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 * Make the address space limit on FreeBSD amd64 128Gbytes
14 (the same as Linux and Solaris, it was 32Gbytes)
16 * ==================== TOOL CHANGES ===================
19 - When doing a delta leak_search, it is now possible to only
20 output the new loss records compared to the previous leak search.
21 This is available in the memcheck monitor command 'leak_search'
22 by specifying the "new" keyword or in your program by using
23 the client request VALGRIND_DO_NEW_LEAK_CHECK.
24 Whenever a "delta" leak search is done (i.e. when specifying
25 "new" or "increased" or "changed" in the monitor command),
26 the new loss records have a "new" marker.
29 - The option ---history-backtrace-size=<number> allows to configure
30 the number of entries to record in the stack traces of "old"
31 accesses. Previously, this number was hardcoded to 8.
33 * ==================== FIXED BUGS ====================
35 The following bugs have been fixed or resolved. Note that "n-i-bz"
36 stands for "not in bugzilla" -- that is, a bug that was reported to us
37 but never got a bugzilla entry. We encourage you to file bugs in
38 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
39 than mailing the developers (or mailing lists) directly -- bugs that
40 are not entered into bugzilla tend to get forgotten about or ignored.
42 170510 Don't warn about ioctl of size 0 without direction hint
43 327548 false positive while destroying mutex
44 351857 confusing error message about valid command line option
45 392331 Spurious lock not held error from inside pthread_cond_timedwait
46 400793 pthread_rwlock_timedwrlock false positive
47 444110 priv/guest_ppc_toIR.c:36198:31: warning: duplicated 'if' condition.
48 444488 Use glibc.pthread.stack_cache_size tunable
49 444568 drd/tests/pth_barrier_thr_cr fails on Fedora 38
50 459476 vgdb: allow address reuse to avoid "address already in use" errorsuse" errors
51 462830 WARNING: unhandled amd64-freebsd syscall: 474
52 463027 broken check for MPX instruction support in assembler
54 To see details of a given bug, visit
55 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
56 where XXXXXX is the bug number as listed above.
58 (3.21.0.RC1: ?? Apr 2023)
60 Release 3.20.0 (24 Oct 2022)
61 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
63 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
64 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
65 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
66 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
67 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
68 AMD64/macOS 10.13 and nanoMIPS/Linux.
70 * ==================== CORE CHANGES ===================
72 * The option "--vgdb-stop-at=event1,event2,..." accepts the new value abexit.
73 This indicates to invoke gdbserver when your program exits abnormally
74 (i.e. with a non zero exit code).
75 * Fix Rust v0 name demangling.
76 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
77 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
78 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with
80 * The option --enable-debuginfod=<no|yes> [default: yes] has been added on
82 * More DWARF5 support as generated by clang14.
84 * ==================== FIXED BUGS ====================
86 The following bugs have been fixed or resolved. Note that "n-i-bz"
87 stands for "not in bugzilla" -- that is, a bug that was reported to us
88 but never got a bugzilla entry. We encourage you to file bugs in
89 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
90 than mailing the developers (or mailing lists) directly -- bugs that
91 are not entered into bugzilla tend to get forgotten about or ignored.
93 131186 writev reports error in (vector[...])
94 434764 iconv_open causes ld.so v2.28+ to use optimised strncmp
95 446754 Improve error codes from alloc functions under memcheck
96 452274 memcheck crashes with Assertion 'sci->status.what == SsIdle' failed
97 452779 Valgrind fails to build on FreeBSD 13.0 with llvm-devel (15.0.0)
98 453055 shared_timed_mutex drd test fails with "Lock shared failed" message
99 453602 Missing command line option to enable/disable debuginfod
100 452802 Handle lld 9+ split RW PT_LOAD segments correctly
101 454040 s390x: False-positive memcheck:cond in memmem on arch13 systems
102 456171 [PATCH] FreeBSD: Don't record address errors when accessing the 'kern.ps_strings' sysctl struct
103 n-i-bz Implement vgdb invoker on FreeBSD
104 458845 PowerPC: The L field for the dcbf and sync instruction should be
106 458915 Remove register cache to fix 458915 gdbserver causes wrong syscall return
107 459031 Documentation on --error-exitcode incomplete
108 459477 XERROR messages lacks ending '\n' in vgdb
109 462007 Implicit int in none/tests/faultstatus.c
111 To see details of a given bug, visit
112 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
113 where XXXXXX is the bug number as listed above.
115 (3.20.0.RC1: 20 Oct 2022)
118 Release 3.19.0 (11 Apr 2022)
119 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
121 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
122 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
123 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
124 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
125 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
126 AMD64/macOS 10.13 and nanoMIPS/Linux.
128 * ==================== CORE CHANGES ===================
130 * Fix Rust v0 name demangling.
131 * The Linux rseq syscall is now implemented as (silently) returning ENOSYS.
132 * Add FreeBSD syscall wrappers for __specialfd and __realpathat.
133 * Remove FreeBSD dependencies on COMPAT10, which fixes compatibility with HardenedBSD
135 * ================== PLATFORM CHANGES =================
138 - ignore the "v8.x" architecture levels, only look at actual CPU features
139 present. Fixes mismatch detected between RDMA and atomics features
140 preventing startup on some QEMU configurations.
141 - Implement LD{,A}XP and ST{,L}XP
142 - Fix incorrect code emitted for doubleword CAS.
145 - Fix sys_ipc semtimedop syscall
146 - Fix VFLRX and WFLRX instructions
147 - Fix EXRL instruction with negative offset
150 - Reimplement the vbpermq instruction support to generate less Iops and
151 avoid overflowing internal buffers.
152 - Fix checking for scv support to avoid "Facility 'SCV' unavailable (12),
153 exception" messages in dmsg.
154 - Fix setting condition code for Vector Compare quad word instructions.
155 - Fix fix lxsibzx, lxsihzx and lxsihzx instructions so they only load
156 their respective sized data.
157 - Fix the prefixed stq instruction in PC relative mode.
159 * ==================== TOOL CHANGES ===================
162 - Speed up --track-origins=yes for large (in the range of hundreds to
163 thousands of megabytes) mmap/munmaps.
165 - Several fixes for new versions of libstd++ using new posix try_lock
168 * ==================== FIXED BUGS ====================
170 The following bugs have been fixed or resolved. Note that "n-i-bz"
171 stands for "not in bugzilla" -- that is, a bug that was reported to us
172 but never got a bugzilla entry. We encourage you to file bugs in
173 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
174 than mailing the developers (or mailing lists) directly -- bugs that
175 are not entered into bugzilla tend to get forgotten about or ignored.
177 403802 leak_cpp_interior fails with some reachable blocks different than expected
178 435732 memcheck/tests/leak_cpp_interior fails with gcc11
179 444242 s390x: Valgrind crashes on EXRL with negative offset
180 444399 arm64: unhandled instruction 0xC87F2D89 (LD{,A}XP and ST{,L}XP).
182 444481 gdb_server test failures on s390x
183 444495 dhat/tests/copy fails on s390x
184 444552 memcheck/tests/sem fails on s390x with glibc 2.34
185 444571 PPC, fix the lxsibzx and lxsihzx so they only load their respective
187 444836 PPC, pstq instruction for R=1 is not storing to the correct address.
188 444925 fexecve syscall wrapper not properly implemented
189 445032 valgrind/memcheck crash with SIGSEGV when SIGVTALRM timer used and
191 445211 Fix out of tree builds
192 445300 [PATCH] Fix building tests with Musl
193 445011 SIGCHLD is sent when valgrind uses debuginfod-find
194 445354 arm64 backend: incorrect code emitted for doubleword CAS
195 445415 arm64 front end: alignment checks missing for atomic instructions
196 445504 Using C++ condition_variable results in bogus "mutex is locked simultaneously by two threads" warning
197 445607 Unhandled amd64-freebsd syscall: 247
198 445668 Inline stack frame generation is broken for Rust binaries
199 445916 Demangle Rust v0 symbols with .llvm suffix
200 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false positives
201 446138 DRD/Helgrind with std::timed_mutex::try_lock_until false positives
202 446281 Add a DRD suppression for fwrite
203 446103 Memcheck: `--track-origins=yes` causes extreme slowdowns for large mmap/munmap
204 446139 DRD/Helgrind with std::shared_timed_mutex::try_lock_until and try_lock_shared_until false
205 446251 TARGET_SIGNAL_THR added to enum target_signal
206 446823 FreeBSD - missing syscalls when using libzm4
207 447991 s390x: Valgrind indicates illegal instruction on wflrx
208 447995 Valgrind segfault on power10 due to hwcap checking code
209 449483 Powerpc: vcmpgtsq., vcmpgtuq,, vcmpequq. instructions not setting the
210 condition code correctly.
211 449672 ppc64 --track-origins=yes failures because of bad cmov addHRegUse
212 449838 sigsegv liburing the 'impossible' happened for io_uring_setup
213 450025 Powerc: ACC file not implemented as a logical overlay of the VSR
215 450437 Warn for execve syscall with argv or argv[0] being NULL
216 450536 Powerpc: valgrind throws 'facility scv unavailable exception'
217 451626 Syscall param bpf(attr->raw_tracepoint.name) points to unaddressable byte(s)
218 451827 [ppc64le] VEX temporary storage exhausted with several vbpermq instructions
219 451843 valgrind fails to start on a FreeBSD system which enforces W^X
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 (3.19.0.RC1: 02 Apr 2022)
226 (3.19.0.RC2: 08 Apr 2022)
229 Release 3.18.0 (15 Oct 2021)
230 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
232 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
233 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
234 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
235 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
236 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
237 AMD64/macOS 10.13 and nanoMIPS/Linux.
239 * ==================== CORE CHANGES ===================
241 * The libiberty demangler has been updated, which brings support for
242 Rust v0 name demangling. [Update: alas, due to a bug, this support
243 isn't working in 3.18.0.]
245 * __libc_freeres isn't called anymore after the program recieves a
246 fatal signal. Causing some internal glibc resources to hang around,
247 but preventing any crashes after the program has ended.
249 * The DWARF reader is now very much faster at startup when just
250 --read-inline-info=yes (the default in most cases) is given.
252 * glibc 2.34, which moved various functions from libpthread.so into
253 libc.so, is now supported.
255 * ================== PLATFORM CHANGES =================
259 - v8.2 scalar and vector FABD, FACGE, FACGT and FADD.
260 - v8.2 FP compare & conditional compare instructions.
261 - Zero variants of v8.2 FP compare instructions.
265 - Support the miscellaneous-instruction-extensions facility 3 and
266 the vector-enhancements facility 2. This enables programs
267 compiled with "-march=arch13" or "-march=z15" to be executed
272 - ISA 3.1 support is now complete
273 - ISA 3.0 support for the darn instruction added.
274 - ISA 3.0 support for the vector system call instruction scv added.
275 - ISA 3.0 support for the copy, paste and cpabort instructions added.
277 * Support for X86/FreeBSD and AMD64/FreeBSD has been added.
279 * ==================== OTHER CHANGES ====================
281 * Memcheck on amd64: minor fixes to remove some false positive
284 * ==================== FIXED BUGS ====================
286 The following bugs have been fixed or resolved. Note that "n-i-bz"
287 stands for "not in bugzilla" -- that is, a bug that was reported to us
288 but never got a bugzilla entry. We encourage you to file bugs in
289 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
290 than mailing the developers (or mailing lists) directly -- bugs that
291 are not entered into bugzilla tend to get forgotten about or ignored.
293 208531 [PATCH]: FreeBSD support for valgrind
294 368960 WARNING: unhandled amd64-linux syscall: 163 (acct)
295 407589 [Linux] Add support for C11 aligned_alloc() and GNU reallocarray()
296 423963 Error in child thread when CLONE_PIDFD is used
297 426148 crash with "impossible happened" when running BPF CO-RE programs
298 429375 PPC ISA 3.1 support is missing, part 9
299 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
300 431306 Update demangler to support Rust v0 name mangling
301 432387 s390x: z15 instructions support
302 433437 FreeBSD support, part 1
303 433438 FreeBSD support, part 2
304 433439 FreeBSD support, part 3
305 433469 FreeBSD support, part 4
306 433473 FreeBSD support, part 5
307 433477 FreeBSD support, part 6
308 433479 FreeBSD support, part 7
309 433504 FreeBSD support, part 8
310 433506 FreeBSD support, part 9
311 433507 FreeBSD support, part 10
312 433508 FreeBSD support, part 11
313 433510 FreeBSD support, part 12
314 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
315 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
316 434296 s390x: False-positive memcheck diagnostics from vector string
318 434840 PPC64 darn instruction not supported
319 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
320 435908 valgrind tries to fetch from deubginfod for files which already
321 have debug information
322 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2
323 439046 valgrind is unusably large when linked with lld
324 439090 Implement close_range(2)
325 439326 Valgrind 3.17.0 won't compile with Intel 2021 oneAPI compilers
326 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
327 440670 unhandled ppc64le-linux syscall: 252 statfs64 and 253 fstatfs64
328 440906 Fix impossible constraint issue in P10 testcase.
329 441512 Remove a unneeded / unnecessary prefix check.
330 441534 Update the expected output for test_isa_3_1_VRT.
331 442061 very slow execution under Fedora 34 (readdwarf3)
332 443031 Gcc -many change requires explicit .machine directives
333 443033 Add support for the ISA 3.0 mcrxrx instruction
334 443034 Sraw, srawi, srad, sradi, mfs
335 443178 Powerpc, test jm-mfspr expected output needs to be updated.
336 443179 Need new test for the lxvx and stxvx instructions on ISA 2.07 and
338 443180 The subnormal test and the ISA 3.0 test generate compiler warnings
339 443314 In the latest GIT version, Valgrind with "--trace-flags" crashes
341 443605 Don't call final_tidyup (__libc_freeres) on FatalSignal
343 To see details of a given bug, visit
344 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
345 where XXXXXX is the bug number as listed below.
347 (3.18.0.RC1: 12 Oct 2021)
348 (3.18.0: 15 Oct 2021)
352 Release 3.17.0 (19 Mar 2021)
353 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
355 3.17.0 fixes a number of bugs and adds some functional changes: support for
356 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
357 some new instructions for Arm64, S390 and POWER. There are also some tool
360 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
361 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
362 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
363 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
364 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
366 * ==================== CORE CHANGES ===================
368 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
371 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
372 debugging information. When a debuginfo file cannot be found locally,
373 Valgrind is able to query debuginfod servers for the file using its
374 build-id. See the user manual for more information about debuginfod support.
376 * ================== PLATFORM CHANGES =================
380 - Inaccuracies resulting from double-rounding in the simulation of
381 floating-point multiply-add/subtract instructions have been fixed. These
382 should now behave exactly as the hardware does.
384 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
385 ongoing. Support for the half-word variants of at least the following
386 instructions has been added:
388 FABS <Vd>.<T>, <Vn>.<T>
390 FNEG <Vd>.<T>, <Vn>.<T>
392 FSQRT <Vd>.<T>, <Vn>.<T>
397 - Implement the new instructions/features that were added to z/Architecture
398 with the vector-enhancements facility 1. Also cover the instructions from
399 the vector-packed-decimal facility that are defined outside the chapter
400 "Vector Decimal Instructions", but not the ones from that chapter itself.
402 For a detailed list of newly supported instructions see the updates to
403 `docs/internals/s390-opcodes.csv'.
405 Since the miscellaneous instruction extensions facility 2 was already
406 added in Valgrind 3.16.0, this completes the support necessary to run
407 general programs built with `--march=z14' under Valgrind. The
408 vector-packed-decimal facility is currently not exploited by the standard
409 toolchain and libraries.
413 - Various bug fixes. Fix for the sync field to limit setting just two of
414 the two bits in the L-field. Fix the write size for the stxsibx and
415 stxsihx instructions. Fix the modsw and modsd instructions.
417 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
418 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
419 conversion instructions are still missing.
421 * ==================== TOOL CHANGES ====================
423 * General tool changes
425 - All the tools and their vgpreload libraries are now installed under
426 libexec because they cannot be executed directly and should be run through
427 the valgrind executable. This should be an internal, not user visible,
428 change, but might impact valgrind packagers.
430 - The --track-fds option now respects -q, --quiet and won't output anything
431 if no file descriptors are leaked. It also won't report the standard stdin
432 (0), stdout (1) or stderr (2) descriptors as being leaked with
433 --trace-fds=yes anymore. To track whether the standard file descriptors
434 are still open at the end of the program run use --trace-fds=all.
438 - DHAT has been extended, with two new modes of operation. The new
439 --mode=copy flag triggers copy profiling, which records calls to memcpy,
440 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
441 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
442 the new dhat/dhat.h file. This is useful for learning more about hot code
443 paths. See the user manual for more information about the new modes.
445 - Because of these changes, DHAT's file format has changed. DHAT output
446 files produced with earlier versions of DHAT will not work with this
447 version of DHAT's viewer, and DHAT output files produced with this version
448 of DHAT will not work with earlier versions of DHAT's viewer.
450 * ==================== FIXED BUGS ====================
452 The following bugs have been fixed or resolved. Note that "n-i-bz"
453 stands for "not in bugzilla" -- that is, a bug that was reported to us
454 but never got a bugzilla entry. We encourage you to file bugs in
455 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
456 than mailing the developers (or mailing lists) directly -- bugs that
457 are not entered into bugzilla tend to get forgotten about or ignored.
459 To see details of a given bug, visit
460 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
461 where XXXXXX is the bug number as listed below.
463 140178 open("/proc/self/exe", ...); doesn't quite work
464 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
465 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
466 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
467 345077 linux syscall execveat support (linux 3.19)
468 361770 Missing F_ADD_SEALS
469 369029 handle linux syscalls sched_getattr and sched_setattr
470 384729 __libc_freeres inhibits cross-platform valgrind
471 388787 Support for C++17 new/delete
472 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
473 396656 Warnings while reading debug info
474 397605 ioctl FICLONE mishandled
475 401416 Compile failure with openmpi 4.0
476 408663 Suppression file for musl libc
477 404076 s390x: z14 vector instructions not implemented
478 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
479 (actually affected all x86 and nanomips regardless of host bitness)
480 413547 regression test does not check for Arm 64 features.
481 414268 Enable AArch64 feature detection and decoding for v8.x instructions
482 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
483 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
484 422261 platform selection fails for unqualified client name
485 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
486 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
487 423195 PPC ISA 3.1 support is missing, part 1
488 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
489 424012 crash with readv/writev having invalid but not NULL arg2 iovec
490 424298 amd64: Implement RDSEED
491 425232 PPC ISA 3.1 support is missing, part 2
492 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
493 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
494 426123 PPC ISA 3.1 support is missing, part 3
495 426144 Fix "condition variable has not been initialized" on Fedora 33.
496 427400 PPC ISA 3.1 support is missing, part 4
497 427401 PPC ISA 3.1 support is missing, part 5
498 427404 PPC ISA 3.1 support is missing, part 6
499 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
500 427787 Support new faccessat2 linux syscall (439)
501 427969 debuginfo section duplicates a section in the main ELF file
502 428035 drd: Unbreak the musl build
503 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
504 428716 cppcheck detects potential leak in VEX/useful/smchash.c
505 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
506 429352 PPC ISA 3.1 support is missing, part 7
507 429354 PPC ISA 3.1 support is missing, part 8
508 429692 unhandled ppc64le-linux syscall: 147 (getsid)
509 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
511 429952 Errors when building regtest with clang
512 430354 ppc stxsibx and stxsihx instructions write too much data
513 430429 valgrind.h doesn't compile on s390x with clang
514 430485 expr_is_guardable doesn't handle Iex_Qop
515 431556 Complete arm64 FADDP v8.2 instruction support
516 432102 Add support for DWARF5 as produced by GCC11
517 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
518 432381 drd: Process STACK_REGISTER client requests
519 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
520 432672 vg_regtest: test-specific environment variables not reset between tests
521 432809 VEX should support REX.W + POPF
522 432861 PPC modsw and modsd give incorrect results for 1 mod 12
523 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
524 432215 Add debuginfod functionality
525 433323 Use pkglibexecdir as vglibdir
526 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
527 433629 valgrind/README has type "abd" instead of "and"
528 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
529 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
530 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
531 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
532 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
534 (3.17.0.RC1: 13 Mar 2021)
535 (3.17.0.RC2: 17 Mar 2021)
536 (3.17.0: 19 Mar 2021)
540 Release 3.16.1 (22 June 2020)
541 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
543 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
544 fixes character encoding problems in the documentation HTML.
