1 Release 3.??.? (?? 2021)
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 * ================== PLATFORM CHANGES =================
19 - Support the miscellaneous-instruction-extensions facility 3 and the
20 vector-enhancements facility 2. This enables programs compiled with
21 "-march=arch13" or "-march=z15" to be executed under Valgrind.
25 - ISA 3.1 support is now complete
26 - ISA 3.0 support for the darn instruction added.
27 - ISA 3.0 support for the vector system call instruction scv added.
28 - ISA 3.0 support for the copy, paste and cpabort instructions added.
30 * Support for X86/FreeBSD and AMD64/FreeBSD has been added.
32 * ==================== TOOL CHANGES ====================
34 * ==================== FIXED BUGS ====================
36 The following bugs have been fixed or resolved. Note that "n-i-bz"
37 stands for "not in bugzilla" -- that is, a bug that was reported to us
38 but never got a bugzilla entry. We encourage you to file bugs in
39 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
40 than mailing the developers (or mailing lists) directly -- bugs that
41 are not entered into bugzilla tend to get forgotten about or ignored.
43 208531 [PATCH]: FreeBSD support for valgrind
44 368960 WARNING: unhandled amd64-linux syscall: 163 (acct)
45 407589 [Linux] Add support for C11 aligned_alloc() and GNU reallocarray()
46 423963 Error in child thread when CLONE_PIDFD is used
47 429375 PPC ISA 3.1 support is missing, part 9
48 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
49 431306 Update demangler to support Rust v0 name mangling
50 432387 s390x: z15 instructions support
51 433437 FreeBSD support, part 1
52 433438 FreeBSD support, part 2
53 433439 FreeBSD support, part 3
54 433469 FreeBSD support, part 4
55 433473 FreeBSD support, part 5
56 433477 FreeBSD support, part 6
57 433479 FreeBSD support, part 7
58 433504 FreeBSD support, part 8
59 433506 FreeBSD support, part 9
60 433507 FreeBSD support, part 10
61 433508 FreeBSD support, part 11
62 433510 FreeBSD support, part 12
63 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
64 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
65 434296 s390x: False-positive memcheck diagnostics from vector string
67 434840 PPC64 darn instruction not supported
68 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
69 435908 valgrind tries to fetch from deubginfod for files which already
70 have debug information
71 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2 0x0 0x0
72 439046 valgrind is unusably large when linked with lld
73 439090 Implement close_range(2)
74 439326 Valgrind 3.17.0 won't compile with Intel 2021 oneAPI compilers
75 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
76 440670 unhandled ppc64le-linux syscall: 252 (statfs64) and 253 (fstatfs64)
77 440906 Fix impossible constraint issue in P10 testcase.
78 441512 Remove a unneeded / unnecessary prefix check.
79 441534 Update the expected output for test_isa_3_1_VRT.
80 442061 very slow execution under Fedora 34 (readdwarf3)
81 443031 Gcc -many change requires explicit .machine directives
82 443033 Add support for the ISA 3.0 mcrxrx instruction
83 443034 Sraw, srawi, srad, sradi, mfs
84 443178 Powerpc, test jm-mfspr expected output needs to be updated.
85 443179 Need new test for the lxvx and stxvx instructions on ISA 2.07 and
87 443180 The subnormal test and the ISA 3.0 test generate compiler warnings.
88 443314 In the latest GIT version, Valgrind with "--trace-flags" crashes at
90 443605 Don't call final_tidyup (__libc_freeres) on FatalSignal
92 To see details of a given bug, visit
93 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
94 where XXXXXX is the bug number as listed below.
97 Release 3.17.0 (19 Mar 2021)
98 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
100 3.17.0 fixes a number of bugs and adds some functional changes: support for
101 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
102 some new instructions for Arm64, S390 and POWER. There are also some tool
105 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
106 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
107 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
108 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
109 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
111 * ==================== CORE CHANGES ===================
113 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
116 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
117 debugging information. When a debuginfo file cannot be found locally,
118 Valgrind is able to query debuginfod servers for the file using its
119 build-id. See the user manual for more information about debuginfod support.
