 compatible ANNOTATE_* macros.
4214 - More controllable tradeoffs between performance and the level of
4215 detail of "previous" accesses in a race. There are now three
4218 * --history-level=full. This is the default, and was also the
4219 default in 3.4.x. It shows both stacks involved in a race, but
4220 requires a lot of memory and can be very slow in programs that
4221 do many inter-thread synchronisation events.
4223 * --history-level=none. This only shows the later stack involved
4224 in a race. This can be much faster than --history-level=full,
4225 but makes it much more difficult to find the other access
4226 involved in the race.
4228 The new intermediate setting is
4230 * --history-level=approx
4232 For the earlier (other) access, two stacks are presented. The
4233 earlier access is guaranteed to be somewhere in between the two
4234 program points denoted by those stacks. This is not as useful
4235 as showing the exact stack for the previous access (as per
4236 --history-level=full), but it is better than nothing, and it's
4237 almost as fast as --history-level=none.
4240 * New features and improvements in DRD:
4242 - The error messages printed by DRD are now easier to interpret.
4243 Instead of using two different numbers to identify each thread
4244 (Valgrind thread ID and DRD thread ID), DRD does now identify
4245 threads via a single number (the DRD thread ID). Furthermore
4246 "first observed at" information is now printed for all error
4247 messages related to synchronization objects.
4249 - Added support for named semaphores (sem_open() and sem_close()).
4251 - Race conditions between pthread_barrier_wait() and
4252 pthread_barrier_destroy() calls are now reported.
4254 - Added support for custom allocators through the macros
4255 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
4256 in <valgrind/valgrind.h>). An alternative for these two macros is
4257 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
4260 - Added support for annotating non-POSIX synchronization objects
4261 through several new ANNOTATE_*() macros.
4263 - OpenMP: added support for the OpenMP runtime (libgomp) included
4264 with gcc versions 4.3.0 and 4.4.0.
4268 - Added two new command-line options (--first-race-only and
4269 --segment-merging-interval).
4272 * Genuinely atomic support for x86/amd64/ppc atomic instructions
4274 Valgrind will now preserve (memory-access) atomicity of LOCK-
4275 prefixed x86/amd64 instructions, and any others implying a global
4276 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
4278 This means that Valgrinded processes will "play nicely" in
4279 situations where communication with other processes, or the kernel,
4280 is done through shared memory and coordinated with such atomic
4281 instructions. Prior to this change, such arrangements usually
4282 resulted in hangs, races or other synchronisation failures, because
4283 Valgrind did not honour atomicity of such instructions.
4286 * A new experimental tool, BBV, has been added. BBV generates basic
4287 block vectors for use with the SimPoint analysis tool, which allows
4288 a program's overall behaviour to be approximated by running only a
4289 fraction of it. This is useful for computer architecture
4290 researchers. You can run BBV by specifying --tool=exp-bbv (the
4291 "exp-" prefix is short for "experimental"). BBV was written by
4295 * Modestly improved support for running Windows applications under
4296 Wine. In particular, initial support for reading Windows .PDB debug
4297 information has been added.
4300 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
4301 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
4305 * The Valgrind client requests VALGRIND_PRINTF and
4306 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
4307 the string was always printed immediately on its own line. Now, the
4308 string will be added to a buffer but not printed until a newline is
4309 encountered, or other Valgrind output is printed (note that for
4310 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
4311 "other Valgrind output"). This allows you to use multiple
4312 VALGRIND_PRINTF calls to build up a single output line, and also to
4313 print multiple output lines with a single request (by embedding
4314 multiple newlines in the string).
4317 * The graphs drawn by Massif's ms_print program have changed slightly:
4319 - The half-height chars '.' and ',' are no longer drawn, because
4320 they are confusing. The --y option can be used if the default
4321 y-resolution is not high enough.
4323 - Horizontal lines are now drawn after the top of a snapshot if
4324 there is a gap until the next snapshot. This makes it clear that
4325 the memory usage has not dropped to zero between snapshots.