546 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
547 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
549 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
553 Release 3.16.0 (27 May 2020)
554 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
556 3.16.0 is a feature release with many improvements and the usual collection of
559 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
560 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
561 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
562 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
563 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
565 * ==================== CORE CHANGES ===================
567 * It is now possible to dynamically change the value of many command line
568 options while your program (or its children) are running under Valgrind.
570 To see the list of dynamically changeable options, run
571 "valgrind --help-dyn-options".
573 You can change the options from the shell by using vgdb to launch
574 the monitor command "v.clo <clo option>...".
575 The same monitor command can be used from a gdb connected
576 to the valgrind gdbserver.
577 Your program can also change the dynamically changeable options using
578 the client request VALGRIND_CLO_CHANGE(option).
580 * ================== PLATFORM CHANGES =================
582 * MIPS: preliminary support for nanoMIPS instruction set has been added.
584 * ==================== TOOL CHANGES ====================
588 - The implicit memcpy done by each call to realloc now counts towards the
589 read and write counts of resized heap blocks, making those counts higher
594 - cg_annotate's --auto and --show-percs options now default to 'yes', because
595 they are usually wanted.
599 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
600 because they are usually wanted.
602 - The command option --collect-systime has been enhanced to specify
603 the unit used to record the elapsed time spent during system calls.
604 The command option now accepts the values no|yes|msec|usec|nsec,
605 where yes is a synonym of msec. When giving the value nsec, the
606 system cpu time of system calls is also recorded.
610 - Several memcheck options are now dynamically changeable.
611 Use valgrind --help-dyn-options to list them.
613 - The release 3.15 introduced a backward incompatible change for
614 some suppression entries related to preadv and pwritev syscalls.
615 When reading a suppression entry using the unsupported 3.14 format,
616 valgrind will now produce a warning to say the suppression entry will not
617 work, and suggest the needed change.
619 - Significantly fewer false positive errors on optimised code generated by
620 Clang and GCC. In particular, Memcheck now deals better with the
621 situation where the compiler will transform C-level "A && B" into "B && A"
622 under certain circumstances (in which the transformation is valid).
623 Handling of integer equality/non-equality checks on partially defined
624 values is also improved on some architectures.
628 - The exprimental Stack and Global Array Checking tool has been removed.
629 It only ever worked on x86 and amd64, and even on those it had a
630 high false positive rate and was slow. An alternative for detecting
631 stack and global array overruns is using the AddressSanitizer (ASAN)
632 facility of the GCC and Clang compilers, which require you to rebuild
633 your code with -fsanitize=address.
635 * ==================== OTHER CHANGES ====================
637 * New and modified GDB server monitor features:
639 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
641 - The gdbserver monitor commands that require an address and an optional
642 length argument now accepts the alternate 'C like' syntax "address[length]".
643 For example, the memcheck command "monitor who_points_at 0x12345678 120"
644 can now also be given as "monitor who_points_at 0x12345678[120]".
646 * ==================== FIXED BUGS ====================
648 The following bugs have been fixed or resolved. Note that "n-i-bz"
649 stands for "not in bugzilla" -- that is, a bug that was reported to us
650 but never got a bugzilla entry. We encourage you to file bugs in
651 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
652 than mailing the developers (or mailing lists) directly -- bugs that
653 are not entered into bugzilla tend to get forgotten about or ignored.
655 To see details of a given bug, visit
656 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
657 where XXXXXX is the bug number as listed below.
659 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
660 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
661 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
662 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
663 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
664 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
665 400872 Add nanoMIPS support to Valgrind
666 403212 drd/tests/trylock hangs on FreeBSD
667 404406 s390x: z14 miscellaneous instructions not implemented
668 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
669 406561 mcinfcallWSRU gdbserver_test fails on ppc64
670 406824 Unsupported baseline
671 407218 Add support for the copy_file_range syscall
672 407307 Intercept stpcpy also in ld.so for arm64
673 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
675 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
676 408009 Expose rdrand and f16c even on avx if host cpu supports them
677 408091 Missing pkey syscalls
678 408414 Add support for missing for preadv2 and pwritev2 syscalls
679 409141 Valgrind hangs when SIGKILLed
680 409206 Support for Linux PPS and PTP ioctls
681 409367 exit_group() after signal to thread waiting in futex() causes hangs
682 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
683 409780 References to non-existent configure.in
684 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
685 410599 Non-deterministic behaviour of pth_self_kill_15_other test
686 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
687 411134 Allow the user to change a set of command line options during execution
688 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
689 412344 Problem setting mips flags with specific paths
690 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
691 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
692 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
693 413603 callgrind_annotate/cg_annotate truncate function names at '#'
694 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
695 415136 ARMv8.1 Compare-and-Swap instructions are not supported
696 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
697 416239 valgrind crashes when handling clock_adjtime
698 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
699 416286 DRD reports "conflicting load" error on std::mutex::lock()
700 416301 s390x: "compare and signal" not supported
701 416387 finit_module and bpf syscalls are unhandled on arm64
702 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
703 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
704 416753 new 32bit time syscalls for 2038+
705 417075 pwritev(vector[...]) suppression ignored
706 417075 is not fixed, but incompatible supp entries are detected
707 and a warning is produced for these.
708 417187 [MIPS] Conditional branch problem since 'grail' changes
709 417238 Test memcheck/tests/vbit-test fails on mips64 BE
710 417266 Make memcheck/tests/linux/sigqueue usable with musl
711 417281 s390x: /bin/true segfaults with "grail" enabled
712 417427 commit to fix vki_siginfo_t definition created numerous regression
714 417452 s390_insn_store_emit: dst->tag for HRcVec128
715 417578 Add suppressions for glibc DTV leaks
716 417906 clone with CLONE_VFORK and no CLONE_VM fails
717 418004 Grail code additions break ppc64.
718 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
719 418997 s390x: Support Iex_ITE for float and vector types
720 419503 s390x: Avoid modifying registers returned from isel functions
721 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
722 421570 std_mutex fails on Arm v8.1 h/w
723 434035 vgdb might crash if valgrind is killed
724 n-i-bz Fix minor one time leaks in dhat.
725 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
726 n-i-bz Add support for the Linux io_uring system calls
727 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
728 n-i-bz Fix non-glibc build of test suite with s390x_features
729 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
730 423195 PPC ISA 3.1 support is missing, part 1
732 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
733 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
734 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
738 Release 3.15.0 (12 April 2019)
739 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
741 3.15.0 is a feature release with many improvements and the usual collection of
744 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
745 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
746 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
747 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
748 support for X86/macOS 10.13 and AMD64/macOS 10.13.
750 * ==================== CORE CHANGES ===================
752 * The XTree Massif output format now makes use of the information obtained
753 when specifying --read-inline-info=yes.
755 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
757 * ==================== TOOL CHANGES ====================
761 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
762 it has been promoted from an experimental tool to a regular tool. Run it
763 with --tool=dhat instead of --tool=exp-dhat.
765 - DHAT now prints only minimal data when the program ends, instead writing
766 the bulk of the profiling data to a file. As a result, the --show-top-n
767 and --sort-by options have been removed.
769 - Profile results can be viewed with the new viewer, dh_view.html. When
770 a run ends, a short message is printed, explaining how to view the result.
772 - See the documentation for more details.
776 - cg_annotate has a new option, --show-percs, which prints percentages next
781 - callgrind_annotate has a new option, --show-percs, which prints percentages
782 next to all event counts.
784 - callgrind_annotate now inserts commas in call counts, and
785 sort the caller/callee lists in the call tree.
789 - The default value for --read-inline-info is now "yes" on
790 Linux/Android/Solaris. It is still "no" on other OS.
794 - The option --xtree-leak=yes (to output leak result in xtree format)
795 automatically activates the option --show-leak-kinds=all, as xtree
796 visualisation tools such as kcachegrind can in any case select what kind
797 of leak to visualise.
799 - There has been further work to avoid false positives. In particular,
800 integer equality on partially defined inputs (C == and !=) is now handled
803 * ==================== OTHER CHANGES ====================
805 * The new option --show-error-list=no|yes displays, at the end of the run, the
806 list of detected errors and the used suppressions. Prior to this change,
807 showing this information could only be done by specifying "-v -v", but that
808 also produced a lot of other possibly-non-useful messages. The option -s is
809 equivalent to --show-error-list=yes.
811 * ==================== FIXED BUGS ====================
813 The following bugs have been fixed or resolved. Note that "n-i-bz"
814 stands for "not in bugzilla" -- that is, a bug that was reported to us
815 but never got a bugzilla entry. We encourage you to file bugs in
816 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
817 than mailing the developers (or mailing lists) directly -- bugs that
818 are not entered into bugzilla tend to get forgotten about or ignored.
820 To see details of a given bug, visit
821 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
822 where XXXXXX is the bug number as listed below.
824 385411 s390x: z13 vector floating-point instructions not implemented
825 397187 z13 vector register support for vgdb gdbserver
826 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
827 398870 Please add support for instruction vcvtps2ph
828 399287 amd64 front end: Illegal Instruction vcmptrueps
829 399301 Use inlined frames in Massif XTree output.
830 399322 Improve callgrind_annotate output
831 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
832 400164 helgrind test encounters mips x-compiler warnings and assembler error
833 400490 s390x: VRs allocated as if separate from FPRs
834 400491 s390x: Operand of LOCH treated as unsigned integer
835 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
836 options, which specify a mips64 processor
837 401112 LLVM 5.0 generates comparison against partially initialized data
838 401277 More bugs in z13 support
839 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
840 401578 drd: crashes sometimes on fork()
841 401627 memcheck errors with glibc avx2 optimized wcsncmp
842 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
843 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
844 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
846 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
847 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
848 402123 invalid assembler opcodes for mips32r2
849 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
850 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
851 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
852 402351 mips64 libvexmultiarch_test fails on s390x
854 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
855 402480 Do not use %rsp in clobber list
856 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
857 402515 Implement new option --show-error-list=no|yes / -s
858 402519 POWER 3.0 addex instruction incorrectly implemented
859 402781 Redo the cache used to process indirect branch targets
860 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
861 403552 s390x: wrong facility bit checked for vector facility
862 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
863 404638 Add VG_(replaceIndexXA)
864 404843 s390x: backtrace sometimes ends prematurely
865 404888 autotools cleanup series
866 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
867 405182 Valgrind fails to build with Clang
868 405205 filter_libc: remove the line holding the futex syscall error entirely
869 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
870 the upper and lower 32-bits of the 64-bit result
871 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
872 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
873 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
875 405403 s390x disassembler cannot be used on x86
876 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
877 405458 MIPS mkFormVEC arguments swapped?
878 405716 drd: Fix an integer overflow in the stack margin calculation
879 405722 Support arm64 core dump
880 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
881 of the 64-bit destination field.
882 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
883 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
884 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
886 406256 PPC64, vector floating point instructions don't handle subnormal
887 according to VSCR[NJ] bit setting.
888 406352 cachegrind/callgrind fails ann tests because of missing a.c
889 406354 dhat is broken on x86 (32bit)
890 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
891 406357 gdbserver_tests fails because of gdb output change
892 406360 memcheck/tests/libstdc++.supp needs more supression variants
893 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
894 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
895 407340 PPC64, does not support the vlogefp, vexptefp instructions.
896 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
897 n-i-bz Fix callgrind_annotate non deterministic order for equal total
898 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
899 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
900 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
902 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
903 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
904 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
908 Release 3.14.0 (9 October 2018)
909 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
911 3.14.0 is a feature release with many improvements and the usual collection of
914 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
915 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
916 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
917 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
918 support for X86/macOS 10.13, AMD64/macOS 10.13.
920 * ==================== CORE CHANGES ===================
922 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
923 debug info for unloaded code. This allows saved stack traces (e.g. for
924 memory leaks) to include file/line info for code that has been dlclose'd (or
925 similar). See the user manual for more information and known limitations.
927 * Ability to specify suppressions based on source file name and line number.
929 * Majorly overhauled register allocator. No end-user changes, but the JIT
930 generates code a bit more quickly now.
932 * ================== PLATFORM CHANGES =================
934 * Preliminary support for macOS 10.13 has been added.
936 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
938 * mips: support for MIPS SIMD architecture (MSA) has been added.
940 * mips: support for MIPS N32 ABI has been added.
942 * s390: partial support for vector instructions (integer and string) has been
945 * ==================== TOOL CHANGES ====================
947 * Helgrind: Addition of a flag
948 --delta-stacktrace=no|yes [yes on linux amd64/x86]
949 which specifies how full history stack traces should be computed.
950 Setting this to =yes can speed up Helgrind by 25% when using
951 --history-level=full.
953 * Memcheck: reduced false positive rate for optimised code created by Clang 6
954 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
955 blocks more carefully to determine where it can avoid expensive definedness
956 checks without loss of precision. This is controlled by the flag
957 --expensive-definedness-checks=no|auto|yes [auto].
959 * ==================== OTHER CHANGES ====================
961 * Valgrind is now buildable with link-time optimisation (LTO). A new
962 configure option --enable-lto=yes allows building Valgrind with LTO. If the
963 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
964 Note that if you are doing Valgrind development, --enable-lto=yes massively
965 slows down the build process.
967 * ==================== FIXED BUGS ====================
969 The following bugs have been fixed or resolved. Note that "n-i-bz"
970 stands for "not in bugzilla" -- that is, a bug that was reported to us
971 but never got a bugzilla entry. We encourage you to file bugs in
972 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
973 than mailing the developers (or mailing lists) directly -- bugs that
974 are not entered into bugzilla tend to get forgotten about or ignored.
976 To see details of a given bug, visit
977 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
978 where XXXXXX is the bug number as listed below.
980 79362 Debug info is lost for .so files when they are dlclose'd
981 208052 strlcpy error when n = 0
982 255603 exp-sgcheck Assertion '!already_present' failed
983 338252 building valgrind with -flto (link time optimisation) fails
984 345763 MIPS N32 ABI support
985 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
986 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
987 372347 Replacement problem of the additional c++14/c++17 new/delete operators
988 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
989 376257 helgrind history full speed up using a cached stack
990 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
992 379748 Fix missing pselect syscall (OS X 10.11)
993 379754 Fix missing syscall ulock_wait (OS X 10.12)
994 380397 s390x: __GI_strcspn() replacemenet needed
995 381162 possible array overrun in VEX register allocator
996 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
997 381274 powerpc too chatty even with --sigill-diagnostics=no
998 381289 epoll_pwait can have a NULL sigmask
999 381553 VEX register allocator v3
1000 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
1001 381769 Use ucontext_t instead of struct ucontext
1002 381805 arm32 needs ld.so index hardwire for new glibc security fixes
1003 382256 gz compiler flag test doesn't work for gold
1004 382407 vg_perf needs "--terse" command line option
1005 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
1006 382563 MIPS MSA ASE support
1007 382998 xml-socket doesn't work
1008 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
1009 383723 Fix missing kevent_qos syscall (macOS 10.11)
1010 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
1011 384096 Mention AddrCheck at Memcheck's command line option [..]
1012 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
1013 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
1014 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
1015 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
1016 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
1017 384337 performance improvements to VEX register allocator v2 and v3
1018 384526 reduce number of spill insns generated by VEX register allocator v3
1019 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
1020 384631 Sanitise client args as printed with -v
1021 384633 Add a simple progress-reporting facility
1022 384987 VEX regalloc: allocate caller-save registers for short lived vregs
1023 385055 PPC VEX temporary storage exhausted
1024 385182 PPC64 is missing support for the DSCR
1025 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
1026 385207 PPC64, generate_store_FPRF() generates too many Iops
1027 385208 PPC64, xxperm instruction exhausts temporary memory
1028 385210 PPC64, vpermr instruction could exhaust temporary memory
1029 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
1030 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
1031 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
1032 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
1033 385408 s390x: z13 vector "support" instructions not implemented
1034 385409 s390x: z13 vector integer instructions not implemented
1035 385410 s390x: z13 vector string instructions not implemented
1036 385412 s390x: new non-vector z13 instructions not implemented
1037 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
1038 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
1039 385939 Optionally exit on the first error
1040 386318 valgrind.org/info/tools.html is missing SGCheck
1041 386425 running valgrind + wine on armv7l gives illegal opcode
1042 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
1043 387410 MIPSr6 support
1044 387664 Memcheck: make expensive-definedness-checks be the default
1045 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
1046 387766 asm shifts cause false positive "Conditional jump or move depends
1047 on uninitialised value"
1048 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
1049 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
1050 388786 Support bpf syscall in amd64 Linux
1051 388862 Add replacements for wmemchr and wcsnlen on Linux
1052 389065 valgrind meets gcc flag -Wlogical-op
1053 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
1054 390471 suppression by specification of source-file line number
1055 390723 make xtree dump files world wide readable, similar to log files
1056 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
1057 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
1058 392118 unhandled amd64-linux syscall: 332 (statx)
1059 392449 callgrind not clearing the number of calls properly
1060 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
1061 lxssp, stxssp and stxvl instructions.
1062 393023 callgrind_control risks using the wrong vgdb
1063 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
1064 393099 posix_memalign() invalid write if alignment == 0
1065 393146 failing assert "is_DebugInfo_active(di)"
1066 395709 PPC64 is missing support for the xvnegsp instruction
1067 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
1069 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
1070 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
1071 396839 s390x: Trap instructions not implemented
1072 396887 arch_prctl should return EINVAL on unknown option
1073 == 397286 crash before launching binary (Unsupported arch_prctl option)
1074 == 397393 valgrind: the 'impossible' happened: (Archlinux)
1075 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
1076 396906 compile tests failure on mips32-linux: broken inline asm in tests on
1078 397012 glibc ld.so uses arch_prctl on i386
1079 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
1080 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
1081 397424 glibc 2.27 and gdb_server tests
1082 398028 Assertion `cfsi_fits` failing in simple C program
1083 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
1085 n-i-bz Fix missing workq_ops operations (macOS)
1086 n-i-bz fix bug in strspn replacement
1087 n-i-bz Add support for the Linux BLKFLSBUF ioctl
1088 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
1089 n-i-bz Fix possible stack trashing by semctl syscall wrapping
1090 n-i-bz Add support for the Linux membarrier() system call
1091 n-i-bz x86 front end: recognise and handle UD2 correctly
1092 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
1093 correctly aligned before entering the handler.
1095 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
1096 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
1097 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
1101 Release 3.13.0 (15 June 2017)
1102 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1104 3.13.0 is a feature release with many improvements and the usual collection of
1107 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1108 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1109 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
1110 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
1112 * ==================== CORE CHANGES ===================
1114 * The translation cache size has been increased to keep up with the demands of
1115 large applications. The maximum number of sectors has increased from 24 to
1116 48. The default number of sectors has increased from 16 to 32 on all
1117 targets except Android, where the increase is from 6 to 12.
1119 * The amount of memory that Valgrind can use has been increased from 64GB to
1120 128GB. In particular this means your application can allocate up to about
1121 60GB when running on Memcheck.
1123 * Valgrind's default load address has been changed from 0x3800'0000 to
1124 0x5800'0000, so as to make it possible to load larger executables. This
1125 should make it possible to load executables of size at least 1200MB.
1127 * A massive spaceleak caused by reading compressed debuginfo files has been
1128 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
1131 * The C++ demangler has been updated.
1133 * Support for demangling Rust symbols has been added.
1135 * A new representation of stack traces, the "XTree", has been added. An XTree
1136 is a tree of stacktraces with data associated with the stacktraces. This is
1137 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
1138 consumption of your program. Reporting is controlled by the new options
1139 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
1141 A report can also be produced on demand using the gdbserver monitor command
1142 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
1143 format' and 'massif format. The existing visualisers for these formats (e.g.
1144 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
1145 analyse these reports.
1147 Memcheck can also produce XTree leak reports using the Callgrind file
1148 format. For more details, see the user manual.
1150 * ================== PLATFORM CHANGES =================
1152 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
1154 * amd64: fixes for JIT failure problems on long AVX2 code blocks
1156 * amd64 and x86: support for CET prefixes has been added
1158 * arm32: a few missing ARMv8 instructions have been implemented
1160 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
1161 Store-Conditional instructions has been added. This is to deal with
1162 processor implementations that implement the LL/SC specifications strictly
1163 and as a result cause Valgrind to hang in certain situations. The
1164 alternative implementation is automatically enabled at startup, as required.