121 * ================== PLATFORM CHANGES =================
125 - Inaccuracies resulting from double-rounding in the simulation of
126 floating-point multiply-add/subtract instructions have been fixed. These
127 should now behave exactly as the hardware does.
129 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
130 ongoing. Support for the half-word variants of at least the following
131 instructions has been added:
133 FABS <Vd>.<T>, <Vn>.<T>
135 FNEG <Vd>.<T>, <Vn>.<T>
137 FSQRT <Vd>.<T>, <Vn>.<T>
142 - Implement the new instructions/features that were added to z/Architecture
143 with the vector-enhancements facility 1. Also cover the instructions from
144 the vector-packed-decimal facility that are defined outside the chapter
145 "Vector Decimal Instructions", but not the ones from that chapter itself.
147 For a detailed list of newly supported instructions see the updates to
148 `docs/internals/s390-opcodes.csv'.
150 Since the miscellaneous instruction extensions facility 2 was already
151 added in Valgrind 3.16.0, this completes the support necessary to run
152 general programs built with `--march=z14' under Valgrind. The
153 vector-packed-decimal facility is currently not exploited by the standard
154 toolchain and libraries.
158 - Various bug fixes. Fix for the sync field to limit setting just two of
159 the two bits in the L-field. Fix the write size for the stxsibx and
160 stxsihx instructions. Fix the modsw and modsd instructions.
162 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
163 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
164 conversion instructions are still missing.
166 * ==================== TOOL CHANGES ====================
168 * General tool changes
170 - All the tools and their vgpreload libraries are now installed under
171 libexec because they cannot be executed directly and should be run through
172 the valgrind executable. This should be an internal, not user visible,
173 change, but might impact valgrind packagers.
175 - The --track-fds option now respects -q, --quiet and won't output anything
176 if no file descriptors are leaked. It also won't report the standard stdin
177 (0), stdout (1) or stderr (2) descriptors as being leaked with
178 --trace-fds=yes anymore. To track whether the standard file descriptors
179 are still open at the end of the program run use --trace-fds=all.
183 - DHAT has been extended, with two new modes of operation. The new
184 --mode=copy flag triggers copy profiling, which records calls to memcpy,
185 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
186 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
187 the new dhat/dhat.h file. This is useful for learning more about hot code
188 paths. See the user manual for more information about the new modes.
190 - Because of these changes, DHAT's file format has changed. DHAT output
191 files produced with earlier versions of DHAT will not work with this
192 version of DHAT's viewer, and DHAT output files produced with this version
193 of DHAT will not work with earlier versions of DHAT's viewer.
195 * ==================== FIXED BUGS ====================
197 The following bugs have been fixed or resolved. Note that "n-i-bz"
198 stands for "not in bugzilla" -- that is, a bug that was reported to us
199 but never got a bugzilla entry. We encourage you to file bugs in
200 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
201 than mailing the developers (or mailing lists) directly -- bugs that
202 are not entered into bugzilla tend to get forgotten about or ignored.
204 To see details of a given bug, visit
205 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
206 where XXXXXX is the bug number as listed below.
208 140178 open("/proc/self/exe", ...); doesn't quite work
209 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
210 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
211 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
212 345077 linux syscall execveat support (linux 3.19)
213 361770 Missing F_ADD_SEALS
214 369029 handle linux syscalls sched_getattr and sched_setattr
215 384729 __libc_freeres inhibits cross-platform valgrind
216 388787 Support for C++17 new/delete
217 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
218 396656 Warnings while reading debug info
219 397605 ioctl FICLONE mishandled
220 401416 Compile failure with openmpi 4.0
221 408663 Suppression file for musl libc
222 404076 s390x: z14 vector instructions not implemented
223 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
224 (actually affected all x86 and nanomips regardless of host bitness)
225 413547 regression test does not check for Arm 64 features.