4328 * Something that happened in 3.4.0, but wasn't clearly announced: the
4329 option --read-var-info=yes can be used by some tools (Memcheck,
4330 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
4331 variable type and location information. This makes those tools
4332 start up more slowly and increases memory consumption, but
4333 descriptions of data addresses in error messages become more
4337 * exp-Omega, an experimental instantaneous leak-detecting tool, was
4338 disabled in 3.4.0 due to a lack of interest and maintenance,
4339 although the source code was still in the distribution. The source
4340 code has now been removed from the distribution. For anyone
4341 interested, the removal occurred in SVN revision r10247.
4344 * Some changes have been made to the build system.
4346 - VEX/ is now integrated properly into the build system. This means
4347 that dependency tracking within VEX/ now works properly, "make
4348 install" will work without requiring "make" before it, and
4349 parallel builds (ie. 'make -j') now work (previously a
4350 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
4351 was effectively ignored).
4353 - The --with-vex configure option has been removed. It was of
4354 little use and removing it simplified the build system.
4356 - The location of some install files has changed. This should not
4357 affect most users. Those who might be affected:
4359 * For people who use Valgrind with MPI programs, the installed
4360 libmpiwrap.so library has moved from
4361 $(INSTALL)/<platform>/libmpiwrap.so to
4362 $(INSTALL)/libmpiwrap-<platform>.so.
4364 * For people who distribute standalone Valgrind tools, the
4365 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
4366 have moved to $(INSTALL)/libcoregrind-<platform>.a.
4368 These changes simplify the build system.
4370 - Previously, all the distributed suppression (*.supp) files were
4371 installed. Now, only default.supp is installed. This should not
4372 affect users as the other installed suppression files were not
4373 read; the fact that they were installed was a mistake.
4376 * KNOWN LIMITATIONS:
4378 - Memcheck is unusable with the Intel compiler suite version 11.1,
4379 when it generates code for SSE2-and-above capable targets. This
4380 is because of icc's use of highly optimised inlined strlen
4381 implementations. It causes Memcheck to report huge numbers of
4382 false errors even in simple programs. Helgrind and DRD may also
4385 Versions 11.0 and earlier may be OK, but this has not been
4389 The following bugs have been fixed or resolved. Note that "n-i-bz"
4390 stands for "not in bugzilla" -- that is, a bug that was reported to us
4391 but never got a bugzilla entry. We encourage you to file bugs in
4392 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4393 mailing the developers (or mailing lists) directly -- bugs that are
4394 not entered into bugzilla tend to get forgotten about or ignored.
4396 To see details of a given bug, visit
4397 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4398 where XXXXXX is the bug number as listed below.
4400 84303 How about a LockCheck tool?
4401 91633 dereference of null ptr in vgPlain_st_basetype
4402 97452 Valgrind doesn't report any pthreads problems
4403 100628 leak-check gets assertion failure when using
4404 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4405 108528 NPTL pthread cleanup handlers not called
4406 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4407 110128 mallinfo is not implemented...
4408 110770 VEX: Generated files not always updated when making valgrind
4409 111102 Memcheck: problems with large (memory footprint) applications
4410 115673 Vex's decoder should never assert
4411 117564 False positive: Syscall param clone(child_tidptr) contains
4412 uninitialised byte(s)
4413 119404 executing ssh from inside valgrind fails
4414 133679 Callgrind does not write path names to sources with dwarf debug
4416 135847 configure.in problem with non gnu compilers (and possible fix)
4417 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4418 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4419 136230 memcheck reports "possibly lost", should be "still reachable"
4420 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4421 137904 Valgrind reports a memory leak when using POSIX threads,
4423 139076 valgrind VT_GETSTATE error
4424 142228 complaint of elf_dynamic_do_rela in trivial usage
4425 145347 spurious warning with USBDEVFS_REAPURB
4426 148441 (wine) can't find memory leak in Wine, win32 binary
4428 148742 Leak-check fails assert on exit
4429 149878 add (proper) check for calloc integer overflow
4430 150606 Call graph is broken when using callgrind control
4431 152393 leak errors produce an exit code of 0. I need some way to
4432 cause leak errors to result in a nonzero exit code.