1165 You can use the option --sim-hints=fallback-llsc to force-enable it if you
1168 * Support for OSX 10.12 has been improved.
1170 * On Linux, clone handling has been improved to honour CLONE_VFORK that
1171 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
1172 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
1173 CLONE_VM exact semantics will (still) not work.
1175 * The TileGX/Linux port has been removed because it appears to be both unused
1178 * ==================== TOOL CHANGES ====================
1182 - Memcheck should give fewer false positives when running optimised
1183 Clang/LLVM generated code.
1185 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1187 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
1188 to produce the end of execution leak report in a xtree callgrind format
1191 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
1192 the leak report in an xtree file.
1196 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1198 - For some workloads (typically, for big applications), Massif memory
1199 consumption and CPU consumption has decreased significantly.
1203 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1205 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
1206 for Ada gnat compiled applications.
1208 * ==================== OTHER CHANGES ====================
1210 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
1211 append the inner guest stacktrace to the inner host stacktrace. This helps
1212 to investigate the errors reported by the outer, when they are caused by the
1213 inner guest program (such as an inner regtest). See README_DEVELOPERS for
1216 * To allow fast detection of callgrind files by desktop environments and file
1217 managers, the format was extended to have an optional first line that
1218 uniquely identifies the format ("# callgrind format"). Callgrind creates
1219 this line now, as does the new xtree functionality.
1221 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
1222 have a new %n format letter that is replaced by a sequence number.
1224 * "--version -v" now shows the SVN revision numbers from which Valgrind was
1227 * ==================== FIXED BUGS ====================
1229 The following bugs have been fixed or resolved. Note that "n-i-bz"
1230 stands for "not in bugzilla" -- that is, a bug that was reported to us
1231 but never got a bugzilla entry. We encourage you to file bugs in
1232 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1233 than mailing the developers (or mailing lists) directly -- bugs that
1234 are not entered into bugzilla tend to get forgotten about or ignored.
1236 To see details of a given bug, visit
1237 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1238 where XXXXXX is the bug number as listed below.
1240 162848 --log-file output isn't split when a program forks
1241 340777 Illegal instruction on mips (ar71xx)
1242 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
1243 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
1244 to a different stack.
1245 344139 x86 stack-seg overrides, needed by the Wine people
1246 344524 store conditional of guest applications always fail - observed on
1248 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
1249 352395 Please provide SVN revision info in --version -v
1250 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
1251 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
1252 INVALID_POSIX_THREADID' failed
1253 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
1254 barrier implementation
1255 358697 valgrind.h: Some code remains even when defining NVALGRIND
1256 359202 Add musl libc configure/compile
1257 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
1258 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
1259 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
1260 362223 assertion failed when .valgrindrc is a directory instead of a file
1261 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
1262 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
1263 368507 can't malloc chunks larger than about 34GB
1264 368529 Android arm target link error, missing atexit and pthread_atfork
1265 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
1266 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
1267 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
1268 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
1269 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
1270 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
1271 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
1272 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
1273 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
1274 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
1275 370028 Reduce the number of compiler warnings on MIPS platforms
1276 370635 arm64 missing syscall getcpu
1277 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
1278 371227 Clean AArch64 syscall table
1279 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
1280 371471 Valgrind complains about non legit memory leaks on placement new (C++)
1281 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
1282 371503 disInstr(arm64): unhandled instruction 0xF89F0000
1283 371869 support '%' in symbol Z-encoding
1284 371916 execution tree xtree concept
1285 372120 c++ demangler demangles symbols which are not c++
1286 372185 Support of valgrind on ARMv8 with 32 bit executable
1287 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1288 372195 Power PC, xxsel instruction is not always recognized.
1289 372504 Hanging on exit_group
1290 372600 process loops forever when fatal signals are arriving quickly
1291 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1292 373046 Stacks registered by core are never deregistered
1293 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1294 373086 Implement additional Xen hypercalls
1295 373192 Calling posix_spawn in glibc 2.24 completely broken
1296 373488 Support for fanotify API on ARM64 architecture
1297 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1298 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1299 373938 const IRExpr arguments for matchIRExpr()
1300 374719 some spelling fixes
1301 374963 increase valgrind's load address to prevent mmap failure
1302 375514 valgrind_get_tls_addr() does not work in case of static TLS
1303 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1304 for ML_(find_rx_mapping)()
1305 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1306 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1307 == 377159 "vex: the `impossible' happened" still present
1308 == 375150 Assertion 'tres.status == VexTransOK' failed
1309 == 378068 valgrind crashes on AVX2 function in FFmpeg
1310 376142 Segfaults on MIPS Cavium Octeon boards
1311 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1312 376455 Solaris: unhandled syscall lgrpsys(180)
1313 376518 Solaris: unhandled fast trap getlgrp(6)
1314 376611 ppc64 and arm64 don't know about prlimit64 syscall
1315 376729 PPC64, remove R2 from the clobber list
1317 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1318 to be wrongly marked as addressable
1319 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1320 PIE enabled by default
1321 377376 memcheck/tests/linux/getregset fails with glibc2.24
1322 377427 PPC64, lxv instruction failing on odd destination register
1323 377478 PPC64: ISA 3.0 setup fixes
1324 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1325 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1326 and 2 args for FUTEX_TRYLOCK_PI
1327 377717 Fix massive space leak when reading compressed debuginfo sections
1328 377891 Update Xen 4.6 domctl wrappers
1329 377930 fcntl syscall wrapper is missing flock structure check
1330 378524 libvexmultiarch_test regression on s390x and ppc64
1331 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1332 378673 Update libiberty demangler
1333 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1334 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1335 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1336 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1337 (task_register_dyld_image_infos)
1338 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1339 (task_register_dyld_shared_cache_image_info)
1340 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1341 379473 MIPS: add support for rdhwr cycle counter register
1342 379504 remove TileGX/Linux port
1343 379525 Support more x86 nop opcodes
1344 379838 disAMode(x86): not an addr!
1345 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1346 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1347 379895 clock_gettime does not execute POST syscall wrapper
1348 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1349 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1350 380200 xtree generated callgrind files refer to files without directory name
1351 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1352 380397 s390x: __GI_strcspn() replacement needed
1353 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1355 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1356 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1357 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1361 Release 3.12.0 (20 October 2016)
1362 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1364 3.12.0 is a feature release with many improvements and the usual
1365 collection of bug fixes.
1367 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1368 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1369 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1370 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1371 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1372 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1374 * ================== PLATFORM CHANGES =================
1376 * POWER: Support for ISA 3.0 has been added
1378 * mips: support for O32 FPXX ABI has been added.
1379 * mips: improved recognition of different processors
1380 * mips: determination of page size now done at run time
1382 * amd64: Partial support for AMD FMA4 instructions.
1384 * arm, arm64: Support for v8 crypto and CRC instructions.
1386 * Improvements and robustification of the Solaris port.
1388 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1390 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1391 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1392 where possible. Valgrind's support for 32-bit x86 has stagnated in
1393 recent years and has fallen far behind that for 64-bit x86
1394 instructions. By contrast 64-bit x86 is well supported, up to and
1397 * ==================== TOOL CHANGES ====================
1401 - Added meta mempool support for describing a custom allocator which:
1402 - Auto-frees all chunks assuming that destroying a pool destroys all
1404 - Uses itself to allocate other memory blocks
1406 - New flag --ignore-range-below-sp to ignore memory accesses below
1407 the stack pointer, if you really have to. The related flag
1408 --workaround-gcc296-bugs=yes is now deprecated. Use
1409 --ignore-range-below-sp=1024-1 as a replacement.
1413 - Improved thread startup time significantly on non-Linux platforms.
1417 - Added collection of the metric "tot-blocks-allocd"
1419 * ==================== OTHER CHANGES ====================
1421 * Replacement/wrapping of malloc/new related functions is now done not just
1422 for system libraries by default, but for any globally defined malloc/new
1423 related function (both in shared libraries and statically linked alternative
1424 malloc implementations). The dynamic (runtime) linker is excluded, though.
1425 To only intercept malloc/new related functions in
1426 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1427 "nouserintercepts" can be any non-existing library name).
1428 This new functionality is not implemented for MacOS X.
1430 * The maximum number of callers in a suppression entry is now equal to
1431 the maximum size for --num-callers (500).
1432 Note that --gen-suppressions=yes|all similarly generates suppressions
1433 containing up to --num-callers frames.
1435 * New and modified GDB server monitor features:
1437 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1438 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1441 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1442 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1445 * Valgrind is able to read compressed debuginfo sections in two formats:
1446 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1447 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1449 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1450 for the most common use case (x86_64-linux, Memcheck) has been
1453 * Improved performance for programs that do a lot of discarding of
1454 instruction address ranges of 8KB or less.
1456 * The C++ symbol demangler has been updated.
1458 * More robustness against invalid syscall parameters on Linux.
1460 * ==================== FIXED BUGS ====================
1462 The following bugs have been fixed or resolved. Note that "n-i-bz"
1463 stands for "not in bugzilla" -- that is, a bug that was reported to us
1464 but never got a bugzilla entry. We encourage you to file bugs in
1465 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1466 than mailing the developers (or mailing lists) directly -- bugs that
1467 are not entered into bugzilla tend to get forgotten about or ignored.
1469 To see details of a given bug, visit
1470 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1471 where XXXXXX is the bug number as listed below.
1473 191069 Exiting due to signal not reported in XML output
1474 199468 Suppressions: stack size limited to 25
1475 while --num-callers allows more frames
1476 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1477 278744 cvtps2pd with redundant RexW
1478 303877 valgrind doesn't support compressed debuginfo sections.
1479 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1480 348345 Assertion fails for negative lineno
1481 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1482 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1483 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1484 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1485 352197 mips: mmap2() not wrapped correctly for page size > 4096
1486 353083 arm64 doesn't implement various xattr system calls
1487 353084 arm64 doesn't support sigpending system call
1488 353137 www: update info for Supported Platforms
1489 353138 www: update "The Valgrind Developers" page
1490 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1493 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1494 353398 WARNING: unhandled amd64-solaris syscall: 207
1495 353660 XML in auxwhat tag not escaping reserved symbols properly
1496 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1497 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1498 353802 ELF debug info reader confused with multiple .rodata sections
1499 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1500 353917 unhandled amd64-solaris syscall fchdir(120)
1501 353920 unhandled amd64-solaris syscall: 170
1502 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1503 354392 unhandled amd64-solaris syscall: 171
1504 354797 Vbit test does not include Iops for Power 8 instruction support
1505 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1509 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1510 355188 valgrind should intercept all malloc related global functions
1511 355454 do not intercept malloc related symbols from the runtime linker
1512 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1513 356044 Dwarf line info reader misinterprets is_stmt register
1514 356112 mips: replace addi with addiu
1515 356393 valgrind (vex) crashes because isZeroU happened
1518 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1519 356678 arm64-linux: unhandled syscall 232 (mincore)
1520 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1521 356823 Unsupported ARM instruction: stlex
1522 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1523 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1524 357673 crash if I try to run valgrind with a binary link with libcurl
1525 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1526 357871 pthread_spin_destroy not properly wrapped
1527 357887 Calls to VG_(fclose) do not close the file descriptor
1528 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1529 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1530 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1531 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1532 359181 Buffer Overflow during Demangling
1533 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1534 359289 s390x: popcnt (B9E1) not implemented
1535 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1536 359503 Add missing syscalls for aarch64 (arm64)
1537 359645 "You need libc6-dbg" help message could be more helpful
1538 359703 s390: wire up separate socketcalls system calls
1539 359724 getsockname might crash - deref_UInt should call safe_to_deref
1540 359733 amd64 implement ld.so strchr/index override like x86
1541 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1542 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1544 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1545 359871 Incorrect mask handling in ppoll
1546 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1547 360008 Contents of Power vr registers contents is not printed correctly when
1548 the --vgdb-shadow-registers=yes option is used
1549 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1550 non-zero shadow bits
1551 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1552 360425 arm64 unsupported instruction ldpsw
1554 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1555 360571 Error about the Android Runtime reading below the stack pointer on ARM
1556 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1557 360749 kludge for multiple .rodata sections on Solaris no longer needed
1558 360752 raise the number of reserved fds in m_main.c from 10 to 12
1559 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1560 361226 s390x: risbgn (EC59) not implemented
1561 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1562 361354 ppc64[le]: wire up separate socketcalls system calls
1563 361615 Inconsistent termination for multithreaded process terminated by signal
1564 361926 Unhandled Solaris syscall: sysfs(84)
1565 362009 V dumps core on unimplemented functionality before threads are created
1566 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1567 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1568 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1569 362953 Request for an update to the Valgrind Developers page
1570 363680 add renameat2() support
1571 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1572 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1573 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1574 364058 clarify in manual limitations of array overruns detections
1575 364413 pselect sycallwrapper mishandles NULL sigmask
1576 364728 Power PC, missing support for several HW registers in
1577 get_otrack_shadow_offset_wrk()
1578 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1579 365273 Invalid write to stack location reported after signal handler runs
1580 365912 ppc64BE segfault during jm-insns test (RELRO)
1581 366079 FPXX Support for MIPS32 Valgrind
1582 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1583 366344 Multiple unhandled instruction for Aarch64
1584 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1585 367995 Integration of memcheck with custom memory allocator
1586 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1587 368412 False positive result for altivec capability check
1588 368416 Add tc06_two_races_xml.exp output for ppc64
1589 368419 Perf Events ioctls not implemented
1590 368461 mmapunmap test fails on ppc64
1591 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1592 369000 AMD64 fma4 instructions unsupported.
1593 369169 ppc64 fails jm_int_isa_2_07 test
1594 369175 jm_vec_isa_2_07 test crashes on ppc64
1595 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1596 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1597 369359 msghdr_foreachfield can crash when handling bad iovec
1598 369360 Bad sigprocmask old or new sets can crash valgrind
1599 369361 vmsplice syscall wrapper crashes on bad iovec
1600 369362 Bad sigaction arguments crash valgrind
1601 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1602 369402 Bad set/get_thread_area pointer crashes valgrind
1603 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1604 369446 valgrind crashes on unknown fcntl command
1605 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1606 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1607 370265 ISA 3.0 HW cap stuff needs updating
1608 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1609 372195 Power PC, xxsel instruction is not always recognized
1611 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1612 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1613 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1614 n-i-bz Fix ppoll_alarm exclusion on OS X
1615 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1616 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1617 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1618 n-i-bz amd64: memcheck false positive with shr %edx
1619 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1620 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1621 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1622 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1624 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1625 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1626 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1630 Release 3.11.0 (22 September 2015)
1631 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1633 3.11.0 is a feature release with many improvements and the usual
1634 collection of bug fixes.
1636 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1637 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1638 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1639 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1640 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1641 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1643 * ================== PLATFORM CHANGES =================
1645 * Support for Solaris/x86 and Solaris/amd64 has been added.
1647 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1649 * Preliminary support for the Tilera TileGX architecture has been added.
1651 * s390x: It is now required for the host to have the "long displacement"
1652 facility. The oldest supported machine model is z990.
1654 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1655 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1657 * The JIT's register allocator is significantly faster, making the JIT
1658 as a whole somewhat faster, so JIT-intensive activities, for example
1659 program startup, are modestly faster, around 5%.
1661 * There have been changes to the default settings of several command
1662 line flags, as detailed below.
1664 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1665 capable hosts, the simulated CPUID will now indicate AVX2 support.
1667 * ==================== TOOL CHANGES ====================
1671 - The default value for --leak-check-heuristics has been changed from
1672 "none" to "all". This helps to reduce the number of possibly
1673 lost blocks, in particular for C++ applications.
1675 - The default value for --keep-stacktraces has been changed from
1676 "malloc-then-free" to "malloc-and-free". This has a small cost in
1677 memory (one word per malloc-ed block) but allows Memcheck to show the
1678 3 stacktraces of a dangling reference: where the block was allocated,
1679 where it was freed, and where it is acccessed after being freed.
1681 - The default value for --partial-loads-ok has been changed from "no" to
1682 "yes", so as to avoid false positive errors resulting from some kinds
1683 of vectorised loops.
1685 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1686 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1687 than get_vbits when you need to associate byte data value with
1688 their corresponding validity bits.
1690 - The 'block_list' monitor command has been enhanced:
1691 o it can print a range of loss records
1692 o it now accepts an optional argument 'limited <max_blocks>'
1693 to control the number of blocks printed.
1694 o if a block has been found using a heuristic, then
1695 'block_list' now shows the heuristic after the block size.
1696 o the loss records/blocks to print can be limited to the blocks
1697 found via specified heuristics.
1699 - The C helper functions used to instrument loads on
1700 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1701 replaced by handwritten assembly sequences. This gives speedups
1702 in the region of 0% to 7% for those targets only.
1704 - A new command line option, --expensive-definedness-checks=yes|no,
1705 has been added. This is useful for avoiding occasional invalid
1706 uninitialised-value errors in optimised code. Watch out for
1707 runtime degradation, as this can be up to 25%. As always, though,
1708 the slowdown is highly application specific. The default setting
1713 - A new monitor command 'all_snapshots <filename>' dumps all
1714 snapshots taken so far.
1718 - Significant memory reduction and moderate speedups for
1719 --history-level=full for applications accessing a lot of memory
1720 with many different stacktraces.
1722 - The default value for --conflict-cache-size=N has been doubled to
1723 2000000. Users that were not using the default value should
1724 preferably also double the value they give.
1726 The default was changed due to the changes in the "full history"
1727 implementation. Doubling the value gives on average a slightly more
1728 complete history and uses similar memory (or significantly less memory
1729 in the worst case) than the previous implementation.
1731 - The Helgrind monitor command 'info locks' now accepts an optional
1732 argument 'lock_addr', which shows information about the lock at the
1735 - When using --history-level=full, the new Helgrind monitor command
1736 'accesshistory <addr> [<len>]' will show the recorded accesses for
1737 <len> (or 1) bytes at <addr>.
1739 * ==================== OTHER CHANGES ====================
1741 * The default value for the --smc-check option has been changed from
1742 "stack" to "all-non-file" on targets that provide automatic D-I
1743 cache coherence (x86, amd64 and s390x). The result is to provide,
1744 by default, transparent support for JIT generated and self-modifying
1745 code on all targets.
1747 * Mac OS X only: the default value for the --dsymutil option has been
1748 changed from "no" to "yes", since any serious usage on Mac OS X
1749 always required it to be "yes".
1751 * The command line options --db-attach and --db-command have been removed.
1752 They were deprecated in 3.10.0.
1754 * When a process dies due to a signal, Valgrind now shows the signal
1755 and the stacktrace at default verbosity (i.e. verbosity 1).
1757 * The address description logic used by Memcheck and Helgrind now
1758 describes addresses in anonymous segments, file mmap-ed segments,
1759 shared memory segments and the brk data segment.
1761 * The new option --error-markers=<begin>,<end> can be used to mark the
1762 begin/end of errors in textual output mode, to facilitate
1763 searching/extracting errors in output files that mix valgrind errors
1764 with program output.
1766 * The new option --max-threads=<number> can be used to change the number
1767 of threads valgrind can handle. The default is 500 threads which
1768 should be more than enough for most applications.
1770 * The new option --valgrind-stacksize=<number> can be used to change the
1771 size of the private thread stacks used by Valgrind. This is useful
1772 for reducing memory use or increasing the stack size if Valgrind
1773 segfaults due to stack overflow.
1775 * The new option --avg-transtab-entry-size=<number> can be used to specify
1776 the expected instrumented block size, either to reduce memory use or
1777 to avoid excessive retranslation.
1779 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1781 * New and modified GDB server monitor features:
1783 - When a signal is reported in GDB, you can now use the GDB convenience
1784 variable $_siginfo to examine detailed signal information.
1786 - Valgrind's gdbserver now allows the user to change the signal
1787 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1788 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1789 continue without passing the signal to the process.
1791 - With GDB >= 7.10, the command 'target remote'
1792 will automatically load the executable file of the process running
1793 under Valgrind. This means you do not need to specify the executable
1794 file yourself, GDB will discover it itself. See GDB documentation about
1795 'qXfer:exec-file:read' packet for more info.
1797 * ==================== FIXED BUGS ====================
1799 The following bugs have been fixed or resolved. Note that "n-i-bz"
1800 stands for "not in bugzilla" -- that is, a bug that was reported to us
1801 but never got a bugzilla entry. We encourage you to file bugs in
1802 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1803 than mailing the developers (or mailing lists) directly -- bugs that
1804 are not entered into bugzilla tend to get forgotten about or ignored.