226 414268 Enable AArch64 feature detection and decoding for v8.x instructions
227 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
228 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
229 422261 platform selection fails for unqualified client name
230 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
231 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
232 423195 PPC ISA 3.1 support is missing, part 1
233 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
234 424012 crash with readv/writev having invalid but not NULL arg2 iovec
235 424298 amd64: Implement RDSEED
236 425232 PPC ISA 3.1 support is missing, part 2
237 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
238 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
239 426123 PPC ISA 3.1 support is missing, part 3
240 426144 Fix "condition variable has not been initialized" on Fedora 33.
241 427400 PPC ISA 3.1 support is missing, part 4
242 427401 PPC ISA 3.1 support is missing, part 5
243 427404 PPC ISA 3.1 support is missing, part 6
244 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
245 427787 Support new faccessat2 linux syscall (439)
246 427969 debuginfo section duplicates a section in the main ELF file
247 428035 drd: Unbreak the musl build
248 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
249 428716 cppcheck detects potential leak in VEX/useful/smchash.c
250 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
251 429352 PPC ISA 3.1 support is missing, part 7
252 429354 PPC ISA 3.1 support is missing, part 8
253 429692 unhandled ppc64le-linux syscall: 147 (getsid)
254 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
256 429952 Errors when building regtest with clang
257 430354 ppc stxsibx and stxsihx instructions write too much data
258 430429 valgrind.h doesn't compile on s390x with clang
259 430485 expr_is_guardable doesn't handle Iex_Qop
260 431556 Complete arm64 FADDP v8.2 instruction support
261 432102 Add support for DWARF5 as produced by GCC11
262 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
263 432381 drd: Process STACK_REGISTER client requests
264 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
265 432672 vg_regtest: test-specific environment variables not reset between tests
266 432809 VEX should support REX.W + POPF
267 432861 PPC modsw and modsd give incorrect results for 1 mod 12
268 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
269 432215 Add debuginfod functionality
270 433323 Use pkglibexecdir as vglibdir
271 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
272 433629 valgrind/README has type "abd" instead of "and"
273 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
274 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
275 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
276 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
277 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
279 (3.17.0.RC1: 13 Mar 2021)
280 (3.17.0.RC2: 17 Mar 2021)
281 (3.17.0: 19 Mar 2021)
285 Release 3.16.1 (22 June 2020)
286 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
288 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
289 fixes character encoding problems in the documentation HTML.
291 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
292 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
294 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
298 Release 3.16.0 (27 May 2020)
299 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
301 3.16.0 is a feature release with many improvements and the usual collection of
304 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
305 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
306 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
307 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
308 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
310 * ==================== CORE CHANGES ===================
312 * It is now possible to dynamically change the value of many command line
313 options while your program (or its children) are running under Valgrind.
315 To see the list of dynamically changeable options, run
316 "valgrind --help-dyn-options".
318 You can change the options from the shell by using vgdb to launch
319 the monitor command "v.clo <clo option>...".
320 The same monitor command can be used from a gdb connected
321 to the valgrind gdbserver.
322 Your program can also change the dynamically changeable options using
323 the client request VALGRIND_CLO_CHANGE(option).
325 * ================== PLATFORM CHANGES =================
327 * MIPS: preliminary support for nanoMIPS instruction set has been added.
329 * ==================== TOOL CHANGES ====================
333 - The implicit memcpy done by each call to realloc now counts towards the
334 read and write counts of resized heap blocks, making those counts higher
339 - cg_annotate's --auto and --show-percs options now default to 'yes', because
340 they are usually wanted.
344 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
345 because they are usually wanted.
347 - The command option --collect-systime has been enhanced to specify
348 the unit used to record the elapsed time spent during system calls.
349 The command option now accepts the values no|yes|msec|usec|nsec,
350 where yes is a synonym of msec. When giving the value nsec, the
351 system cpu time of system calls is also recorded.