4433 157154 documentation (leak-resolution doc speaks about num-callers
4434 def=4) + what is a loss record
4435 159501 incorrect handling of ALSA ioctls
4436 162020 Valgrinding an empty/zero-byte file crashes valgrind
4437 162482 ppc: Valgrind crashes while reading stabs information
4438 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4439 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4440 163560 VEX/test_main.c is missing from valgrind-3.3.1
4441 164353 malloc_usable_size() doesn't return a usable size
4442 165468 Inconsistent formatting in memcheck manual -- please fix
4443 169505 main.c:286 (endOfInstr):
4444 Assertion 'ii->cost_offset == *cost_offset' failed
4445 177206 Generate default.supp during compile instead of configure
4446 177209 Configure valt_load_address based on arch+os
4447 177305 eventfd / syscall 323 patch lost
4448 179731 Tests fail to build because of inlining of non-local asm labels
4449 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4450 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4451 181594 Bogus warning for empty text segment
4452 181707 dwarf doesn't require enumerations to have name
4453 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4454 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4455 Assertion '!already_present' failed.
4456 185359 exp-ptrcheck: unhandled syscall getresuid()
4457 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4458 185816 Valgrind is unable to handle debug info for files with split
4459 debug info that are prelinked afterwards
4460 185980 [darwin] unhandled syscall: sem_open
4461 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4462 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4463 186790 Suppression pattern used for leaks are not reported
4464 186796 Symbols with length>200 in suppression files are ignored
4465 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4466 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4467 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4468 188046 bashisms in the configure script
4469 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4470 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4471 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4472 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4473 assertion fail "!lock->heldBy"
4474 188427 Add support for epoll_create1 (with patch)
4475 188530 Support for SIOCGSTAMPNS
4476 188560 Include valgrind.spec in the tarball
4477 188572 Valgrind on Mac should suppress setenv() mem leak
4478 189054 Valgrind fails to build because of duplicate non-local asm labels
4479 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4480 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4481 189763 drd assertion failure: s_threadinfo[tid].is_recording
4482 190219 unhandled syscall: 328 (x86-linux)
4483 190391 dup of 181394; see above
4484 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4485 190820 No debug information on powerpc-linux
4486 191095 PATCH: Improve usbdevfs ioctl handling
4487 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4489 191189 --xml=yes should obey --gen-suppressions=all
4490 191192 syslog() needs a suppression on macosx
4491 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4492 191761 getrlimit on MacOSX
4493 191992 multiple --fn-skip only works sometimes; dependent on order
4494 192634 V. reports "aspacem sync_check_mapping_callback:
4495 segment mismatch" on Darwin
4496 192954 __extension__ missing on 2 client requests
4497 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4498 194474 "INSTALL" file has different build instructions than "README"
4499 194671 Unhandled syscall (sem_wait?) from mac valgrind
4500 195069 memcheck: reports leak (memory still reachable) for
4502 195169 drd: (vgDrd_barrier_post_wait):
4503 Assertion 'r->sg[p->post_iteration]' failed.
4504 195268 valgrind --log-file doesn't accept ~/...
4505 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4506 195860 WARNING: unhandled syscall: unix:223
4507 196528 need a error suppression for pthread_rwlock_init under os x?