1806 To see details of a given bug, visit
1807 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1808 where XXXXXX is the bug number as listed below.
1810 116002 VG_(printf): Problems with justification of strings and integers
1811 155125 avoid cutting away file:lineno after long function name
1812 197259 Unsupported arch_prtctl PR_SET_GS option
1813 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
1814 201216 Fix Valgrind does not support pthread_sigmask() on OS X
1815 201435 Fix Darwin: -v does not show kernel version
1816 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
1817 211256 Fixed an outdated comment regarding the default platform.
1818 211529 Incomplete call stacks for code compiled by newer versions of MSVC
1819 211926 Avoid compilation warnings in valgrind.h with -pedantic
1820 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
1822 226609 Crediting upstream authors in man page
1823 231257 Valgrind omits path when executing script from shebang line
1824 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
1825 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
1826 269360 s390x: Fix addressing mode selection for compare-and-swap
1827 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
1829 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
1830 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
1831 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
1832 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
1833 330147 libmpiwrap PMPI_Get_count returns undefined value
1834 333051 mmap of huge pages fails due to incorrect alignment
1836 334802 valgrind does not always explain why a given option is bad
1837 335618 mov.w rN, pc/sp (ARM32)
1838 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
1843 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
1844 338602 AVX2 bit in CPUID missing
1845 338606 Strange message for scripts with invalid interpreter
1846 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
1847 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
1848 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
1850 339156 gdbsrv not called for fatal signal
1851 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
1852 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
1853 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
1854 339442 Fix testsuite build failure on OS X 10.9
1855 339542 Enable compilation with Intel's ICC compiler
1856 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
1857 339688 Mac-specific ASM does not support .version directive (cpuid,
1858 tronical and pushfpopf tests)
1859 339745 Valgrind crash when check Marmalade app (partial fix)
1860 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
1861 339778 Linux/TileGx platform support for Valgrind
1862 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
1863 339789 Fix none/tests/execve test on Mac OS X 10.9
1864 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
1865 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
1866 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
1867 340392 Allow user to select more accurate definedness checking in memcheck
1868 to avoid invalid complaints on optimised code
1869 340430 Fix some grammatical weirdness in the manual.
1870 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
1871 341419 Signal handler ucontext_t not filled out correctly on OS X
1872 341539 VG_(describe_addr) should not describe address as belonging to client
1873 segment if it is past the heap end
1874 341613 Enable building of manythreads and thread-exits tests on Mac OS X
1875 341615 Fix none/tests/darwin/access_extended test on Mac OS X
1876 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
1877 341789 aarch64: shmat fails with valgrind on ARMv8
1878 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
1879 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
1880 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
1881 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
1882 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
1883 342221 socket connect false positive uninit memory for unknown af family
1884 342353 Allow dumping full massif output while valgrind is still running
1885 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
1889 342603 Add I2C_SMBUS ioctl support
1890 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
1891 342683 Mark memory past the initial brk limit as unaddressable
1892 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
1893 342795 Internal glibc __GI_mempcpy call should be intercepted
1894 342841 s390x: Support instructions fiebr(a) and fidbr(a)
1895 343012 Unhandled syscall 319 (memfd_create)
1896 343069 Patch updating v4l2 API support
1897 343173 helgrind crash during stack unwind
1898 343219 fix GET_STARTREGS for arm
1899 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
1900 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
1901 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
1902 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
1903 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
1904 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
1905 343597 ppc64le: incorrect use of offseof macro
1906 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
1907 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
1908 343732 Unhandled syscall 144 (setgid) on aarch64
1909 343733 Unhandled syscall 187 (msgctl and related) on aarch64
1910 343802 s390x: False positive "conditional jump or move depends on [..]
1911 343902 --vgdb=yes doesn't break when --xml=yes is used
1912 343967 Don't warn about setuid/setgid/setcap executable for directories
1913 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
1914 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
1915 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
1916 344054 www - update info for Solaris/illumos
1917 344416 'make regtest' does not work cleanly on OS X
1918 344235 Remove duplicate include of pub_core_aspacemgr.h
1919 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
1920 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
1921 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
1922 344314 callgrind_annotate ... warnings about commands containing newlines
1923 344318 socketcall should wrap recvmmsg and sendmmsg
1924 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
1925 344416 Fix 'make regtest' does not work cleanly on OS X
1926 344499 Fix compilation for Linux kernel >= 4.0.0
1927 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
1928 unix:349 (__pthread_fchdir)
1929 344559 Garbage collection of unused segment names in address space manager
1930 344560 Fix stack traces missing penultimate frame on OS X
1931 344621 Fix memcheck/tests/err_disable4 test on OS X
1932 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
1933 344702 Fix missing libobjc suppressions on OS X 10.10
1935 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
1936 344939 Fix memcheck/tests/xml1 on OS X 10.10
1937 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
1938 345079 Fix build problems in VEX/useful/test_main.c
1939 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
1940 345177 arm64: prfm (reg) not implemented
1941 345215 Performance improvements for the register allocator
1942 345248 add support for Solaris OS in valgrind
1943 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
1944 345394 Fix memcheck/tests/strchr on OS X
1945 345637 Fix memcheck/tests/sendmsg on OS X
1946 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
1947 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
1948 345887 Fix an assertion in the address space manager
1949 345928 amd64: callstack only contains current function for small stacks
1950 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1951 345987 MIPS64: Implement cavium LHX instruction
1952 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1953 346185 Fix typo saving altivec register v24
1954 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1955 and LibVEX_GuestPPC64_get_CR()
1956 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1957 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1958 346307 fuse filesystem syscall deadlocks
1959 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1960 346411 MIPS: SysRes::_valEx handling is incorrect
1961 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1962 346474 PPC64 Power 8, spr TEXASRU register not supported
1963 346487 Compiler generates "note" about a future ABI change for PPC64
1964 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1965 and causing spurious "invalid read of size 8" warnings
1966 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1967 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1968 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1969 347322 Power PC regression test cleanup
1970 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1972 347389 unhandled syscall: 373 (Linux ARM syncfs)
1973 347686 Patch set to cleanup PPC64 regtests
1974 347978 Remove bash dependencies where not needed
1975 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1976 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1978 348102 Patch updating v4l2 API support
1979 348247 amd64 front end: jno jumps wrongly when overflow is not set
1980 348269 Improve mmap MAP_HUGETLB support.
1981 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1982 348345 Assertion fails for negative lineno
1983 348377 Unsupported ARM instruction: yield
1984 348565 Fix detection of command line option availability for clang
1985 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1986 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1987 348748 Fix redundant condition
1988 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1989 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1990 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1991 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1992 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1993 349626 Implemented additional Xen hypercalls
1994 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1995 349790 Clean up of the hardware capability checking utilities.
1996 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1997 349874 Fix typos in source code
1998 349879 memcheck: add handwritten assembly for helperc_LOADV*
1999 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
2000 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
2001 350202 Add limited param to 'monitor block_list'
2002 350290 s390x: Support instructions fixbr(a)
2003 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
2004 350809 Fix none/tests/async-sigs for Solaris
2005 350811 Remove reference to --db-attach which has been removed.
2006 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
2007 350854 hard-to-understand code in VG_(load_ELF)()
2008 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2009 351386 Solaris: Cannot run ld.so.1 under Valgrind
2010 351474 Fix VG_(iseqsigset) as obvious
2011 351531 Typo in /include/vki/vki-xen-physdev.h header guard
2012 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
2013 351858 ldsoexec support on Solaris
2014 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
2015 352130 helgrind reports false races for printfs using mempcpy on FILE* state
2016 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
2017 352320 arm64 crash on none/tests/nestedfs
2018 352765 Vbit test fails on Power 6
2019 352768 The mbar instruction is missing from the Power PC support
2020 352769 Power PC program priority register (PPR) is not supported
2021 n-i-bz Provide implementations of certain compiler builtins to support
2022 compilers that may not provide those
2023 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
2024 n-i-bz Fix compilation on distros with glibc < 2.5
2025 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
2026 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
2027 n-i-bz mremap did not work properly on shared memory
2028 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
2029 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
2031 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
2032 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
2033 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
2037 Release 3.10.1 (25 November 2014)
2038 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2039 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
2040 and backports fixes for all reported missing AArch64 ARMv8 instructions
2041 and syscalls from the trunk. If you package or deliver 3.10.0 for others
2042 to use, you might want to consider upgrading to 3.10.1 instead.
2044 The following bugs have been fixed or resolved. Note that "n-i-bz"
2045 stands for "not in bugzilla" -- that is, a bug that was reported to us
2046 but never got a bugzilla entry. We encourage you to file bugs in
2047 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2048 than mailing the developers (or mailing lists) directly -- bugs that
2049 are not entered into bugzilla tend to get forgotten about or ignored.
2051 To see details of a given bug, visit
2052 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2053 where XXXXXX is the bug number as listed below.
2055 335440 arm64: ld1 (single structure) is not implemented
2056 335713 arm64: unhanded instruction: prfm (immediate)
2057 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
2058 339182 ppc64: AvSplat ought to load destination vector register with [..]
2059 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
2060 339433 ppc64 lxvw4x instruction uses four 32-byte loads
2061 339645 Use correct tag names in sys_getdents/64 wrappers
2062 339706 Fix false positive for ioctl(TIOCSIG) on linux
2063 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
2064 339853 arm64 times syscall unknown
2065 339855 arm64 unhandled getsid/setsid syscalls
2066 339858 arm64 dmb sy not implemented
2067 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
2068 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
2069 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
2071 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
2072 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
2073 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
2074 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
2075 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
2076 340509 arm64: unhandled instruction fcvtas
2077 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
2078 340632 arm64: unhandled instruction fcvtas
2079 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
2080 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
2081 340788 warning: unhandled syscall: 318 (getrandom)
2082 340807 disInstr(arm): unhandled instruction: 0xEE989B20
2083 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
2084 340922 arm64: unhandled getgroups/setgroups syscalls
2085 350251 Fix typo in VEX utility program (test_main.c).
2086 350407 arm64: unhandled instruction ucvtf (vector, integer)
2087 350809 none/tests/async-sigs breaks when run under cron on Solaris
2088 350811 update README.solaris after r15445
2089 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
2090 350854 strange code in VG_(load_ELF)()
2091 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
2092 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
2093 n-i-bz Add missing ]] to terminate CDATA.
2094 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
2095 n-i-bz Enable sys_fadvise64_64 on arm32.
2096 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
2097 n-i-bz Add test cases for all known arm64 load/store instructions.
2098 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
2099 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
2100 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
2101 n-i-bz arm64: Implement "fcvtpu w, s".
2102 n-i-bz arm64: implement ADDP and various others
2103 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
2104 n-i-bz arm64: enable FCVT{A,N}S X,S.
2106 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
2110 Release 3.10.0 (10 September 2014)
2111 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2113 3.10.0 is a feature release with many improvements and the usual
2114 collection of bug fixes.
2116 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
2117 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
2118 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
2119 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
2120 significantly improved relative to the 3.9.0 release.
2122 * ================== PLATFORM CHANGES =================
2124 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
2125 is mostly complete, and is usable, but some SIMD instructions are as
2128 * Support for little-endian variant of the 64-bit POWER architecture.
2130 * Support for Android on MIPS32.
2132 * Support for 64bit FPU on MIPS32 platforms.
2134 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
2136 * Configuration for and running on Android targets has changed.
2137 See README.android in the source tree for details.
2139 * ================== DEPRECATED FEATURES =================
2141 * --db-attach is now deprecated and will be removed in the next
2142 valgrind feature release. The built-in GDB server capabilities are
2143 superior and should be used instead. Learn more here:
2144 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
2146 * ==================== TOOL CHANGES ====================
2150 - Client code can now selectively disable and re-enable reporting of
2151 invalid address errors in specific ranges using the new client
2152 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
2153 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
2155 - Leak checker: there is a new leak check heuristic called
2156 "length64". This is used to detect interior pointers pointing 8
2157 bytes inside a block, on the assumption that the first 8 bytes
2158 holds the value "block size - 8". This is used by
2159 sqlite3MemMalloc, for example.
2161 - Checking of system call parameters: if a syscall parameter
2162 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
2163 several fields not initialised, an error is now reported for each
2164 field. Previously, an error was reported only for the first
2165 uninitialised field.
2167 - Mismatched alloc/free checking: a new flag
2168 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
2169 such checks if necessary.
2173 - Improvements to error messages:
2175 o Race condition error message involving heap allocated blocks also
2176 show the thread number that allocated the raced-on block.
2178 o All locks referenced by an error message are now announced.
2179 Previously, some error messages only showed the lock addresses.
2181 o The message indicating where a lock was first observed now also
2182 describes the address/location of the lock.
2184 - Helgrind now understands the Ada task termination rules and
2185 creates a happens-before relationship between a terminated task
2186 and its master. This avoids some false positives and avoids a big
2187 memory leak when a lot of Ada tasks are created and terminated.
2188 The interceptions are only activated with forthcoming releases of
2189 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
2191 - A new GDB server monitor command "info locks" giving the list of
2192 locks, their location, and their status.
2196 - callgrind_control now supports the --vgdb-prefix argument,
2197 which is needed if valgrind was started with this same argument.
2199 * ==================== OTHER CHANGES ====================
2201 * Unwinding through inlined function calls. Stack unwinding can now
2202 make use of Dwarf3 inlined-unwind information if it is available.
2203 The practical effect is that inlined calls become visible in stack
2204 traces. The suppression matching machinery has been adjusted
2205 accordingly. This is controlled by the new option
2206 --read-inline-info=yes|no. Currently this is enabled by default
2207 only on Linux and Android targets and only for the tools Memcheck,
2210 * Valgrind can now read EXIDX unwind information on 32-bit ARM
2211 targets. If an object contains both CFI and EXIDX unwind
2212 information, Valgrind will prefer the CFI over the EXIDX. This
2213 facilitates unwinding through system libraries on arm-android
2216 * Address description logic has been improved and is now common
2217 between Memcheck and Helgrind, resulting in better address
2218 descriptions for some kinds of error messages.
2220 * Error messages about dubious arguments (eg, to malloc or calloc) are
2221 output like other errors. This means that they can be suppressed
2222 and they have a stack trace.
2224 * The C++ demangler has been updated for better C++11 support.
2226 * New and modified GDB server monitor features:
2228 - Thread local variables/storage (__thread) can now be displayed.
2230 - The GDB server monitor command "v.info location <address>"
2231 displays information about an address. The information produced
2232 depends on the tool and on the options given to valgrind.
2233 Possibly, the following are described: global variables, local
2234 (stack) variables, allocated or freed blocks, ...
2236 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
2237 ask the GDB server to stop at the start of program execution, at
2238 the end of the program execution and on Valgrind internal errors.
2240 - A new monitor command "v.info stats" shows various Valgrind core
2241 and tool statistics.
2243 - A new monitor command "v.set hostvisibility" allows the GDB server
2244 to provide access to Valgrind internal host status/memory.
2246 * A new option "--aspace-minaddr=<address>" can in some situations
2247 allow the use of more memory by decreasing the address above which
2248 Valgrind maps memory. It can also be used to solve address
2249 conflicts with system libraries by increasing the default value.
2250 See user manual for details.
2252 * The amount of memory used by Valgrind to store debug info (unwind
2253 info, line number information and symbol data) has been
2254 significantly reduced, even though Valgrind now reads more
2255 information in order to support unwinding of inlined function calls.
2257 * Dwarf3 handling with --read-var-info=yes has been improved:
2259 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
2261 - Code compiled with
2262 -ffunction-sections -fdata-sections -Wl,--gc-sections
2263 no longer causes assertion failures.
2265 * Improved checking for the --sim-hints= and --kernel-variant=
2266 options. Unknown strings are now detected and reported to the user
2269 * The semantics of stack start/end boundaries in the valgrind.h
2270 VALGRIND_STACK_REGISTER client request has been clarified and
2271 documented. The convention is that start and end are respectively
2272 the lowest and highest addressable bytes of the stack.
2274 * ==================== FIXED BUGS ====================
2276 The following bugs have been fixed or resolved. Note that "n-i-bz"
2277 stands for "not in bugzilla" -- that is, a bug that was reported to us
2278 but never got a bugzilla entry. We encourage you to file bugs in
2279 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2280 than mailing the developers (or mailing lists) directly -- bugs that
2281 are not entered into bugzilla tend to get forgotten about or ignored.
2283 To see details of a given bug, visit
2284 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2285 where XXXXXX is the bug number as listed below.
2287 175819 Support for ipv6 socket reporting with --track-fds
2288 232510 make distcheck fails
2289 249435 Analyzing wine programs with callgrind triggers a crash
2290 278972 support for inlined function calls in stacktraces and suppression
2292 291310 FXSAVE instruction marks memory as undefined on amd64
2293 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2294 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2295 315199 vgcore file for threaded app does not show which thread crashed
2296 315952 tun/tap ioctls are not supported
2297 323178 Unhandled instruction: PLDW register (ARM)
2298 323179 Unhandled instruction: PLDW immediate (ARM)
2299 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2300 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2301 325124 [MIPSEL] Compilation error
2302 325477 Phase 4 support for IBM Power ISA 2.07
2303 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2304 325628 Phase 5 support for IBM Power ISA 2.07
2305 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2306 325751 Missing the two privileged Power PC Transactional Memory Instructions
2307 325816 Phase 6 support for IBM Power ISA 2.07
2308 325856 Make SGCheck fail gracefully on unsupported platforms
2309 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2310 326436 DRD: False positive in libstdc++ std::list::push_back
2311 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2312 326462 Refactor vgdb to isolate invoker stuff into separate module
2313 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2314 326623 DRD: false positive conflict report in a field assignment
2315 326724 Valgrind does not compile on OSX 1.9 Mavericks
2316 326816 Intercept for __strncpy_sse2_unaligned missing?
2317 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2318 326983 Clear direction flag after tests on amd64.
2319 327212 Do not prepend the current directory to absolute path names.
2320 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2321 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2322 327284 s390x: Fix translation of the risbg instruction
2323 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2324 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2325 327916 DW_TAG_typedef may have no name
2326 327943 s390x: add a redirection for the 'index' function
2327 328100 XABORT not implemented
2328 328205 Implement additional Xen hypercalls
2329 328454 add support Backtraces with ARM unwind tables (EXIDX)
2330 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2331 328711 valgrind.1 manpage "memcheck options" section is badly generated
2332 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2333 329612 Incorrect handling of AT_BASE for image execution
2334 329694 clang warns about using uninitialized variable
2335 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2336 330228 mmap must align to VKI_SHMLBA on mips32
2337 330257 LLVM does not support `-mno-dynamic-no-pic` option
2338 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2339 330459 --track-fds=yes doesn't track eventfds
2340 330469 Add clock_adjtime syscall support
2341 330594 Missing sysalls on PowerPC / uClibc
2342 330622 Add test to regression suite for POWER instruction: dcbzl
2343 330939 Support for AMD's syscall instruction on x86
2345 330941 Typo in PRE(poll) syscall wrapper
2346 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2347 331254 Fix expected output for memcheck/tests/dw4
2348 331255 Fix race condition in test none/tests/coolo_sigaction
2349 331257 Fix type of jump buffer in test none/tests/faultstatus
2350 331305 configure uses bash specific syntax
2351 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2352 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2353 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2354 331829 Unexpected ioctl opcode sign extension
2355 331830 ppc64: WARNING: unhandled syscall: 96/97
2356 331839 drd/tests/sem_open specifies invalid semaphore name
2357 331847 outcome of drd/tests/thread_name is nondeterministic
2358 332037 Valgrind cannot handle Thumb "add pc, reg"
2359 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2360 consistency checks enabled
2361 332263 intercepts for pthread_rwlock_timedrdlock and
2362 pthread_rwlock_timedwrlock are incorrect
2363 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2365 332276 Implement additional Xen hypercalls
2366 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2367 332765 Fix ms_print to create temporary files in a proper directory
2368 333072 drd: Add semaphore annotations
2369 333145 Tests for missaligned PC+#imm access for arm
2370 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2371 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2372 333248 WARNING: unhandled syscall: unix:443
2373 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2374 333501 cachegrind: assertion: Cache set count is not a power of two.
2377 333666 Recognize MPX instructions and bnd prefix.