355 - Several memcheck options are now dynamically changeable.
356 Use valgrind --help-dyn-options to list them.
358 - The release 3.15 introduced a backward incompatible change for
359 some suppression entries related to preadv and pwritev syscalls.
360 When reading a suppression entry using the unsupported 3.14 format,
361 valgrind will now produce a warning to say the suppression entry will not
362 work, and suggest the needed change.
364 - Significantly fewer false positive errors on optimised code generated by
365 Clang and GCC. In particular, Memcheck now deals better with the
366 situation where the compiler will transform C-level "A && B" into "B && A"
367 under certain circumstances (in which the transformation is valid).
368 Handling of integer equality/non-equality checks on partially defined
369 values is also improved on some architectures.
373 - The exprimental Stack and Global Array Checking tool has been removed.
374 It only ever worked on x86 and amd64, and even on those it had a
375 high false positive rate and was slow. An alternative for detecting
376 stack and global array overruns is using the AddressSanitizer (ASAN)
377 facility of the GCC and Clang compilers, which require you to rebuild
378 your code with -fsanitize=address.
380 * ==================== OTHER CHANGES ====================
382 * New and modified GDB server monitor features:
384 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
386 - The gdbserver monitor commands that require an address and an optional
387 length argument now accepts the alternate 'C like' syntax "address[length]".
388 For example, the memcheck command "monitor who_points_at 0x12345678 120"
389 can now also be given as "monitor who_points_at 0x12345678[120]".
391 * ==================== FIXED BUGS ====================
393 The following bugs have been fixed or resolved. Note that "n-i-bz"
394 stands for "not in bugzilla" -- that is, a bug that was reported to us
395 but never got a bugzilla entry. We encourage you to file bugs in
396 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
397 than mailing the developers (or mailing lists) directly -- bugs that
398 are not entered into bugzilla tend to get forgotten about or ignored.
400 To see details of a given bug, visit
401 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
402 where XXXXXX is the bug number as listed below.
404 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
405 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
406 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
407 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
408 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
409 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
410 400872 Add nanoMIPS support to Valgrind
411 403212 drd/tests/trylock hangs on FreeBSD
412 404406 s390x: z14 miscellaneous instructions not implemented
413 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
414 406561 mcinfcallWSRU gdbserver_test fails on ppc64
415 406824 Unsupported baseline
416 407218 Add support for the copy_file_range syscall
417 407307 Intercept stpcpy also in ld.so for arm64
418 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
420 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
421 408009 Expose rdrand and f16c even on avx if host cpu supports them
422 408091 Missing pkey syscalls
423 408414 Add support for missing for preadv2 and pwritev2 syscalls
424 409141 Valgrind hangs when SIGKILLed
425 409206 Support for Linux PPS and PTP ioctls
426 409367 exit_group() after signal to thread waiting in futex() causes hangs
427 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
428 409780 References to non-existent configure.in
429 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
430 410599 Non-deterministic behaviour of pth_self_kill_15_other test
431 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
432 411134 Allow the user to change a set of command line options during execution
433 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
434 412344 Problem setting mips flags with specific paths
435 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
436 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
437 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
438 413603 callgrind_annotate/cg_annotate truncate function names at '#'
439 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
440 415136 ARMv8.1 Compare-and-Swap instructions are not supported
441 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
442 416239 valgrind crashes when handling clock_adjtime
443 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
444 416286 DRD reports "conflicting load" error on std::mutex::lock()
445 416301 s390x: "compare and signal" not supported
446 416387 finit_module and bpf syscalls are unhandled on arm64
447 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
448 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
449 416753 new 32bit time syscalls for 2038+
450 417075 pwritev(vector[...]) suppression ignored
451 417075 is not fixed, but incompatible supp entries are detected
452 and a warning is produced for these.