4508 197227 Support aio_* syscalls on Darwin
4509 197456 valgrind should reject --suppressions=(directory)
4510 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4511 197591 unhandled syscall 27 (mincore)
4512 197793 Merge DCAS branch to the trunk == 85756, 142103
4513 197794 Avoid duplicate filenames in Vex
4514 197898 make check fails on current SVN
4515 197901 make check fails also under exp-ptrcheck in current SVN
4516 197929 Make --leak-resolution=high the default
4517 197930 Reduce spacing between leak reports
4518 197933 Print command line of client at start-up, and shorten preamble
4519 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4520 198395 add BBV to the distribution as an experimental tool
4521 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4522 198649 callgrind_annotate doesn't cumulate counters
4523 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4524 199977 Valgrind complains about an unrecognized instruction in the
4525 atomic_incs test program
4526 200029 valgrind isn't able to read Fedora 12 debuginfo
4527 200760 darwin unhandled syscall: unix:284
4528 200827 DRD doesn't work on Mac OS X
4529 200990 VG_(read_millisecond_timer)() does not work correctly
4530 201016 Valgrind does not support pthread_kill() on Mac OS
4531 201169 Document --read-var-info
4532 201323 Pre-3.5.0 performance sanity checking
4533 201384 Review user manual for the 3.5.0 release
4534 201585 mfpvr not implemented on ppc
4535 201708 tests failing because x86 direction flag is left set
4536 201757 Valgrind doesn't handle any recent sys_futex additions
4537 204377 64-bit valgrind can not start a shell script
4538 (with #!/path/to/shell) if the shell is a 32-bit executable
4539 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4540 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4541 about memory allocation events with memory access tracing enabled
4542 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4544 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4548 Release 3.4.1 (28 February 2009)
4549 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4550 3.4.1 is a bug-fix release that fixes some regressions and assertion
4551 failures in debug info reading in 3.4.0, most notably incorrect stack
4552 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4553 other debug info problems are also fixed. A number of bugs in the
4554 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4556 In view of the fact that 3.4.0 contains user-visible regressions
4557 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4558 encouraged to ship 3.4.1 in preference to 3.4.0.
4560 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4561 bugzilla" -- that is, a bug that was reported to us but never got a
4562 bugzilla entry. We encourage you to file bugs in bugzilla
4563 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4564 developers (or mailing lists) directly -- bugs that are not entered
4565 into bugzilla tend to get forgotten about or ignored.
4567 n-i-bz Fix various bugs reading icc-11 generated debug info
4568 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4569 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4570 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4571 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4572 179618 exp-ptrcheck crashed / exit prematurely
4573 179624 helgrind: false positive races with pthread_create and
4574 recv/open/close/read
4575 134207 pkg-config output contains @VG_PLATFORM@
4576 176926 floating point exception at valgrind startup with PPC 440EPX
4577 181594 Bogus warning for empty text segment
4578 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4579 181707 Dwarf3 doesn't require enumerations to have name
4580 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4581 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4582 Assertion '!already_present' failed.
4583 185359 exp-ptrcheck unhandled syscall getresuid()
4585 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4586 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4590 Release 3.4.0 (2 January 2009)
4591 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4592 3.4.0 is a feature release with many significant improvements and the
4593 usual collection of bug fixes. This release supports X86/Linux,
4594 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4595 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4597 3.4.0 brings some significant tool improvements. Memcheck can now
4598 report the origin of uninitialised values, the thread checkers
4599 Helgrind and DRD are much improved, and we have a new experimental
4600 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4601 global arrays. In detail:
4603 * Memcheck is now able to track the origin of uninitialised values.
4604 When it reports an uninitialised value error, it will try to show
4605 the origin of the value, as either a heap or stack allocation.
4606 Origin tracking is expensive and so is not enabled by default. To
4607 use it, specify --track-origins=yes. Memcheck's speed will be
4608 essentially halved, and memory usage will be significantly
4609 increased. Nevertheless it can drastically reduce the effort
4610 required to identify the root cause of uninitialised value errors,
4611 and so is often a programmer productivity win, despite running more
4614 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4615 3.4.0, will be released shortly.
4617 * Helgrind's race detection algorithm has been completely redesigned
4618 and reimplemented, to address usability and scalability concerns:
4620 - The new algorithm has a lower false-error rate: it is much less
4621 likely to report races that do not really exist.
4623 - Helgrind will display full call stacks for both accesses involved
4624 in a race. This makes it easier to identify the root causes of
4627 - Limitations on the size of program that can run have been removed.
4629 - Performance has been modestly improved, although that is very
4632 - Direct support for Qt4 threading has been added.
4634 - pthread_barriers are now directly supported.
4636 - Helgrind works well on all supported Linux targets.
4638 * The DRD thread debugging tool has seen major improvements:
4640 - Greatly improved performance and significantly reduced memory
4643 - Support for several major threading libraries (Boost.Thread, Qt4,
4644 glib, OpenMP) has been added.
4646 - Support for atomic instructions, POSIX semaphores, barriers and
4647 reader-writer locks has been added.
4649 - Works now on PowerPC CPUs too.
4651 - Added support for printing thread stack usage at thread exit time.