2378 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2379 333817 Valgrind reports the memory areas written to by the SG_IO
2381 334049 lzcnt fails silently (x86_32)
2382 334384 Valgrind does not have support Little Endian support for
2384 334585 recvmmsg unhandled (+patch) (arm)
2385 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2386 334727 Build fails with -Werror=format-security
2387 334788 clarify doc about --log-file initial program directory
2388 334834 PPC64 Little Endian support, patch 2
2389 334836 PPC64 Little Endian support, patch 3 testcase fixes
2390 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2391 335034 Unhandled ioctl: HCIGETDEVLIST
2392 335155 vgdb, fix error print statement.
2393 335262 arm64: movi 8bit version is not supported
2394 335263 arm64: dmb instruction is not implemented
2395 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2396 335496 arm64: sbc/abc instructions are not implemented
2397 335554 arm64: unhandled instruction: abs
2398 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2399 335735 arm64: unhandled instruction: cnt
2400 335736 arm64: unhandled instruction: uaddlv
2401 335848 arm64: unhandled instruction: {s,u}cvtf
2402 335902 arm64: unhandled instruction: sli
2403 335903 arm64: unhandled instruction: umull (vector)
2404 336055 arm64: unhandled instruction: mov (element)
2405 336062 arm64: unhandled instruction: shrn{,2}
2406 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2407 336189 arm64: unhandled Instruction: mvn
2408 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2409 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2410 336772 Make moans about unknown ioctls more informative
2411 336957 Add a section about the Solaris/illumos port on the webpage
2412 337094 ifunc wrapper is broken on ppc64
2413 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2414 337528 leak check heuristic for block prefixed by length as 64bit number
2415 337740 Implement additional Xen hypercalls
2416 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2417 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2418 337871 deprecate --db-attach
2419 338023 Add support for all V4L2/media ioctls
2420 338024 inlined functions are not shown if DW_AT_ranges is used
2421 338106 Add support for 'kcmp' syscall
2422 338115 DRD: computed conflict set differs from actual after fork
2423 338160 implement display of thread local storage in gdbsrv
2424 338205 configure.ac and check for -Wno-tautological-compare
2425 338300 coredumps are missing one byte of every segment
2426 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2427 338499 --sim-hints parsing broken due to wrong order in tokens
2428 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2429 338681 Unable to unwind through clone thread created on i386-linux
2430 338698 race condition between gdbsrv and vgdb on startup
2431 338703 helgrind on arm-linux gets false positives in dynamic loader
2432 338791 alt dwz files can be relative of debug/main file
2433 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2434 338932 build V-trunk with gcc-trunk
2435 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2436 345079 Fix build problems in VEX/useful/test_main.c
2437 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2438 n-i-bz s390x: Fix memory corruption for multithreaded applications
2439 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2440 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2441 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2442 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2443 n-i-bz Add command-line option --default-suppressions
2444 n-i-bz Add support for BLKDISCARDZEROES ioctl
2445 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2446 n-i-bz Add support for sys_pivot_root and sys_unshare
2448 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2449 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2450 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2454 Release 3.9.0 (31 October 2013)
2455 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2456 3.9.0 is a feature release with many improvements and the usual
2457 collection of bug fixes.
2459 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2460 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2461 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2462 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2464 * ================== PLATFORM CHANGES =================
2466 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2467 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2469 * Support for MIPS DSP ASE on MIPS32 platforms.
2471 * Support for s390x Decimal Floating Point instructions on hosts that
2472 have the DFP facility installed.
2474 * Support for POWER8 (Power ISA 2.07) instructions
2476 * Support for Intel AVX2 instructions. This is available only on 64
2479 * Initial support for Intel Transactional Synchronization Extensions,
2482 * Initial support for Hardware Transactional Memory on POWER.
2484 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2485 run large GUI apps tolerably well.
2487 * ==================== TOOL CHANGES ====================
2491 - Improvements in handling of vectorised code, leading to
2492 significantly fewer false error reports. You need to use the flag
2493 --partial-loads-ok=yes to get the benefits of these changes.
2495 - Better control over the leak checker. It is now possible to
2496 specify which leak kinds (definite/indirect/possible/reachable)
2497 should be displayed, which should be regarded as errors, and which
2498 should be suppressed by a given leak suppression. This is done
2499 using the options --show-leak-kinds=kind1,kind2,..,
2500 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2501 "match-leak-kinds:" line in suppression entries, respectively.
2503 Note that generated leak suppressions contain this new line and
2504 are therefore more specific than in previous releases. To get the
2505 same behaviour as previous releases, remove the "match-leak-kinds:"
2506 line from generated suppressions before using them.
2508 - Reduced "possible leak" reports from the leak checker by the use
2509 of better heuristics. The available heuristics provide detection
2510 of valid interior pointers to std::stdstring, to new[] allocated
2511 arrays with elements having destructors and to interior pointers
2512 pointing to an inner part of a C++ object using multiple
2513 inheritance. They can be selected individually using the
2514 option --leak-check-heuristics=heur1,heur2,...
2516 - Better control of stacktrace acquisition for heap-allocated
2517 blocks. Using the --keep-stacktraces option, it is possible to
2518 control independently whether a stack trace is acquired for each
2519 allocation and deallocation. This can be used to create better
2520 "use after free" errors or to decrease Valgrind's resource
2521 consumption by recording less information.
2523 - Better reporting of leak suppression usage. The list of used
2524 suppressions (shown when the -v option is given) now shows, for
2525 each leak suppressions, how many blocks and bytes it suppressed
2526 during the last leak search.
2530 - False errors resulting from the use of statically initialised
2531 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2534 - False errors resulting from the use of pthread_cond_waits that
2535 timeout, have been removed.
2537 * ==================== OTHER CHANGES ====================
2539 * Some attempt to tune Valgrind's space requirements to the expected
2540 capabilities of the target:
2542 - The default size of the translation cache has been reduced from 8
2543 sectors to 6 on Android platforms, since each sector occupies
2544 about 40MB when using Memcheck.
2546 - The default size of the translation cache has been increased to 16
2547 sectors on all other platforms, reflecting the fact that large
2548 applications require instrumentation and storage of huge amounts
2549 of code. For similar reasons, the number of memory mapped
2550 segments that can be tracked has been increased by a factor of 6.
2552 - In all cases, the maximum number of sectors in the translation
2553 cache can be controlled by the new flag --num-transtab-sectors.
2555 * Changes in how debug info (line numbers, etc) is read:
2557 - Valgrind no longer temporarily mmaps the entire object to read
2558 from it. Instead, reading is done through a small fixed sized
2559 buffer. This avoids virtual memory usage spikes when Valgrind
2560 reads debuginfo from large shared objects.
2562 - A new experimental remote debug info server. Valgrind can read
2563 debug info from a different machine (typically, a build host)
2564 where debuginfo objects are stored. This can save a lot of time
2565 and hassle when running Valgrind on resource-constrained targets
2566 (phones, tablets) when the full debuginfo objects are stored
2567 somewhere else. This is enabled by the --debuginfo-server=
2570 - Consistency checking between main and debug objects can be
2571 disabled using the --allow-mismatched-debuginfo option.
2573 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2574 scanning can recover stack traces in some cases when the normal
2575 unwind mechanisms fail. Stack scanning is best described as "a
2576 nasty, dangerous and misleading hack" and so is disabled by default.
2577 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2580 * Detection and merging of recursive stack frame cycles. When your
2581 program has recursive algorithms, this limits the memory used by
2582 Valgrind for recorded stack traces and avoids recording
2583 uninteresting repeated calls. This is controlled by the command
2584 line option --merge-recursive-frame and by the monitor command
2585 "v.set merge-recursive-frames".
2587 * File name and line numbers for used suppressions. The list of used
2588 suppressions (shown when the -v option is given) now shows, for each
2589 used suppression, the file name and line number where the suppression
2592 * New and modified GDB server monitor features:
2594 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2595 that can be used to execute gdbserver monitor commands from the
2598 - A new monitor command, "v.info open_fds", that gives the list of
2599 open file descriptors and additional details.
2601 - An optional message in the "v.info n_errs_found" monitor command,
2602 for example "v.info n_errs_found test 1234 finished", allowing a
2603 comment string to be added to the process output, perhaps for the
2604 purpose of separating errors of different tests or test phases.
2606 - A new monitor command "v.info execontext" that shows information
2607 about the stack traces recorded by Valgrind.
2609 - A new monitor command "v.do expensive_sanity_check_general" to run
2610 some internal consistency checks.
2612 * New flag --sigill-diagnostics to control whether a diagnostic
2613 message is printed when the JIT encounters an instruction it can't
2614 translate. The actual behavior -- delivery of SIGILL to the
2615 application -- is unchanged.
2617 * The maximum amount of memory that Valgrind can use on 64 bit targets
2618 has been increased from 32GB to 64GB. This should make it possible
2619 to run applications on Memcheck that natively require up to about 35GB.
2621 * ==================== FIXED BUGS ====================
2623 The following bugs have been fixed or resolved. Note that "n-i-bz"
2624 stands for "not in bugzilla" -- that is, a bug that was reported to us
2625 but never got a bugzilla entry. We encourage you to file bugs in
2626 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2627 than mailing the developers (or mailing lists) directly -- bugs that
2628 are not entered into bugzilla tend to get forgotten about or ignored.
2630 To see details of a given bug, visit
2631 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2632 where XXXXXX is the bug number as listed below.
2634 123837 system call: 4th argument is optional, depending on cmd
2635 135425 memcheck should tell you where Freed blocks were Mallocd
2636 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2637 207815 Adds some of the drm ioctls to syswrap-linux.c
2638 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2639 252955 Impossible to compile with ccache
2640 253519 Memcheck reports auxv pointer accesses as invalid reads.
2641 263034 Crash when loading some PPC64 binaries
2642 269599 Increase deepest backtrace
2643 274695 s390x: Support "compare to/from logical" instructions (z196)
2644 275800 s390x: Autodetect cache info (part 2)
2645 280271 Valgrind reports possible memory leaks on still-reachable std::string
2646 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2647 289578 Backtraces with ARM unwind tables (stack scan flags)
2648 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2649 304832 ppc32: build failure
2650 305431 Use find_buildid shdr fallback for separate .debug files
2651 305728 Add support for AVX2 instructions
2652 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2653 306035 s390x: Fix IR generation for LAAG and friends
2654 306054 s390x: Condition code computation for convert-to-int/logical
2655 306098 s390x: alternate opcode form for convert to/from fixed
2656 306587 Fix cache line detection from auxiliary vector for PPC.
2657 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2658 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2659 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2660 307101 sys_capget second argument can be NULL
2661 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2662 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2663 307113 s390x: DFP support
2664 307141 valgrind does't work in mips-linux system
2665 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2666 307285 x86_amd64 feature test for avx in test suite is wrong
2667 307290 memcheck overlap testcase needs memcpy version filter
2668 307463 Please add "&limit=0" to the "all open bugs" link
2669 307465 --show-possibly-lost=no should reduce the error count / exit code
2670 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2671 307729 pkgconfig support broken valgrind.pc
2672 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2673 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2674 308089 Unhandled syscall on ppc64: prctl
2675 308135 PPC32 MPC8xx has 16 bytes cache size
2676 308321 testsuite memcheck filter interferes with gdb_filter
2678 308341 vgdb should report process exit (or fatal signal)
2679 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2680 308495 Remove build dependency on installed Xen headers
2681 308573 Internal error on 64-bit instruction executed in 32-bit mode
2683 308627 pmovmskb validity bit propagation is imprecise
2684 308644 vgdb command for having the info for the track-fds option
2685 308711 give more info about aspacemgr and arenas in out_of_memory
2686 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2687 308718 ARM implement SMLALBB family of instructions
2688 308886 Missing support for PTRACE_SET/GETREGSET
2689 308930 syscall name_to_handle_at (303 on amd64) not handled
2690 309229 V-bit tester does not report number of tests generated
2691 309323 print unrecognized instuction on MIPS
2692 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2693 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2694 309430 Self hosting ppc64 encounters a vassert error on operand type
2695 309600 valgrind is a bit confused about 0-sized sections
2696 309823 Generate errors for still reachable blocks
2697 309921 PCMPISTRI validity bit propagation is imprecise
2698 309922 none/tests/ppc64/test_dfp5 sometimes fails
2699 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2700 310424 --read-var-info does not properly describe static variables
2701 310792 search additional path for debug symbols
2702 310931 s390x: Message-security assist (MSA) instruction extension [..]
2703 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2704 311318 ARM: "128-bit constant is not implemented" error message
2705 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2706 311690 V crashes because it redirects branches inside of a redirected function
2707 311880 x86_64: make regtest hangs at shell_valid1
2708 311922 WARNING: unhandled syscall: 170
2710 312171 ppc: insn selection for DFP
2711 312571 Rounding mode call wrong for the DFP Iops [..]
2712 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2713 312913 Dangling pointers error should also report the alloc stack trace
2714 312980 Building on Mountain Lion generates some compiler warnings
2715 313267 Adding MIPS64/Linux port to Valgrind
2718 313811 Buffer overflow in assert_fail
2719 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2720 314269 ppc: dead code in insn selection
2721 314718 ARM: implement integer divide instruction (sdiv and udiv)
2722 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2723 315441 sendmsg syscall should ignore unset msghdr msg_flags
2724 315534 msgrcv inside a thread causes valgrind to hang (block)
2725 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2726 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2727 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2728 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2729 316144 valgrind.1 manpage contains unknown ??? strings [..]
2730 316145 callgrind command line options in manpage reference (unknown) [..]
2731 316145 callgrind command line options in manpage reference [..]
2732 316181 drd: Fixed a 4x slowdown for certain applications
2733 316503 Valgrind does not support SSE4 "movntdqa" instruction
2734 316535 Use of |signed int| instead of |size_t| in valgrind messages
2735 316696 fluidanimate program of parsec 2.1 stuck
2736 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2737 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2738 317186 "Impossible happens" when occurs VCVT instruction on ARM
2739 317318 Support for Threading Building Blocks "scalable_malloc"
2740 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2741 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2742 317463 bmi testcase IR SANITY CHECK FAILURE
2743 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2744 318050 libmpiwrap fails to compile with out-of-source build
2745 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2746 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2747 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2748 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2749 318932 Add missing PPC64 and PPC32 system call support
2750 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2751 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2752 319494 VEX Makefile-gcc standalone build update after r2702
2753 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2754 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2755 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2756 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2757 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2758 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2759 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2760 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2761 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2762 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2763 320895 add fanotify support (patch included)
2764 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2765 321065 Valgrind updates for Xen 4.3
2766 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2767 321363 Unhandled instruction: SSAX (ARM + Thumb)
2768 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2769 321466 Unhandled instruction: SHASX (ARM + Thumb)
2770 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2771 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2772 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2773 321620 Unhandled instruction: UASX (ARM + Thumb)
2774 321621 Unhandled instruction: USAX (ARM + Thumb)
2775 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2776 321693 Unhandled instruction: LDRSBT (Thumb)
2777 321694 Unhandled instruction: UQASX (ARM + Thumb)
2778 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2779 321697 Unhandled instruction: UHASX (ARM + Thumb)
2780 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2781 321704 Unhandled instruction: REVSH (ARM + Thumb)
2782 321730 Add cg_diff and cg_merge man pages
2783 321738 Add vgdb and valgrind-listener man pages
2785 321891 Unhandled instruction: LDRHT (Thumb)
2786 321960 pthread_create() then alloca() causing invalid stack write errors
2787 321969 ppc32 and ppc64 don't support [lf]setxattr
2788 322254 Show threadname together with tid if set by application
2789 322294 Add initial support for IBM Power ISA 2.07
2790 322368 Assertion failure in wqthread_hijack under OS X 10.8
2791 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2792 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2793 322851 0bXXX binary literal syntax is not standard
2794 323035 Unhandled instruction: LDRSHT(Thumb)
2795 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2796 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2797 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2798 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2799 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2800 323437 Phase 2 support for IBM Power ISA 2.07
2801 323713 Support mmxext (integer sse) subset on i386 (athlon)
2802 323803 Transactional memory instructions are not supported for Power
2803 323893 SSE3 not available on amd cpus in valgrind
2804 323905 Probable false positive from Valgrind/drd on close()
2805 323912 valgrind.h header isn't compatible for mingw64
2806 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
2807 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
2808 324181 mmap does not handle MAP_32BIT
2809 324227 memcheck false positive leak when a thread calls exit+block [..]
2810 324421 Support for fanotify API on ARM architecture
2811 324514 gdbserver monitor cmd output behaviour consistency [..]
2812 324518 ppc64: Emulation of dcbt instructions does not handle [..]
2813 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
2814 324582 When access is made to freed memory, report both allocation [..]
2815 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
2816 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
2817 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
2818 324834 Unhandled instructions in Microsoft C run-time for x86_64
2819 324894 Phase 3 support for IBM Power ISA 2.07
2820 326091 drd: Avoid false race reports from optimized strlen() impls
2821 326113 valgrind libvex hwcaps error on AMD64
2822 n-i-bz Some wrong command line options could be ignored
2823 n-i-bz patch to allow fair-sched on android
2824 n-i-bz report error for vgdb snapshot requested before execution
2825 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
2827 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
2831 Release 3.8.1 (19 September 2012)
2832 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2833 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
2834 that occur moderately frequently in real use cases, adds support for
2835 some missing instructions on ARM, and fixes a deadlock condition on
2836 MacOSX. If you package or deliver 3.8.0 for others to use, you might
2837 want to consider upgrading to 3.8.1 instead.
2839 The following bugs have been fixed or resolved. Note that "n-i-bz"
2840 stands for "not in bugzilla" -- that is, a bug that was reported to us
2841 but never got a bugzilla entry. We encourage you to file bugs in
2842 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2843 than mailing the developers (or mailing lists) directly -- bugs that
2844 are not entered into bugzilla tend to get forgotten about or ignored.
2846 To see details of a given bug, visit
2847 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2848 where XXXXXX is the bug number as listed below.
2851 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
2852 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
2853 298281 wcslen causes false(?) uninitialised value warnings
2854 301281 valgrind hangs on OS X when the process calls system()
2855 304035 disInstr(arm): unhandled instruction 0xE1023053
2856 304867 implement MOVBE instruction in x86 mode
2857 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
2858 305042 amd64: implement 0F 7F encoding of movq between two registers
2859 305199 ARM: implement QDADD and QDSUB
2860 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
2861 305513 killed by fatal signal: SIGSEGV
2862 305690 DRD reporting invalid semaphore when sem_trywait fails
2863 305926 Invalid alignment checks for some AVX instructions
2864 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
2865 306310 3.8.0 release tarball missing some files
2866 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
2867 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
2868 n-i-bz shmat of a segment > 4Gb does not work
2869 n-i-bz simulate_control_c script wrong USR1 signal number on mips
2870 n-i-bz vgdb ptrace calls wrong on mips [...]
2871 n-i-bz Fixes for more MPI false positives
2872 n-i-bz exp-sgcheck's memcpy causes programs to segfault
2873 n-i-bz OSX build w/ clang: asserts at startup
2874 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
2875 n-i-bz fix a couple of union tag-vs-field mixups
2876 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
2878 The following bugs were fixed in 3.8.0 but not listed in this NEWS
2881 254088 Valgrind should know about UD2 instruction
2884 304754 NEWS blows TeX's little mind
2886 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
2890 Release 3.8.0 (10 August 2012)
2891 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2892 3.8.0 is a feature release with many improvements and the usual
2893 collection of bug fixes.
2895 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2896 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
2897 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
2898 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
2899 There is initial support for MacOSX 10.8, but it is not usable for
2900 serious work at present.
2902 * ================== PLATFORM CHANGES =================
2904 * Support for MIPS32 platforms running Linux. Valgrind has been
2905 tested on MIPS32 and MIPS32r2 platforms running different Debian
2906 Squeeze and MeeGo distributions. Both little-endian and big-endian
2907 cores are supported. The tools Memcheck, Massif and Lackey have
2908 been tested and are known to work. See README.mips for more details.
2910 * Preliminary support for Android running on x86.
2912 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
2914 * Support for Intel AVX instructions and for AES instructions. This
2915 support is available only for 64 bit code.
2917 * Support for POWER Decimal Floating Point instructions.