453 417187 [MIPS] Conditional branch problem since 'grail' changes
454 417238 Test memcheck/tests/vbit-test fails on mips64 BE
455 417266 Make memcheck/tests/linux/sigqueue usable with musl
456 417281 s390x: /bin/true segfaults with "grail" enabled
457 417427 commit to fix vki_siginfo_t definition created numerous regression
459 417452 s390_insn_store_emit: dst->tag for HRcVec128
460 417578 Add suppressions for glibc DTV leaks
461 417906 clone with CLONE_VFORK and no CLONE_VM fails
462 418004 Grail code additions break ppc64.
463 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
464 418997 s390x: Support Iex_ITE for float and vector types
465 419503 s390x: Avoid modifying registers returned from isel functions
466 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
467 421570 std_mutex fails on Arm v8.1 h/w
468 434035 vgdb might crash if valgrind is killed
469 n-i-bz Fix minor one time leaks in dhat.
470 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
471 n-i-bz Add support for the Linux io_uring system calls
472 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
473 n-i-bz Fix non-glibc build of test suite with s390x_features
474 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
475 423195 PPC ISA 3.1 support is missing, part 1
477 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
478 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
479 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
483 Release 3.15.0 (12 April 2019)
484 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
486 3.15.0 is a feature release with many improvements and the usual collection of
489 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
490 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
491 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
492 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
493 support for X86/macOS 10.13 and AMD64/macOS 10.13.
495 * ==================== CORE CHANGES ===================
497 * The XTree Massif output format now makes use of the information obtained
498 when specifying --read-inline-info=yes.
500 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
502 * ==================== TOOL CHANGES ====================
506 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
507 it has been promoted from an experimental tool to a regular tool. Run it
508 with --tool=dhat instead of --tool=exp-dhat.
510 - DHAT now prints only minimal data when the program ends, instead writing
511 the bulk of the profiling data to a file. As a result, the --show-top-n
512 and --sort-by options have been removed.
514 - Profile results can be viewed with the new viewer, dh_view.html. When
515 a run ends, a short message is printed, explaining how to view the result.
517 - See the documentation for more details.
521 - cg_annotate has a new option, --show-percs, which prints percentages next
526 - callgrind_annotate has a new option, --show-percs, which prints percentages
527 next to all event counts.
529 - callgrind_annotate now inserts commas in call counts, and
530 sort the caller/callee lists in the call tree.
534 - The default value for --read-inline-info is now "yes" on
535 Linux/Android/Solaris. It is still "no" on other OS.
539 - The option --xtree-leak=yes (to output leak result in xtree format)
540 automatically activates the option --show-leak-kinds=all, as xtree
541 visualisation tools such as kcachegrind can in any case select what kind
542 of leak to visualise.
544 - There has been further work to avoid false positives. In particular,
545 integer equality on partially defined inputs (C == and !=) is now handled
548 * ==================== OTHER CHANGES ====================
550 * The new option --show-error-list=no|yes displays, at the end of the run, the
551 list of detected errors and the used suppressions. Prior to this change,
552 showing this information could only be done by specifying "-v -v", but that
553 also produced a lot of other possibly-non-useful messages. The option -s is
554 equivalent to --show-error-list=yes.
556 * ==================== FIXED BUGS ====================
558 The following bugs have been fixed or resolved. Note that "n-i-bz"
559 stands for "not in bugzilla" -- that is, a bug that was reported to us
560 but never got a bugzilla entry. We encourage you to file bugs in
561 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
562 than mailing the developers (or mailing lists) directly -- bugs that
563 are not entered into bugzilla tend to get forgotten about or ignored.
565 To see details of a given bug, visit
566 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
567 where XXXXXX is the bug number as listed below.
569 385411 s390x: z13 vector floating-point instructions not implemented
570 397187 z13 vector register support for vgdb gdbserver
571 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
572 398870 Please add support for instruction vcvtps2ph
573 399287 amd64 front end: Illegal Instruction vcmptrueps
574 399301 Use inlined frames in Massif XTree output.