4653 - Added support for debugging lock contention.
4655 - Added a manual for Drd.
4657 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4658 checks for misuses of pointers. In that sense it is a bit like
4659 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4660 detect overruns of stack and global arrays, it can detect
4661 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4662 detect accesses heap blocks that have been freed a very long time
4663 ago (millions of blocks in the past).
4665 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4666 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4667 of the main Valgrind documentation. As this is an experimental
4668 tool, we would be particularly interested in hearing about your
4669 experiences with it.
4671 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4672 longer built by default, although the code remains in the repository
4673 and the tarball. This is due to three factors: a perceived lack of
4674 users, a lack of maintenance, and concerns that it may not be
4675 possible to achieve reliable operation using the existing design.
4677 * As usual, support for the latest Linux distros and toolchain
4678 components has been added. It should work well on Fedora Core 10,
4679 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4680 state) is supported, as is glibc-2.9. The C++ demangler has been
4681 updated so as to work well with C++ compiled by even the most recent
4684 * You can now use frame-level wildcards in suppressions. This was a
4685 frequently-requested enhancement. A line "..." in a suppression now
4686 matches zero or more frames. This makes it easier to write
4687 suppressions which are precise yet insensitive to changes in
4690 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4692 * Very basic support for IBM Power6 has been added (64-bit processes only).
4694 * Valgrind is now cross-compilable. For example, it is possible to
4695 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4696 on a ppc32/64-linux target.
4698 * You can set the main thread's stack size at startup using the
4699 new --main-stacksize= flag (subject of course to ulimit settings).
4700 This is useful for running apps that need a lot of stack space.
4702 * The limitation that you can't use --trace-children=yes together
4703 with --db-attach=yes has been removed.
4705 * The following bugs have been fixed. Note that "n-i-bz" stands for
4706 "not in bugzilla" -- that is, a bug that was reported to us but
4707 never got a bugzilla entry. We encourage you to file bugs in
4708 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4709 mailing the developers (or mailing lists) directly.
4711 n-i-bz Make return types for some client requests 64-bit clean
4712 n-i-bz glibc 2.9 support
4713 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4714 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4715 n-i-bz Building in an env without gdb gives bogus gdb attach
4716 92456 Tracing the origin of uninitialised memory
4717 106497 Valgrind does not demangle some C++ template symbols
4719 151612 Suppression with "..." (frame-level wildcards in .supp files)
4720 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4721 159285 unhandled syscall:25 (stime, on x86-linux)
4722 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4723 160954 ppc build of valgrind crashes with illegal instruction (isel)
4724 160956 mallinfo implementation, w/ patch
4725 162092 Valgrind fails to start gnome-system-monitor
4726 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4727 163794 assertion failure with "--track-origins=yes"
4728 163933 sigcontext.err and .trapno must be set together
4729 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4730 164476 Missing kernel module loading system calls
4731 164669 SVN regression: mmap() drops posix file locks
4732 166581 Callgrind output corruption when program forks
4733 167288 Patch file for missing system calls on Cell BE
4734 168943 unsupported scas instruction pentium
4735 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4736 172417 x86->IR: 0x82 ...
4737 172563 amd64->IR: 0xD9 0xF5 - fprem1
4738 173099 .lds linker script generation error
4739 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4740 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4742 174908 --log-file value not expanded correctly for core file
4743 175044 Add lookup_dcookie for amd64
4744 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4746 Developer-visible changes:
4748 * Valgrind's debug-info reading machinery has been majorly overhauled.
4749 It can now correctly establish the addresses for ELF data symbols,
4750 which is something that has never worked properly before now.
4752 Also, Valgrind can now read DWARF3 type and location information for
4753 stack and global variables. This makes it possible to use the
4754 framework to build tools that rely on knowing the type and locations
4755 of stack and global variables, for example exp-Ptrcheck.
4757 Reading of such information is disabled by default, because most
4758 tools don't need it, and because it is expensive in space and time.
4759 However, you can force Valgrind to read it, using the
4760 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4761 make use of such information, if present, to provide source-level
4762 descriptions of data addresses in the error messages they create.
4764 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4765 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).