2919 * ==================== TOOL CHANGES ====================
2921 * Non-libc malloc implementations are now supported. This is useful
2922 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
2923 Using the new option --soname-synonyms, such tools can be informed
2924 that the malloc implementation is either linked statically into the
2925 executable, or is present in some other shared library different
2926 from libc.so. This makes it possible to process statically linked
2927 programs, and programs using other malloc libraries, for example
2928 TCMalloc or JEMalloc.
2930 * For tools that provide their own replacement for malloc et al, the
2931 option --redzone-size=<number> allows users to specify the size of
2932 the padding blocks (redzones) added before and after each client
2933 allocated block. Smaller redzones decrease the memory needed by
2934 Valgrind. Bigger redzones increase the chance to detect blocks
2935 overrun or underrun. Prior to this change, the redzone size was
2936 hardwired to 16 bytes in Memcheck.
2940 - The leak_check GDB server monitor command now can
2941 control the maximum nr of loss records to output.
2943 - Reduction of memory use for applications allocating
2944 many blocks and/or having many partially defined bytes.
2946 - Addition of GDB server monitor command 'block_list' that lists
2947 the addresses/sizes of the blocks of a leak search loss record.
2949 - Addition of GDB server monitor command 'who_points_at' that lists
2950 the locations pointing at a block.
2952 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2953 detect an invalid access of these redzones, by marking them
2954 noaccess. Similarly, if a redzone size is given for a memory
2955 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2956 This still allows to find some bugs if the user has forgotten to
2957 mark the pool superblock noaccess.
2959 - Performance of memory leak check has been improved, especially in
2960 cases where there are many leaked blocks and/or many suppression
2961 rules used to suppress leak reports.
2963 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2964 more precise analysis, which is important for LLVM/Clang
2965 generated code. This is at the cost of somewhat reduced
2966 performance. Note there is no change to analysis precision or
2967 costs on Linux targets.
2971 - Added even more facilities that can help finding the cause of a data
2972 race, namely the command-line option --ptrace-addr and the macro
2973 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2975 - Fixed a subtle bug that could cause false positive data race reports.
2977 * ==================== OTHER CHANGES ====================
2979 * The C++ demangler has been updated so as to work well with C++
2980 compiled by up to at least g++ 4.6.
2982 * Tool developers can make replacement/wrapping more flexible thanks
2983 to the new option --soname-synonyms. This was reported above, but
2984 in fact is very general and applies to all function
2985 replacement/wrapping, not just to malloc-family functions.
2987 * Round-robin scheduling of threads can be selected, using the new
2988 option --fair-sched= yes. Prior to this change, the pipe-based
2989 thread serialisation mechanism (which is still the default) could
2990 give very unfair scheduling. --fair-sched=yes improves
2991 responsiveness of interactive multithreaded applications, and
2992 improves repeatability of results from the thread checkers Helgrind
2995 * For tool developers: support to run Valgrind on Valgrind has been
2996 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2998 * gdbserver now shows the float shadow registers as integer
2999 rather than float values, as the shadow values are mostly
3000 used as bit patterns.
3002 * Increased limit for the --num-callers command line flag to 500.
3004 * Performance improvements for error matching when there are many
3005 suppression records in use.
3007 * Improved support for DWARF4 debugging information (bug 284184).
3009 * Initial support for DWZ compressed Dwarf debug info.
3011 * Improved control over the IR optimiser's handling of the tradeoff
3012 between performance and precision of exceptions. Specifically,
3013 --vex-iropt-precise-memory-exns has been removed and replaced by
3014 --vex-iropt-register-updates, with extended functionality. This
3015 allows the Valgrind gdbserver to always show up to date register
3018 * Modest performance gains through the use of translation chaining for
3021 * ==================== FIXED BUGS ====================
3023 The following bugs have been fixed or resolved. Note that "n-i-bz"
3024 stands for "not in bugzilla" -- that is, a bug that was reported to us
3025 but never got a bugzilla entry. We encourage you to file bugs in
3026 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
3027 than mailing the developers (or mailing lists) directly -- bugs that
3028 are not entered into bugzilla tend to get forgotten about or ignored.
3030 To see details of a given bug, visit
3031 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3032 where XXXXXX is the bug number as listed below.
3034 197914 Building valgrind from svn now requires automake-1.10
3035 203877 increase to 16Mb maximum allowed alignment for memalign et al
3036 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
3037 247386 make perf does not run all performance tests
3038 270006 Valgrind scheduler unfair
3039 270777 Adding MIPS/Linux port to Valgrind
3040 270796 s390x: Removed broken support for the TS insn
3041 271438 Fix configure for proper SSE4.2 detection
3042 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
3043 273475 Add support for AVX instructions
3044 274078 improved configure logic for mpicc
3045 276993 fix mremap 'no thrash checks'
3046 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
3047 281482 memcheck incorrect byte allocation count in realloc() for silly argument
3048 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
3049 283413 Fix wrong sanity check
3050 283671 Robustize alignment computation in LibVEX_Alloc
3051 283961 Adding support for some HCI IOCTLs
3052 284124 parse_type_DIE: confused by: DWARF 4
3053 284864 == 273475 (Add support for AVX instructions)
3054 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
3055 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
3056 285725 == 273475 (Add support for AVX instructions)
3057 286261 add wrapper for linux I2C_RDWR ioctl
3058 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
3059 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
3060 286384 configure fails "checking for a supported version of gcc"
3061 286497 == 273475 (Add support for AVX instructions)
3062 286596 == 273475 (Add support for AVX instructions)
3063 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
3064 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
3065 287260 Incorrect conditional jump or move depends on uninitialised value(s)
3066 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
3067 287307 == 273475 (Add support for AVX instructions)
3068 287858 VG_(strerror): unknown error
3069 288298 (MacOSX) unhandled syscall shm_unlink
3070 288995 == 273475 (Add support for AVX instructions)
3071 289470 Loading of large Mach-O thin binaries fails.
3072 289656 == 273475 (Add support for AVX instructions)
3073 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
3074 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
3075 289839 s390x: Provide support for unicode conversion instructions
3076 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
3077 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
3078 290655 Add support for AESKEYGENASSIST instruction
3079 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
3080 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
3081 291253 ES register not initialised in valgrind simulation
3082 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
3083 291865 s390x: Support the "Compare Double and Swap" family of instructions
3084 292300 == 273475 (Add support for AVX instructions)
3085 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
3086 292493 == 273475 (Add support for AVX instructions)
3087 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
3088 292627 Missing support for some SCSI ioctls
3089 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
3090 292841 == 273475 (Add support for AVX instructions)
3091 292993 implement the getcpu syscall on amd64-linux
3092 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
3093 293088 Add some VEX sanity checks for ppc64 unhandled instructions
3094 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
3095 293754 PCMPxSTRx not implemented for 16-bit characters
3096 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
3097 293808 CLFLUSH not supported by latest VEX for amd64
3098 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
3099 294048 MPSADBW instruction not implemented
3100 294055 regtest none/tests/shell fails when locale is not set to C
3101 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
3102 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
3103 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
3104 294260 disInstr_AMD64: disInstr miscalculated next %rip
3105 294523 --partial-loads-ok=yes causes false negatives
3106 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
3107 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
3108 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
3109 295089 can not annotate source for both helgrind and drd
3110 295221 POWER Processor decimal floating point instruction support missing
3111 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
3112 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
3113 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
3114 295617 ARM - Add some missing syscalls
3115 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
3116 296229 Linux user input device ioctls missing wrappers
3117 296318 ELF Debug info improvements (more than one rx/rw mapping)
3118 296422 Add translation chaining support
3119 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
3120 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
3121 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
3122 297078 gdbserver signal handling problems [..]
3123 297147 drd false positives on newly allocated memory
3124 297329 disallow decoding of IBM Power DFP insns on some machines
3125 297497 POWER Processor decimal floating point instruction support missing
3126 297701 Another alias for strncasecmp_l in libc-2.13.so
3127 297911 'invalid write' not reported when using APIs for custom mem allocators.
3128 297976 s390x: revisit EX implementation
3129 297991 Valgrind interferes with mmap()+ftell()
3130 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
3131 297993 Fix compilation of valgrind with gcc -g3.
3132 298080 POWER Processor DFP support missing, part 3
3133 298227 == 273475 (Add support for AVX instructions)
3134 298335 == 273475 (Add support for AVX instructions)
3135 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
3136 298394 s390x: Don't bail out on an unknown machine model. [..]
3137 298421 accept4() syscall (366) support is missing for ARM
3138 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
3139 298732 valgrind installation problem in ubuntu with kernel version 3.x
3140 298862 POWER Processor DFP instruction support missing, part 4
3141 298864 DWARF reader mis-parses DW_FORM_ref_addr
3142 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
3143 299053 Support DWARF4 DW_AT_high_pc constant form
3144 299104 == 273475 (Add support for AVX instructions)
3145 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
3146 299629 dup3() syscall (358) support is missing for ARM
3147 299694 POWER Processor DFP instruction support missing, part 5
3148 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
3149 299803 == 273475 (Add support for AVX instructions)
3150 299804 == 273475 (Add support for AVX instructions)
3151 299805 == 273475 (Add support for AVX instructions)
3152 300140 ARM - Missing (T1) SMMUL
3153 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
3154 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
3155 300414 FCOM and FCOMP unimplemented for amd64 guest
3156 301204 infinite loop in canonicaliseSymtab with ifunc symbol
3157 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
3158 301265 add x86 support to Android build
3159 301984 configure script doesn't detect certain versions of clang
3160 302205 Fix compiler warnings for POWER VEX code and POWER test cases
3161 302287 Unhandled movbe instruction on Atom processors
3162 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
3163 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
3164 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
3165 302656 == 273475 (Add support for AVX instructions)
3166 302709 valgrind for ARM needs extra tls support for android emulator [..]
3167 302827 add wrapper for CDROM_GET_CAPABILITY
3168 302901 Valgrind crashes with dwz optimized debuginfo
3169 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
3170 303116 Add support for the POWER instruction popcntb
3171 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
3172 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
3173 303466 == 273475 (Add support for AVX instructions)
3174 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
3175 303963 strstr() function produces wrong results under valgrind callgrind
3176 304054 CALL_FN_xx macros need to enforce stack alignment
3177 304561 tee system call not supported
3178 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
3179 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
3180 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
3181 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
3182 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
3183 n-i-bz s390x: Shadow registers can now be examined using vgdb
3185 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
3186 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
3190 Release 3.7.0 (5 November 2011)
3191 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3192 3.7.0 is a feature release with many significant improvements and the
3193 usual collection of bug fixes.
3195 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3196 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
3197 Support for recent distros and toolchain components (glibc 2.14, gcc
3198 4.6, MacOSX 10.7) has been added.
3200 * ================== PLATFORM CHANGES =================
3202 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
3203 analyse 64-bit programs running on z/Architecture. Most user space
3204 instructions up to and including z10 are supported. Valgrind has
3205 been tested extensively on z9, z10, and z196 machines running SLES
3206 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
3207 known to work well. Callgrind, Helgrind, and DRD work reasonably
3208 well on z9 and later models. See README.s390 for more details.
3210 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
3211 64-bit processes are supported. Some complex threaded applications
3212 (Firefox) are observed to hang when run as 32 bit applications,
3213 whereas 64-bit versions run OK. The cause is unknown. Memcheck
3214 will likely report some false errors. In general, expect some rough
3215 spots. This release also supports MacOSX 10.6, but drops support
3218 * Preliminary support for Android (on ARM). Valgrind can now run
3219 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
3220 README.android for more details, plus instructions on how to get
3223 * Support for the IBM Power ISA 2.06 (Power7 instructions)
3225 * General correctness and performance improvements for ARM/Linux, and,
3226 by extension, ARM/Android.
3228 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
3229 instruction set support is under development but is not available in
3232 * Support for AIX5 has been removed.
3234 * ==================== TOOL CHANGES ====================
3236 * Memcheck: some incremental changes:
3238 - reduction of memory use in some circumstances
3240 - improved handling of freed memory, which in some circumstances
3241 can cause detection of use-after-free that would previously have
3244 - fix of a longstanding bug that could cause false negatives (missed
3245 errors) in programs doing vector saturated narrowing instructions.
3247 * Helgrind: performance improvements and major memory use reductions,
3248 particularly for large, long running applications which perform many
3249 synchronisation (lock, unlock, etc) events. Plus many smaller
3252 - display of locksets for both threads involved in a race
3254 - general improvements in formatting/clarity of error messages
3256 - addition of facilities and documentation regarding annotation
3257 of thread safe reference counted C++ classes
3259 - new flag --check-stack-refs=no|yes [yes], to disable race checking
3260 on thread stacks (a performance hack)
3262 - new flag --free-is-write=no|yes [no], to enable detection of races
3263 where one thread accesses heap memory but another one frees it,
3264 without any coordinating synchronisation event
3266 * DRD: enabled XML output; added support for delayed thread deletion
3267 in order to detect races that occur close to the end of a thread
3268 (--join-list-vol); fixed a memory leak triggered by repeated client
3269 memory allocatation and deallocation; improved Darwin support.
3271 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
3273 * exp-sgcheck: this tool has been reduced in scope so as to improve
3274 performance and remove checking that Memcheck does better.
3275 Specifically, the ability to check for overruns for stack and global
3276 arrays is unchanged, but the ability to check for overruns of heap
3277 blocks has been removed. The tool has accordingly been renamed to
3278 exp-sgcheck ("Stack and Global Array Checking").
3280 * ==================== OTHER CHANGES ====================
3282 * GDB server: Valgrind now has an embedded GDB server. That means it
3283 is possible to control a Valgrind run from GDB, doing all the usual
3284 things that GDB can do (single stepping, breakpoints, examining
3285 data, etc). Tool-specific functionality is also available. For
3286 example, it is possible to query the definedness state of variables
3287 or memory from within GDB when running Memcheck; arbitrarily large
3288 memory watchpoints are supported, etc. To use the GDB server, start
3289 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3292 * Improved support for unfriendly self-modifying code: a new option
3293 --smc-check=all-non-file is available. This adds the relevant
3294 consistency checks only to code that originates in non-file-backed
3295 mappings. In effect this confines the consistency checking only to
3296 code that is or might be JIT generated, and avoids checks on code
3297 that must have been compiled ahead of time. This significantly
3298 improves performance on applications that generate code at run time.
3300 * It is now possible to build a working Valgrind using Clang-2.9 on
3303 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3304 These enable and disable error reporting on a per-thread, and
3305 nestable, basis. This is useful for hiding errors in particularly
3306 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3307 now uses this facility.
3309 * Added the --mod-funcname option to cg_diff.
3311 * ==================== FIXED BUGS ====================
3313 The following bugs have been fixed or resolved. Note that "n-i-bz"
3314 stands for "not in bugzilla" -- that is, a bug that was reported to us
3315 but never got a bugzilla entry. We encourage you to file bugs in
3316 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3317 mailing the developers (or mailing lists) directly -- bugs that are
3318 not entered into bugzilla tend to get forgotten about or ignored.
3320 To see details of a given bug, visit
3321 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3322 where XXXXXX is the bug number as listed below.
3324 79311 malloc silly arg warning does not give stack trace
3325 210935 port valgrind.h (not valgrind) to win32 to support client requests
3326 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3327 243404 Port to zSeries
3328 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3329 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3330 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3331 253206 Some fixes for the faultstatus testcase
3332 255223 capget testcase fails when running as root
3333 256703 xlc_dbl_u32.c testcase broken
3334 256726 Helgrind tests have broken inline asm
3335 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3336 264800 testcase compile failure on zseries
3337 265762 make public VEX headers compilable by G++ 3.x
3338 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3339 266753 configure script does not give the user the option to not use QtCore
3340 266931 gen_insn_test.pl is broken
3341 266961 ld-linux.so.2 i?86-linux strlen issues
3342 266990 setns instruction causes false positive
3343 267020 Make directory for temporary files configurable at run-time.
3344 267342 == 267997 (segmentation fault on Mac OS 10.6)
3345 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3346 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3347 267488 regtest: darwin support for 64-bit build
3348 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3349 267630 Add support for IBM Power ISA 2.06 -- stage 1
3350 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3351 267819 Add client request for informing the core about reallocation
3352 267925 laog data structure quadratic for a single sequence of lock
3353 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3354 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3355 268513 missed optimizations in fold_Expr
3356 268619 s390x: fpr - gpr transfer facility
3357 268620 s390x: reconsider "long displacement" requirement
3358 268621 s390x: improve IR generation for XC
3359 268715 s390x: FLOGR is not universally available
3360 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3361 268930 s390x: MHY is not universally available
3362 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3363 269079 Support ptrace system call on ARM
3364 269144 missing "Bad option" error message
3365 269209 conditional load and store facility (z196)
3366 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3367 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3368 269736 s390x: minor code generation tweaks
3369 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3370 269863 s390x: remove unused function parameters
3371 269864 s390x: tweak s390_emit_load_cc
3372 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3373 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3374 270115 s390x: rewrite some testcases
3375 270309 == 267997 (valgrind crash on startup)
3376 270320 add support for Linux FIOQSIZE ioctl() call
3377 270326 segfault while trying to sanitize the environment passed to execle
3378 270794 IBM POWER7 support patch causes regression in none/tests
3379 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3380 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3381 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3382 270959 s390x: invalid use of R0 as base register
3383 271042 VSX configure check fails when it should not
3384 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3385 271259 s390x: fix code confusion
3386 271337 == 267997 (Valgrind segfaults on MacOS X)
3387 271385 s390x: Implement Ist_MBE
3388 271501 s390x: misc cleanups
3389 271504 s390x: promote likely and unlikely
3390 271579 ppc: using wrong enum type
3391 271615 unhandled instruction "popcnt" (arch=amd10h)
3392 271730 Fix bug when checking ioctls: duplicate check
3393 271776 s390x: provide STFLE instruction support
3394 271779 s390x: provide clock instructions like STCK
3395 271799 Darwin: ioctls without an arg report a memory error
3396 271820 arm: fix type confusion
3397 271917 pthread_cond_timedwait failure leads to not-locked false positive
3398 272067 s390x: fix DISP20 macro
3399 272615 A typo in debug output in mc_leakcheck.c
3400 272661 callgrind_annotate chokes when run from paths containing regex chars
3401 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3402 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3403 272967 make documentation build-system more robust
3404 272986 Fix gcc-4.6 warnings with valgrind.h
3405 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3406 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3407 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3408 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3409 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3410 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3411 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3412 273778 exp-ptrcheck: unhandled sysno == 259
3413 274089 exp-ptrcheck: unhandled sysno == 208
3414 274378 s390x: Various dispatcher tweaks
3415 274447 WARNING: unhandled syscall: 340
3416 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3417 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3418 274926 valgrind does not build against linux-3
3419 275148 configure FAIL with glibc-2.14
3420 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3421 275168 Make Valgrind work for MacOSX 10.7 Lion
3422 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3423 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3424 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3425 275308 Fix implementation for ppc64 fres instruc
3426 275339 s390x: fix testcase compile warnings
3427 275517 s390x: Provide support for CKSM instruction
3428 275710 s390x: get rid of redundant address mode calculation
3429 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3430 275852 == 250101 (valgrind uses all swap space and is killed)
3431 276784 Add support for IBM Power ISA 2.06 -- stage 3
3432 276987 gdbsrv: fix tests following recent commits
3433 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3434 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3435 277471 Unhandled syscall: 340
3436 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3437 277653 ARM: support Thumb2 PLD instruction
3438 277663 ARM: NEON float VMUL by scalar incorrect
3439 277689 ARM: tests for VSTn with register post-index are broken
3440 277694 ARM: BLX LR instruction broken in ARM mode
3441 277780 ARM: VMOV.F32 (immediate) instruction is broken
3442 278057 fuse filesystem syscall deadlocks
3443 278078 Unimplemented syscall 280 on ppc32
3444 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3445 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3446 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3447 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3448 279027 Support for MVCL and CLCL instruction
3449 279027 s390x: Provide support for CLCL and MVCL instructions
3450 279062 Remove a redundant check in the insn selector for ppc.
3451 279071 JDK creates PTEST with redundant REX.W prefix
3452 279212 gdbsrv: add monitor cmd v.info scheduler.
3453 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3454 279698 memcheck discards valid-bits for packuswb
3455 279795 memcheck reports uninitialised values for mincore on amd64
3456 279994 Add support for IBM Power ISA 2.06 -- stage 3
3457 280083 mempolicy syscall check errors
3458 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3459 280710 s390x: config files for nightly builds
3460 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3461 280965 Valgrind breaks fcntl locks when program does mmap
3462 281138 WARNING: unhandled syscall: 340
3463 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3464 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3465 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3466 281468 s390x: handle do_clone and gcc clones in call traces
3467 281488 ARM: VFP register corruption
3468 281828 == 275284 (false memmove warning: "Source and destination overlap")
3469 281883 s390x: Fix system call wrapper for "clone".