575 399322 Improve callgrind_annotate output
576 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
577 400164 helgrind test encounters mips x-compiler warnings and assembler error
578 400490 s390x: VRs allocated as if separate from FPRs
579 400491 s390x: Operand of LOCH treated as unsigned integer
580 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
581 options, which specify a mips64 processor
582 401112 LLVM 5.0 generates comparison against partially initialized data
583 401277 More bugs in z13 support
584 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
585 401578 drd: crashes sometimes on fork()
586 401627 memcheck errors with glibc avx2 optimized wcsncmp
587 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
588 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
589 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
591 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
592 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
593 402123 invalid assembler opcodes for mips32r2
594 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
595 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
596 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
597 402351 mips64 libvexmultiarch_test fails on s390x
599 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
600 402480 Do not use %rsp in clobber list
601 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
602 402515 Implement new option --show-error-list=no|yes / -s
603 402519 POWER 3.0 addex instruction incorrectly implemented
604 402781 Redo the cache used to process indirect branch targets
605 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
606 403552 s390x: wrong facility bit checked for vector facility
607 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
608 404638 Add VG_(replaceIndexXA)
609 404843 s390x: backtrace sometimes ends prematurely
610 404888 autotools cleanup series
611 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
612 405182 Valgrind fails to build with Clang
613 405205 filter_libc: remove the line holding the futex syscall error entirely
614 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
615 the upper and lower 32-bits of the 64-bit result
616 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
617 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
618 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
620 405403 s390x disassembler cannot be used on x86
621 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
622 405458 MIPS mkFormVEC arguments swapped?
623 405716 drd: Fix an integer overflow in the stack margin calculation
624 405722 Support arm64 core dump
625 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
626 of the 64-bit destination field.
627 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
628 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
629 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
631 406256 PPC64, vector floating point instructions don't handle subnormal
632 according to VSCR[NJ] bit setting.
633 406352 cachegrind/callgrind fails ann tests because of missing a.c
634 406354 dhat is broken on x86 (32bit)
635 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
636 406357 gdbserver_tests fails because of gdb output change
637 406360 memcheck/tests/libstdc++.supp needs more supression variants
638 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
639 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
640 407340 PPC64, does not support the vlogefp, vexptefp instructions.
641 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
642 n-i-bz Fix callgrind_annotate non deterministic order for equal total
643 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
644 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
645 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
647 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
648 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
649 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
653 Release 3.14.0 (9 October 2018)
654 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
656 3.14.0 is a feature release with many improvements and the usual collection of
659 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
660 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
661 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
662 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
663 support for X86/macOS 10.13, AMD64/macOS 10.13.
665 * ==================== CORE CHANGES ===================
667 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
668 debug info for unloaded code. This allows saved stack traces (e.g. for
669 memory leaks) to include file/line info for code that has been dlclose'd (or
670 similar). See the user manual for more information and known limitations.
672 * Ability to specify suppressions based on source file name and line number.
674 * Majorly overhauled register allocator. No end-user changes, but the JIT
675 generates code a bit more quickly now.
677 * ================== PLATFORM CHANGES =================
679 * Preliminary support for macOS 10.13 has been added.
681 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
683 * mips: support for MIPS SIMD architecture (MSA) has been added.
685 * mips: support for MIPS N32 ABI has been added.
687 * s390: partial support for vector instructions (integer and string) has been
690 * ==================== TOOL CHANGES ====================
692 * Helgrind: Addition of a flag
693 --delta-stacktrace=no|yes [yes on linux amd64/x86]
694 which specifies how full history stack traces should be computed.
695 Setting this to =yes can speed up Helgrind by 25% when using
696 --history-level=full.
698 * Memcheck: reduced false positive rate for optimised code created by Clang 6
699 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
700 blocks more carefully to determine where it can avoid expensive definedness
701 checks without loss of precision. This is controlled by the flag
702 --expensive-definedness-checks=no|auto|yes [auto].
704 * ==================== OTHER CHANGES ====================
706 * Valgrind is now buildable with link-time optimisation (LTO). A new
707 configure option --enable-lto=yes allows building Valgrind with LTO. If the
708 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
709 Note that if you are doing Valgrind development, --enable-lto=yes massively
710 slows down the build process.
712 * ==================== FIXED BUGS ====================
714 The following bugs have been fixed or resolved. Note that "n-i-bz"
715 stands for "not in bugzilla" -- that is, a bug that was reported to us
716 but never got a bugzilla entry. We encourage you to file bugs in
717 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
718 than mailing the developers (or mailing lists) directly -- bugs that
719 are not entered into bugzilla tend to get forgotten about or ignored.
721 To see details of a given bug, visit
722 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
723 where XXXXXX is the bug number as listed below.
725 79362 Debug info is lost for .so files when they are dlclose'd
726 208052 strlcpy error when n = 0
727 255603 exp-sgcheck Assertion '!already_present' failed
728 338252 building valgrind with -flto (link time optimisation) fails
729 345763 MIPS N32 ABI support
730 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
731 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
732 372347 Replacement problem of the additional c++14/c++17 new/delete operators
733 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
734 376257 helgrind history full speed up using a cached stack
735 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
737 379748 Fix missing pselect syscall (OS X 10.11)
738 379754 Fix missing syscall ulock_wait (OS X 10.12)
739 380397 s390x: __GI_strcspn() replacemenet needed
740 381162 possible array overrun in VEX register allocator
741 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
742 381274 powerpc too chatty even with --sigill-diagnostics=no
743 381289 epoll_pwait can have a NULL sigmask
744 381553 VEX register allocator v3
745 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
746 381769 Use ucontext_t instead of struct ucontext
747 381805 arm32 needs ld.so index hardwire for new glibc security fixes
748 382256 gz compiler flag test doesn't work for gold
749 382407 vg_perf needs "--terse" command line option
750 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
751 382563 MIPS MSA ASE support
752 382998 xml-socket doesn't work
753 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
754 383723 Fix missing kevent_qos syscall (macOS 10.11)
755 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
756 384096 Mention AddrCheck at Memcheck's command line option [..]
757 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
758 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
759 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
760 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
761 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
762 384337 performance improvements to VEX register allocator v2 and v3
763 384526 reduce number of spill insns generated by VEX register allocator v3
764 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
765 384631 Sanitise client args as printed with -v
766 384633 Add a simple progress-reporting facility
767 384987 VEX regalloc: allocate caller-save registers for short lived vregs
768 385055 PPC VEX temporary storage exhausted
769 385182 PPC64 is missing support for the DSCR
770 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
771 385207 PPC64, generate_store_FPRF() generates too many Iops
772 385208 PPC64, xxperm instruction exhausts temporary memory
773 385210 PPC64, vpermr instruction could exhaust temporary memory
774 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
775 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
776 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
777 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
778 385408 s390x: z13 vector "support" instructions not implemented
779 385409 s390x: z13 vector integer instructions not implemented
780 385410 s390x: z13 vector string instructions not implemented
781 385412 s390x: new non-vector z13 instructions not implemented
782 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
783 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
784 385939 Optionally exit on the first error
785 386318 valgrind.org/info/tools.html is missing SGCheck
786 386425 running valgrind + wine on armv7l gives illegal opcode
787 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
788 387410 MIPSr6 support
789 387664 Memcheck: make expensive-definedness-checks be the default
790 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
791 387766 asm shifts cause false positive "Conditional jump or move depends
792 on uninitialised value"
793 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
794 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
795 388786 Support bpf syscall in amd64 Linux
796 388862 Add replacements for wmemchr and wcsnlen on Linux
797 389065 valgrind meets gcc flag -Wlogical-op
798 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
799 390471 suppression by specification of source-file line number
800 390723 make xtree dump files world wide readable, similar to log files
801 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
802 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
803 392118 unhandled amd64-linux syscall: 332 (statx)
804 392449 callgrind not clearing the number of calls properly
805 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
806 lxssp, stxssp and stxvl instructions.