3470 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3471 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3472 282238 SLES10: make check fails
3473 282979 strcasestr needs replacement with recent(>=2.12) glibc
3474 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3475 283243 Regression in ppc64 memcheck tests
3476 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3477 283427 re-connect epoll_pwait syscall on ARM linux
3478 283600 gdbsrv: android: port vgdb.c
3479 283709 none/tests/faultstatus needs to account for page size
3480 284305 filter_gdb needs enhancement to work on ppc64
3481 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3482 284472 Thumb2 ROR.W encoding T2 not implemented
3483 284621 XML-escape process command line in XML output
3484 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3485 that have non-power-of-2 sizes (also AMDs)
3486 n-i-bz don't be spooked by libraries mashed by elfhack
3487 n-i-bz don't be spooked by libxul.so linked with gold
3488 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3490 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3491 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3492 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3496 Release 3.6.1 (16 February 2011)
3497 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3498 3.6.1 is a bug fix release. It adds support for some SSE4
3499 instructions that were omitted in 3.6.0 due to lack of time. Initial
3500 support for glibc-2.13 has been added. A number of bugs causing
3501 crashing or assertion failures have been fixed.
3503 The following bugs have been fixed or resolved. Note that "n-i-bz"
3504 stands for "not in bugzilla" -- that is, a bug that was reported to us
3505 but never got a bugzilla entry. We encourage you to file bugs in
3506 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3507 mailing the developers (or mailing lists) directly -- bugs that are
3508 not entered into bugzilla tend to get forgotten about or ignored.
3510 To see details of a given bug, visit
3511 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3512 where XXXXXX is the bug number as listed below.
3514 188572 Valgrind on Mac should suppress setenv() mem leak
3515 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3516 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3517 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3518 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3519 254420 memory pool tracking broken
3520 254957 Test code failing to compile due to changes in memcheck.h
3521 255009 helgrind/drd: crash on chmod with invalid parameter
3522 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3523 255355 helgrind/drd: crash on threaded programs doing fork
3525 255418 (SSE4.x) rint call compiled with ICC
3526 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3527 255888 closing valgrindoutput tag outputted to log-stream on error
3528 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3529 255966 Slowness when using mempool annotations
3530 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3531 256600 super-optimized strcasecmp() false positive
3532 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3533 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3534 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3535 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3536 257276 Missing case in memcheck --track-origins=yes
3537 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3538 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3539 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3540 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3541 263099 callgrind_annotate counts Ir improperly [...]
3542 263877 undefined coprocessor instruction on ARMv7
3543 265964 configure FAIL with glibc-2.13
3544 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3545 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3546 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3547 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3548 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3550 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3554 Release 3.6.0 (21 October 2010)
3555 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3556 3.6.0 is a feature release with many significant improvements and the
3557 usual collection of bug fixes.
3559 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3560 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3561 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3563 -------------------------
3565 Here are some highlights. Details are shown further down:
3567 * Support for ARM/Linux.
3569 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3571 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3573 * Support for the SSE4.2 instruction set.
3575 * Enhancements to the Callgrind profiler, including the ability to
3576 handle CPUs with three levels of cache.
3578 * A new experimental heap profiler, DHAT.
3580 * A huge number of bug fixes and small enhancements.
3582 -------------------------
3584 Here are details of the above changes, together with descriptions of
3585 many other changes, and a list of fixed bugs.
3587 * ================== PLATFORM CHANGES =================
3589 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3590 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3591 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3593 This requires a CPU capable of running the ARMv7-A instruction set
3594 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3595 of the user space instruction set, including ARM and Thumb integer
3596 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3597 Cachegrind and Massif tools work properly; other tools work to
3600 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3601 with support for recent releases of the underlying toolchain
3602 components, notably gcc-4.5 and glibc-2.12.
3604 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3605 support also works much better on OS X 10.5, and is as solid as
3608 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3609 64-bit mode. In 32-bit mode, support is only available up to and
3610 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3611 supported in 64-bit mode, and 32-bit mode does in fact support the
3612 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3613 10.6 on 32-bit targets.
3615 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3616 and including version 2.05 is supported.
3618 * ==================== TOOL CHANGES ====================
3620 * Cachegrind has a new processing script, cg_diff, which finds the
3621 difference between two profiles. It's very useful for evaluating
3622 the performance effects of a change in a program.
3624 Related to this change, the meaning of cg_annotate's (rarely-used)
3625 --threshold option has changed; this is unlikely to affect many
3626 people, if you do use it please see the user manual for details.
3628 * Callgrind now can do branch prediction simulation, similar to
3629 Cachegrind. In addition, it optionally can count the number of
3630 executed global bus events. Both can be used for a better
3631 approximation of a "Cycle Estimation" as derived event (you need to
3632 update the event formula in KCachegrind yourself).
3634 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3635 rather than the L2 cache. This is to accommodate machines with
3636 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3637 cache configuration of such a machine it will run the simulation as
3638 if the L2 cache isn't present. This means the results are less
3639 likely to match the true result for the machine, but
3640 Cachegrind/Callgrind's results are already only approximate, and
3641 should not be considered authoritative. The results are still
3642 useful for giving a general idea about a program's locality.
3644 * Massif has a new option, --pages-as-heap, which is disabled by
3645 default. When enabled, instead of tracking allocations at the level
3646 of heap blocks (as allocated with malloc/new/new[]), it instead
3647 tracks memory allocations at the level of memory pages (as mapped by
3648 mmap, brk, etc). Each mapped page is treated as its own block.
3649 Interpreting the page-level output is harder than the heap-level
3650 output, but this option is useful if you want to account for every
3651 byte of memory used by a program.
3653 * DRD has two new command-line options: --free-is-write and
3654 --trace-alloc. The former allows to detect reading from already freed
3655 memory, and the latter allows tracing of all memory allocations and
3658 * DRD has several new annotations. Custom barrier implementations can
3659 now be annotated, as well as benign races on static variables.
3661 * DRD's happens before / happens after annotations have been made more
3662 powerful, so that they can now also be used to annotate e.g. a smart
3663 pointer implementation.
3665 * Helgrind's annotation set has also been drastically improved, so as
3666 to provide to users a general set of annotations to describe locks,
3667 semaphores, barriers and condition variables. Annotations to
3668 describe thread-safe reference counted heap objects have also been
3671 * Memcheck has a new command-line option, --show-possibly-lost, which
3672 is enabled by default. When disabled, the leak detector will not
3673 show possibly-lost blocks.
3675 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3676 has been added. DHAT keeps track of allocated heap blocks, and also
3677 inspects every memory reference to see which block (if any) is being
3678 accessed. This gives a lot of insight into block lifetimes,
3679 utilisation, turnover, liveness, and the location of hot and cold
3680 fields. You can use DHAT to do hot-field profiling.
3682 * ==================== OTHER CHANGES ====================
3684 * Improved support for unfriendly self-modifying code: the extra
3685 overhead incurred by --smc-check=all has been reduced by
3686 approximately a factor of 5 as compared with 3.5.0.
3688 * Ability to show directory names for source files in error messages.
3689 This is combined with a flexible mechanism for specifying which
3690 parts of the paths should be shown. This is enabled by the new flag
3693 * A new flag, --require-text-symbol, which will stop the run if a
3694 specified symbol is not found it a given shared object when it is
3695 loaded into the process. This makes advanced working with function
3696 intercepting and wrapping safer and more reliable.
3698 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3699 and control of Valgrind is now available for the tools Memcheck and
3700 Helgrind. XML output from Valgrind is available for Memcheck,
3701 Helgrind and exp-Ptrcheck.
3703 * More reliable stack unwinding on amd64-linux, particularly in the
3704 presence of function wrappers, and with gcc-4.5 compiled code.
3706 * Modest scalability (performance improvements) for massive
3707 long-running applications, particularly for those with huge amounts
3710 * Support for analyzing programs running under Wine with has been
3711 improved. The header files <valgrind/valgrind.h>,
3712 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3713 Windows-programs compiled with MinGW or one of the Microsoft Visual
3716 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3717 The 32-bit simulator was not affected. This did not occur often,
3718 but when it did would usually crash the program under test.
3721 * A large number of bugs were fixed. These are shown below.
3723 * A number of bugs were investigated, and were candidates for fixing,
3724 but are not fixed in 3.6.0, due to lack of developer time. They may
3725 get fixed in later releases. They are:
3727 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3728 212419 false positive "lock order violated" (A+B vs A)
3729 213685 Undefined value propagates past dependency breaking instruction
3730 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3731 237920 valgrind segfault on fork failure
3732 242137 support for code compiled by LLVM-2.8
3733 242423 Another unknown Intel cache config value
3734 243232 Inconsistent Lock Orderings report with trylock
3735 243483 ppc: callgrind triggers VEX assertion failure
3736 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3737 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3739 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3740 249435 Analyzing wine programs with callgrind triggers a crash
3741 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3742 250065 Handling large allocations
3743 250101 huge "free" memory usage due to m_mallocfree.c
3744 "superblocks fragmentation"
3745 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3746 252091 Callgrind on ARM does not detect function returns correctly
3747 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3748 254420 memory pool tracking broken
3749 n-i-bz support for adding symbols for JIT generated code
3752 The following bugs have been fixed or resolved. Note that "n-i-bz"
3753 stands for "not in bugzilla" -- that is, a bug that was reported to us
3754 but never got a bugzilla entry. We encourage you to file bugs in
3755 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3756 mailing the developers (or mailing lists) directly -- bugs that are
3757 not entered into bugzilla tend to get forgotten about or ignored.
3759 To see details of a given bug, visit
3760 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3761 where XXXXXX is the bug number as listed below.
3763 135264 dcbzl instruction missing
3765 153699 Valgrind should report unaligned reads with movdqa
3767 190429 Valgrind reports lost of errors in ld.so
3768 with x86_64 2.9.90 glibc
3769 197266 valgrind appears to choke on the xmms instruction
3771 197988 Crash when demangling very large symbol names
3772 202315 unhandled syscall: 332 (inotify_init1)
3773 203256 Add page-level profiling to Massif
3774 205093 dsymutil=yes needs quotes, locking (partial fix)
3775 205241 Snow Leopard 10.6 support (partial fix)
3776 206600 Leak checker fails to upgrade indirect blocks when their
3777 parent becomes reachable
3778 210935 port valgrind.h (not valgrind) to win32 so apps run under
3779 wine can make client requests
3780 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3781 within Linux ip-stack checksum functions
3782 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3784 213685 Undefined value propagates past dependency breaking instruction
3786 215914 Valgrind inserts bogus empty environment variable
3788 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3789 222545 shmat fails under valgind on some arm targets
3790 222560 ARM NEON support
3793 232509 Docs build fails with formatting inside <title></title> elements
3795 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3796 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3797 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3798 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3799 237485 symlink (syscall 57) is not supported on Mac OS
3800 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3802 238208 is_just_below_ESP doesn't take into account red-zone
3803 238345 valgrind passes wrong $0 when executing a shell script
3804 238679 mq_timedreceive syscall doesn't flag the reception buffer
3806 238696 fcntl command F_DUPFD_CLOEXEC not supported
3807 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3808 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3809 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
3811 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
3817 242606 unhandled syscall: setegid (in Ptrcheck)
3818 242814 Helgrind "Impossible has happened" during
3819 QApplication::initInstance();
3820 243064 Valgrind attempting to read debug information from iso
3821 243270 Make stack unwinding in Valgrind wrappers more reliable
3822 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
3823 sysno = 277 (mq_open)
3824 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
3825 244493 ARM VFP d16-d31 registers support
3826 244670 add support for audit_session_self syscall on Mac OS 10.6
3827 244921 The xml report of helgrind tool is not well format
3828 244923 In the xml report file, the <preamble> not escape the
3829 xml char, eg '<','&','>'
3830 245535 print full path names in plain text reports
3831 245925 x86-64 red zone handling problem
3832 246258 Valgrind not catching integer underruns + new [] s
3833 246311 reg/reg cmpxchg doesn't work on amd64
3834 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
3835 246888 Improve Makefile.vex.am
3836 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
3837 to [f]chmod_extended
3838 247526 IBM POWER6 (ISA 2.05) support is incomplete
3839 247561 Some leak testcases fails due to reachable addresses in
3841 247875 sizeofIRType to handle Ity_I128
3842 247894 [PATCH] unhandled syscall sys_readahead
3843 247980 Doesn't honor CFLAGS passed to configure
3844 248373 darwin10.supp is empty in the trunk
3845 248822 Linux FIBMAP ioctl has int parameter instead of long
3846 248893 [PATCH] make readdwarf.c big endianess safe to enable
3847 unwinding on big endian systems
3848 249224 Syscall 336 not supported (SYS_proc_info)
3850 249775 Incorrect scheme for detecting NEON capabilities of host CPU
3851 249943 jni JVM init fails when using valgrind
3852 249991 Valgrind incorrectly declares AESKEYGENASSIST support
3854 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
3855 250799 frexp$fenv_access_off function generates SIGILL
3856 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
3857 251251 support pclmulqdq insn
3858 251362 valgrind: ARM: attach to debugger either fails or provokes
3860 251674 Unhandled syscall 294
3863 254257 Add support for debugfiles found by build-id
3864 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
3865 254646 Wrapped functions cause stack misalignment on OS X
3866 (and possibly Linux)
3867 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
3869 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
3873 Release 3.5.0 (19 August 2009)
3874 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3875 3.5.0 is a feature release with many significant improvements and the
3876 usual collection of bug fixes. The main improvement is that Valgrind
3877 now works on Mac OS X.
3879 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
3880 and X86/Darwin. Support for recent distros and toolchain components
3881 (glibc 2.10, gcc 4.5) has been added.
3883 -------------------------
3885 Here is a short summary of the changes. Details are shown further
3888 * Support for Mac OS X (10.5.x).
3890 * Improvements and simplifications to Memcheck's leak checker.
3892 * Clarification and simplifications in various aspects of Valgrind's
3895 * XML output for Helgrind and Ptrcheck.
3897 * Performance and stability improvements for Helgrind and DRD.
3899 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
3901 * A new experimental tool, BBV, useful for computer architecture
3904 * Improved Wine support, including ability to read Windows PDB
3907 -------------------------
3909 Here are details of the above changes, followed by descriptions of
3910 many other minor changes, and a list of fixed bugs.
3913 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
3914 called "Darwin" because that is the name of the OS core, which is the
3915 level that Valgrind works at.)
3919 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
3920 because it would require work and 10.4 is only becoming less common.
3922 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
3923 fairly well. For 10.5.x, 32-bit programs are the default even on
3924 64-bit machines, so it handles most current programs.
3926 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
3927 officially supported, but simple programs at least will probably work.
3928 However, start-up is slow.
3930 - PowerPC machines are not supported.
3932 Things that don't work:
3934 - The Ptrcheck tool.
3936 - Objective-C garbage collection.
3940 - If you have Rogue Amoeba's "Instant Hijack" program installed,
3941 Valgrind will fail with a SIGTRAP at start-up. See
3942 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
3947 - You will likely find --dsymutil=yes a useful option, as error
3948 messages may be imprecise without it.
3950 - Mac OS X support is new and therefore will be less robust than the
3951 Linux support. Please report any bugs you find.
3953 - Threaded programs may run more slowly than on Linux.
3955 Many thanks to Greg Parker for developing this port over several years.
3958 * Memcheck's leak checker has been improved.
3960 - The results for --leak-check=summary now match the summary results
3961 for --leak-check=full. Previously they could differ because
3962 --leak-check=summary counted "indirectly lost" blocks and
3963 "suppressed" blocks as "definitely lost".
3965 - Blocks that are only reachable via at least one interior-pointer,
3966 but are directly pointed to by a start-pointer, were previously
3967 marked as "still reachable". They are now correctly marked as
3970 - The default value for the --leak-resolution option has been
3971 changed from "low" to "high". In general, this means that more
3972 leak reports will be produced, but each leak report will describe
3973 fewer leaked blocks.
3975 - With --leak-check=full, "definitely lost" and "possibly lost"
3976 leaks are now considered as proper errors, ie. they are counted
3977 for the "ERROR SUMMARY" and affect the behaviour of
3978 --error-exitcode. These leaks are not counted as errors if
3979 --leak-check=summary is specified, however.
3981 - Documentation for the leak checker has been improved.
3984 * Various aspects of Valgrind's text output have changed.
3986 - Valgrind's start-up message has changed. It is shorter but also
3987 includes the command being run, which makes it easier to use
3988 --trace-children=yes. An example:
3990 - Valgrind's shut-down messages have also changed. This is most
3991 noticeable with Memcheck, where the leak summary now occurs before
3992 the error summary. This change was necessary to allow leaks to be
3993 counted as proper errors (see the description of the leak checker
3994 changes above for more details). This was also necessary to fix a
3995 longstanding bug in which uses of suppressions against leaks were
3996 not "counted", leading to difficulties in maintaining suppression
3997 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3999 - Behavior of -v has changed. In previous versions, -v printed out
4000 a mixture of marginally-user-useful information, and tool/core
4001 statistics. The statistics printing has now been moved to its own
4002 flag, --stats=yes. This means -v is less verbose and more likely
4003 to convey useful end-user information.
4005 - The format of some (non-XML) stack trace entries has changed a
4006 little. Previously there were six possible forms:
4008 0x80483BF: really (a.c:20)
4009 0x80483BF: really (in /foo/a.out)
4011 0x80483BF: (within /foo/a.out)
4012 0x80483BF: ??? (a.c:20)
4015 The third and fourth of these forms have been made more consistent
4016 with the others. The six possible forms are now:
4018 0x80483BF: really (a.c:20)
4019 0x80483BF: really (in /foo/a.out)
4020 0x80483BF: really (in ???)
4021 0x80483BF: ??? (in /foo/a.out)
4022 0x80483BF: ??? (a.c:20)
4025 Stack traces produced when --xml=yes is specified are different
4029 * Helgrind and Ptrcheck now support XML output, so they can be used
4030 from GUI tools. Also, the XML output mechanism has been
4033 - The XML format has been overhauled and generalised, so it is more
4034 suitable for error reporting tools in general. The Memcheck
4035 specific aspects of it have been removed. The new format, which
4036 is an evolution of the old format, is described in
4037 docs/internals/xml-output-protocol4.txt.
4039 - Memcheck has been updated to use the new format.
4041 - Helgrind and Ptrcheck are now able to emit output in this format.
4043 - The XML output mechanism has been overhauled. XML is now output
4044 to its own file descriptor, which means that:
4046 * Valgrind can output text and XML independently.
4048 * The longstanding problem of XML output being corrupted by
4049 unexpected un-tagged text messages is solved.
4051 As before, the destination for text output is specified using
4052 --log-file=, --log-fd= or --log-socket=.
4054 As before, XML output for a tool is enabled using --xml=yes.
4056 Because there's a new XML output channel, the XML output
4057 destination is now specified by --xml-file=, --xml-fd= or
4060 Initial feedback has shown this causes some confusion. To
4061 clarify, the two envisaged usage scenarios are:
4063 (1) Normal text output. In this case, do not specify --xml=yes
4064 nor any of --xml-file=, --xml-fd= or --xml-socket=.
4066 (2) XML output. In this case, specify --xml=yes, and one of
4067 --xml-file=, --xml-fd= or --xml-socket= to select the XML
4068 destination, one of --log-file=, --log-fd= or --log-socket=
4069 to select the destination for any remaining text messages,
4070 and, importantly, -q.
4072 -q makes Valgrind completely silent on the text channel,
4073 except in the case of critical failures, such as Valgrind
4074 itself segfaulting, or failing to read debugging information.
4075 Hence, in this scenario, it suffices to check whether or not
4076 any output appeared on the text channel. If yes, then it is
4077 likely to be a critical error which should be brought to the
4078 attention of the user. If no (the text channel produced no
4079 output) then it can be assumed that the run was successful.
4081 This allows GUIs to make the critical distinction they need to
4082 make (did the run fail or not?) without having to search or
4083 filter the text output channel in any way.
4085 It is also recommended to use --child-silent-after-fork=yes in
4089 * Improvements and changes in Helgrind:
4091 - XML output, as described above
4093 - Checks for consistent association between pthread condition
4094 variables and their associated mutexes are now performed.
4096 - pthread_spinlock functions are supported.
4098 - Modest performance improvements.
4100 - Initial (skeletal) support for describing the behaviour of
4101 non-POSIX synchronisation objects through ThreadSanitizer
4102 compatible ANNOTATE_* macros.
4104 - More controllable tradeoffs between performance and the level of
4105 detail of "previous" accesses in a race. There are now three
4108 * --history-level=full. This is the default, and was also the
4109 default in 3.4.x. It shows both stacks involved in a race, but
4110 requires a lot of memory and can be very slow in programs that
4111 do many inter-thread synchronisation events.
4113 * --history-level=none. This only shows the later stack involved
4114 in a race. This can be much faster than --history-level=full,
4115 but makes it much more difficult to find the other access
4116 involved in the race.
4118 The new intermediate setting is
4120 * --history-level=approx
4122 For the earlier (other) access, two stacks are presented. The
4123 earlier access is guaranteed to be somewhere in between the two
4124 program points denoted by those stacks. This is not as useful
4125 as showing the exact stack for the previous access (as per
4126 --history-level=full), but it is better than nothing, and it's
4127 almost as fast as --history-level=none.
4130 * New features and improvements in DRD:
4132 - The error messages printed by DRD are now easier to interpret.
4133 Instead of using two different numbers to identify each thread
4134 (Valgrind thread ID and DRD thread ID), DRD does now identify
4135 threads via a single number (the DRD thread ID). Furthermore
4136 "first observed at" information is now printed for all error
4137 messages related to synchronization objects.
4139 - Added support for named semaphores (sem_open() and sem_close()).
4141 - Race conditions between pthread_barrier_wait() and
4142 pthread_barrier_destroy() calls are now reported.
4144 - Added support for custom allocators through the macros
4145 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
4146 in <valgrind/valgrind.h>). An alternative for these two macros is
4147 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
4150 - Added support for annotating non-POSIX synchronization objects
4151 through several new ANNOTATE_*() macros.
4153 - OpenMP: added support for the OpenMP runtime (libgomp) included
4154 with gcc versions 4.3.0 and 4.4.0.
4158 - Added two new command-line options (--first-race-only and
4159 --segment-merging-interval).
4162 * Genuinely atomic support for x86/amd64/ppc atomic instructions
4164 Valgrind will now preserve (memory-access) atomicity of LOCK-
4165 prefixed x86/amd64 instructions, and any others implying a global
4166 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
4168 This means that Valgrinded processes will "play nicely" in
4169 situations where communication with other processes, or the kernel,
4170 is done through shared memory and coordinated with such atomic
4171 instructions. Prior to this change, such arrangements usually
4172 resulted in hangs, races or other synchronisation failures, because
4173 Valgrind did not honour atomicity of such instructions.
4176 * A new experimental tool, BBV, has been added. BBV generates basic
4177 block vectors for use with the SimPoint analysis tool, which allows
4178 a program's overall behaviour to be approximated by running only a
4179 fraction of it. This is useful for computer architecture
4180 researchers. You can run BBV by specifying --tool=exp-bbv (the
4181 "exp-" prefix is short for "experimental"). BBV was written by
4185 * Modestly improved support for running Windows applications under
4186 Wine. In particular, initial support for reading Windows .PDB debug
4187 information has been added.
4190 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
4191 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
4195 * The Valgrind client requests VALGRIND_PRINTF and
4196 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
4197 the string was always printed immediately on its own line. Now, the
4198 string will be added to a buffer but not printed until a newline is
4199 encountered, or other Valgrind output is printed (note that for
4200 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
4201 "other Valgrind output"). This allows you to use multiple
4202 VALGRIND_PRINTF calls to build up a single output line, and also to
4203 print multiple output lines with a single request (by embedding
4204 multiple newlines in the string).
4207 * The graphs drawn by Massif's ms_print program have changed slightly:
4209 - The half-height chars '.' and ',' are no longer drawn, because
4210 they are confusing. The --y option can be used if the default
4211 y-resolution is not high enough.
4213 - Horizontal lines are now drawn after the top of a snapshot if
4214 there is a gap until the next snapshot. This makes it clear that
4215 the memory usage has not dropped to zero between snapshots.
4218 * Something that happened in 3.4.0, but wasn't clearly announced: the
4219 option --read-var-info=yes can be used by some tools (Memcheck,
4220 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
4221 variable type and location information. This makes those tools
4222 start up more slowly and increases memory consumption, but
4223 descriptions of data addresses in error messages become more
4227 * exp-Omega, an experimental instantaneous leak-detecting tool, was
4228 disabled in 3.4.0 due to a lack of interest and maintenance,
4229 although the source code was still in the distribution. The source
4230 code has now been removed from the distribution. For anyone
4231 interested, the removal occurred in SVN revision r10247.
4234 * Some changes have been made to the build system.
4236 - VEX/ is now integrated properly into the build system. This means
4237 that dependency tracking within VEX/ now works properly, "make
4238 install" will work without requiring "make" before it, and
4239 parallel builds (ie. 'make -j') now work (previously a
4240 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
4241 was effectively ignored).
4243 - The --with-vex configure option has been removed. It was of
4244 little use and removing it simplified the build system.
4246 - The location of some install files has changed. This should not
4247 affect most users. Those who might be affected:
4249 * For people who use Valgrind with MPI programs, the installed
4250 libmpiwrap.so library has moved from
4251 $(INSTALL)/<platform>/libmpiwrap.so to
4252 $(INSTALL)/libmpiwrap-<platform>.so.
4254 * For people who distribute standalone Valgrind tools, the
4255 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
4256 have moved to $(INSTALL)/libcoregrind-<platform>.a.
4258 These changes simplify the build system.
4260 - Previously, all the distributed suppression (*.supp) files were
4261 installed. Now, only default.supp is installed. This should not
4262 affect users as the other installed suppression files were not
4263 read; the fact that they were installed was a mistake.
4266 * KNOWN LIMITATIONS:
4268 - Memcheck is unusable with the Intel compiler suite version 11.1,
4269 when it generates code for SSE2-and-above capable targets. This
4270 is because of icc's use of highly optimised inlined strlen
4271 implementations. It causes Memcheck to report huge numbers of
4272 false errors even in simple programs. Helgrind and DRD may also
4275 Versions 11.0 and earlier may be OK, but this has not been
4279 The following bugs have been fixed or resolved. Note that "n-i-bz"
4280 stands for "not in bugzilla" -- that is, a bug that was reported to us
4281 but never got a bugzilla entry. We encourage you to file bugs in
4282 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4283 mailing the developers (or mailing lists) directly -- bugs that are
4284 not entered into bugzilla tend to get forgotten about or ignored.
4286 To see details of a given bug, visit
4287 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4288 where XXXXXX is the bug number as listed below.
4290 84303 How about a LockCheck tool?
4291 91633 dereference of null ptr in vgPlain_st_basetype
4292 97452 Valgrind doesn't report any pthreads problems
4293 100628 leak-check gets assertion failure when using
4294 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4295 108528 NPTL pthread cleanup handlers not called
4296 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4297 110128 mallinfo is not implemented...
4298 110770 VEX: Generated files not always updated when making valgrind
4299 111102 Memcheck: problems with large (memory footprint) applications
4300 115673 Vex's decoder should never assert
4301 117564 False positive: Syscall param clone(child_tidptr) contains
4302 uninitialised byte(s)
4303 119404 executing ssh from inside valgrind fails
4304 133679 Callgrind does not write path names to sources with dwarf debug
4306 135847 configure.in problem with non gnu compilers (and possible fix)
4307 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4308 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4309 136230 memcheck reports "possibly lost", should be "still reachable"
4310 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4311 137904 Valgrind reports a memory leak when using POSIX threads,
4313 139076 valgrind VT_GETSTATE error
4314 142228 complaint of elf_dynamic_do_rela in trivial usage
4315 145347 spurious warning with USBDEVFS_REAPURB
4316 148441 (wine) can't find memory leak in Wine, win32 binary
4318 148742 Leak-check fails assert on exit
4319 149878 add (proper) check for calloc integer overflow
4320 150606 Call graph is broken when using callgrind control
4321 152393 leak errors produce an exit code of 0. I need some way to
4322 cause leak errors to result in a nonzero exit code.
4323 157154 documentation (leak-resolution doc speaks about num-callers
4324 def=4) + what is a loss record
4325 159501 incorrect handling of ALSA ioctls
4326 162020 Valgrinding an empty/zero-byte file crashes valgrind
4327 162482 ppc: Valgrind crashes while reading stabs information
4328 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4329 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4330 163560 VEX/test_main.c is missing from valgrind-3.3.1
4331 164353 malloc_usable_size() doesn't return a usable size
4332 165468 Inconsistent formatting in memcheck manual -- please fix
4333 169505 main.c:286 (endOfInstr):
4334 Assertion 'ii->cost_offset == *cost_offset' failed
4335 177206 Generate default.supp during compile instead of configure
4336 177209 Configure valt_load_address based on arch+os
4337 177305 eventfd / syscall 323 patch lost
4338 179731 Tests fail to build because of inlining of non-local asm labels
4339 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4340 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4341 181594 Bogus warning for empty text segment
4342 181707 dwarf doesn't require enumerations to have name
4343 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4344 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4345 Assertion '!already_present' failed.
4346 185359 exp-ptrcheck: unhandled syscall getresuid()
4347 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4348 185816 Valgrind is unable to handle debug info for files with split
4349 debug info that are prelinked afterwards
4350 185980 [darwin] unhandled syscall: sem_open
4351 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4352 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4353 186790 Suppression pattern used for leaks are not reported
4354 186796 Symbols with length>200 in suppression files are ignored
4355 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4356 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4357 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4358 188046 bashisms in the configure script
4359 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4360 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4361 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4362 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4363 assertion fail "!lock->heldBy"
4364 188427 Add support for epoll_create1 (with patch)
4365 188530 Support for SIOCGSTAMPNS
4366 188560 Include valgrind.spec in the tarball
4367 188572 Valgrind on Mac should suppress setenv() mem leak
4368 189054 Valgrind fails to build because of duplicate non-local asm labels
4369 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4370 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4371 189763 drd assertion failure: s_threadinfo[tid].is_recording
4372 190219 unhandled syscall: 328 (x86-linux)
4373 190391 dup of 181394; see above
4374 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4375 190820 No debug information on powerpc-linux
4376 191095 PATCH: Improve usbdevfs ioctl handling
4377 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4379 191189 --xml=yes should obey --gen-suppressions=all
4380 191192 syslog() needs a suppression on macosx
4381 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4382 191761 getrlimit on MacOSX
4383 191992 multiple --fn-skip only works sometimes; dependent on order
4384 192634 V. reports "aspacem sync_check_mapping_callback:
4385 segment mismatch" on Darwin
4386 192954 __extension__ missing on 2 client requests
4387 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4388 194474 "INSTALL" file has different build instructions than "README"
4389 194671 Unhandled syscall (sem_wait?) from mac valgrind
4390 195069 memcheck: reports leak (memory still reachable) for
4392 195169 drd: (vgDrd_barrier_post_wait):
4393 Assertion 'r->sg[p->post_iteration]' failed.
4394 195268 valgrind --log-file doesn't accept ~/...
4395 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4396 195860 WARNING: unhandled syscall: unix:223
4397 196528 need a error suppression for pthread_rwlock_init under os x?
4398 197227 Support aio_* syscalls on Darwin
4399 197456 valgrind should reject --suppressions=(directory)
4400 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4401 197591 unhandled syscall 27 (mincore)
4402 197793 Merge DCAS branch to the trunk == 85756, 142103
4403 197794 Avoid duplicate filenames in Vex
4404 197898 make check fails on current SVN
4405 197901 make check fails also under exp-ptrcheck in current SVN
4406 197929 Make --leak-resolution=high the default
4407 197930 Reduce spacing between leak reports
4408 197933 Print command line of client at start-up, and shorten preamble
4409 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4410 198395 add BBV to the distribution as an experimental tool
4411 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4412 198649 callgrind_annotate doesn't cumulate counters
4413 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4414 199977 Valgrind complains about an unrecognized instruction in the
4415 atomic_incs test program
4416 200029 valgrind isn't able to read Fedora 12 debuginfo
4417 200760 darwin unhandled syscall: unix:284
4418 200827 DRD doesn't work on Mac OS X
4419 200990 VG_(read_millisecond_timer)() does not work correctly
4420 201016 Valgrind does not support pthread_kill() on Mac OS
4421 201169 Document --read-var-info
4422 201323 Pre-3.5.0 performance sanity checking
4423 201384 Review user manual for the 3.5.0 release
4424 201585 mfpvr not implemented on ppc
4425 201708 tests failing because x86 direction flag is left set
4426 201757 Valgrind doesn't handle any recent sys_futex additions
4427 204377 64-bit valgrind can not start a shell script
4428 (with #!/path/to/shell) if the shell is a 32-bit executable
4429 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4430 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4431 about memory allocation events with memory access tracing enabled
4432 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4434 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4438 Release 3.4.1 (28 February 2009)
4439 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4440 3.4.1 is a bug-fix release that fixes some regressions and assertion
4441 failures in debug info reading in 3.4.0, most notably incorrect stack
4442 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4443 other debug info problems are also fixed. A number of bugs in the
4444 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4446 In view of the fact that 3.4.0 contains user-visible regressions
4447 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4448 encouraged to ship 3.4.1 in preference to 3.4.0.
4450 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4451 bugzilla" -- that is, a bug that was reported to us but never got a
4452 bugzilla entry. We encourage you to file bugs in bugzilla
4453 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4454 developers (or mailing lists) directly -- bugs that are not entered
4455 into bugzilla tend to get forgotten about or ignored.
4457 n-i-bz Fix various bugs reading icc-11 generated debug info
4458 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4459 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4460 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4461 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4462 179618 exp-ptrcheck crashed / exit prematurely
4463 179624 helgrind: false positive races with pthread_create and
4464 recv/open/close/read
4465 134207 pkg-config output contains @VG_PLATFORM@
4466 176926 floating point exception at valgrind startup with PPC 440EPX
4467 181594 Bogus warning for empty text segment
4468 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4469 181707 Dwarf3 doesn't require enumerations to have name
4470 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4471 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4472 Assertion '!already_present' failed.
4473 185359 exp-ptrcheck unhandled syscall getresuid()
4475 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4476 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4480 Release 3.4.0 (2 January 2009)
4481 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4482 3.4.0 is a feature release with many significant improvements and the
4483 usual collection of bug fixes. This release supports X86/Linux,
4484 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4485 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4487 3.4.0 brings some significant tool improvements. Memcheck can now
4488 report the origin of uninitialised values, the thread checkers
4489 Helgrind and DRD are much improved, and we have a new experimental
4490 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4491 global arrays. In detail:
4493 * Memcheck is now able to track the origin of uninitialised values.
4494 When it reports an uninitialised value error, it will try to show
4495 the origin of the value, as either a heap or stack allocation.
4496 Origin tracking is expensive and so is not enabled by default. To
4497 use it, specify --track-origins=yes. Memcheck's speed will be
4498 essentially halved, and memory usage will be significantly
4499 increased. Nevertheless it can drastically reduce the effort
4500 required to identify the root cause of uninitialised value errors,
4501 and so is often a programmer productivity win, despite running more
4504 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4505 3.4.0, will be released shortly.
4507 * Helgrind's race detection algorithm has been completely redesigned
4508 and reimplemented, to address usability and scalability concerns:
4510 - The new algorithm has a lower false-error rate: it is much less
4511 likely to report races that do not really exist.
4513 - Helgrind will display full call stacks for both accesses involved
4514 in a race. This makes it easier to identify the root causes of
4517 - Limitations on the size of program that can run have been removed.
4519 - Performance has been modestly improved, although that is very
4522 - Direct support for Qt4 threading has been added.
4524 - pthread_barriers are now directly supported.
4526 - Helgrind works well on all supported Linux targets.
4528 * The DRD thread debugging tool has seen major improvements:
4530 - Greatly improved performance and significantly reduced memory
4533 - Support for several major threading libraries (Boost.Thread, Qt4,
4534 glib, OpenMP) has been added.
4536 - Support for atomic instructions, POSIX semaphores, barriers and
4537 reader-writer locks has been added.
4539 - Works now on PowerPC CPUs too.
4541 - Added support for printing thread stack usage at thread exit time.
4543 - Added support for debugging lock contention.
4545 - Added a manual for Drd.
4547 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4548 checks for misuses of pointers. In that sense it is a bit like
4549 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4550 detect overruns of stack and global arrays, it can detect
4551 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4552 detect accesses heap blocks that have been freed a very long time
4553 ago (millions of blocks in the past).
4555 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4556 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4557 of the main Valgrind documentation. As this is an experimental
4558 tool, we would be particularly interested in hearing about your
4559 experiences with it.
4561 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4562 longer built by default, although the code remains in the repository
4563 and the tarball. This is due to three factors: a perceived lack of
4564 users, a lack of maintenance, and concerns that it may not be
4565 possible to achieve reliable operation using the existing design.
4567 * As usual, support for the latest Linux distros and toolchain
4568 components has been added. It should work well on Fedora Core 10,
4569 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4570 state) is supported, as is glibc-2.9. The C++ demangler has been
4571 updated so as to work well with C++ compiled by even the most recent
4574 * You can now use frame-level wildcards in suppressions. This was a
4575 frequently-requested enhancement. A line "..." in a suppression now
4576 matches zero or more frames. This makes it easier to write
4577 suppressions which are precise yet insensitive to changes in
4580 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4582 * Very basic support for IBM Power6 has been added (64-bit processes only).
4584 * Valgrind is now cross-compilable. For example, it is possible to
4585 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4586 on a ppc32/64-linux target.
4588 * You can set the main thread's stack size at startup using the
4589 new --main-stacksize= flag (subject of course to ulimit settings).
4590 This is useful for running apps that need a lot of stack space.
4592 * The limitation that you can't use --trace-children=yes together
4593 with --db-attach=yes has been removed.
4595 * The following bugs have been fixed. Note that "n-i-bz" stands for
4596 "not in bugzilla" -- that is, a bug that was reported to us but
4597 never got a bugzilla entry. We encourage you to file bugs in
4598 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4599 mailing the developers (or mailing lists) directly.
4601 n-i-bz Make return types for some client requests 64-bit clean
4602 n-i-bz glibc 2.9 support
4603 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4604 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4605 n-i-bz Building in an env without gdb gives bogus gdb attach
4606 92456 Tracing the origin of uninitialised memory
4607 106497 Valgrind does not demangle some C++ template symbols
4609 151612 Suppression with "..." (frame-level wildcards in .supp files)
4610 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4611 159285 unhandled syscall:25 (stime, on x86-linux)
4612 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4613 160954 ppc build of valgrind crashes with illegal instruction (isel)
4614 160956 mallinfo implementation, w/ patch
4615 162092 Valgrind fails to start gnome-system-monitor
4616 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4617 163794 assertion failure with "--track-origins=yes"
4618 163933 sigcontext.err and .trapno must be set together
4619 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4620 164476 Missing kernel module loading system calls
4621 164669 SVN regression: mmap() drops posix file locks
4622 166581 Callgrind output corruption when program forks
4623 167288 Patch file for missing system calls on Cell BE
4624 168943 unsupported scas instruction pentium
4625 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4626 172417 x86->IR: 0x82 ...
4627 172563 amd64->IR: 0xD9 0xF5 - fprem1
4628 173099 .lds linker script generation error
4629 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4630 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4632 174908 --log-file value not expanded correctly for core file
4633 175044 Add lookup_dcookie for amd64
4634 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4636 Developer-visible changes:
4638 * Valgrind's debug-info reading machinery has been majorly overhauled.
4639 It can now correctly establish the addresses for ELF data symbols,
4640 which is something that has never worked properly before now.
4642 Also, Valgrind can now read DWARF3 type and location information for
4643 stack and global variables. This makes it possible to use the
4644 framework to build tools that rely on knowing the type and locations
4645 of stack and global variables, for example exp-Ptrcheck.
4647 Reading of such information is disabled by default, because most
4648 tools don't need it, and because it is expensive in space and time.
4649 However, you can force Valgrind to read it, using the
4650 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4651 make use of such information, if present, to provide source-level
4652 descriptions of data addresses in the error messages they create.
4654 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4655 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).