807 393023 callgrind_control risks using the wrong vgdb
808 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
809 393099 posix_memalign() invalid write if alignment == 0
810 393146 failing assert "is_DebugInfo_active(di)"
811 395709 PPC64 is missing support for the xvnegsp instruction
812 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
814 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
815 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
816 396839 s390x: Trap instructions not implemented
817 396887 arch_prctl should return EINVAL on unknown option
818 == 397286 crash before launching binary (Unsupported arch_prctl option)
819 == 397393 valgrind: the 'impossible' happened: (Archlinux)
820 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
821 396906 compile tests failure on mips32-linux: broken inline asm in tests on
823 397012 glibc ld.so uses arch_prctl on i386
824 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
825 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
826 397424 glibc 2.27 and gdb_server tests
827 398028 Assertion `cfsi_fits` failing in simple C program
828 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
830 n-i-bz Fix missing workq_ops operations (macOS)
831 n-i-bz fix bug in strspn replacement
832 n-i-bz Add support for the Linux BLKFLSBUF ioctl
833 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
834 n-i-bz Fix possible stack trashing by semctl syscall wrapping
835 n-i-bz Add support for the Linux membarrier() system call
836 n-i-bz x86 front end: recognise and handle UD2 correctly
837 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
838 correctly aligned before entering the handler.
840 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
841 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
842 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
846 Release 3.13.0 (15 June 2017)
847 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
849 3.13.0 is a feature release with many improvements and the usual collection of
852 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
853 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
854 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
855 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
857 * ==================== CORE CHANGES ===================
859 * The translation cache size has been increased to keep up with the demands of
860 large applications. The maximum number of sectors has increased from 24 to
861 48. The default number of sectors has increased from 16 to 32 on all
862 targets except Android, where the increase is from 6 to 12.
864 * The amount of memory that Valgrind can use has been increased from 64GB to
865 128GB. In particular this means your application can allocate up to about
866 60GB when running on Memcheck.
868 * Valgrind's default load address has been changed from 0x3800'0000 to
869 0x5800'0000, so as to make it possible to load larger executables. This
870 should make it possible to load executables of size at least 1200MB.
872 * A massive spaceleak caused by reading compressed debuginfo files has been
873 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
876 * The C++ demangler has been updated.
878 * Support for demangling Rust symbols has been added.
880 * A new representation of stack traces, the "XTree", has been added. An XTree
881 is a tree of stacktraces with data associated with the stacktraces. This is
882 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
883 consumption of your program. Reporting is controlled by the new options
884 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
886 A report can also be produced on demand using the gdbserver monitor command
887 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
888 format' and 'massif format. The existing visualisers for these formats (e.g.
889 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
890 analyse these reports.
892 Memcheck can also produce XTree leak reports using the Callgrind file
893 format. For more details, see the user manual.
895 * ================== PLATFORM CHANGES =================
897 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
899 * amd64: fixes for JIT failure problems on long AVX2 code blocks
901 * amd64 and x86: support for CET prefixes has been added
903 * arm32: a few missing ARMv8 instructions have been implemented
905 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
906 Store-Conditional instructions has been added. This is to deal with
907 processor implementations that implement the LL/SC specifications strictly
908 and as a result cause Valgrind to hang in certain situations. The
909 alternative implementation is automatically enabled at startup, as required.
910 You can use the option --sim-hints=fallback-llsc to force-enable it if you
913 * Support for OSX 10.12 has been improved.
915 * On Linux, clone handling has been improved to honour CLONE_VFORK that
916 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
917 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
918 CLONE_VM exact semantics will (still) not work.
920 * The TileGX/Linux port has been removed because it appears to be both unused
923 * ==================== TOOL CHANGES ====================
927 - Memcheck should give fewer false positives when running optimised
928 Clang/LLVM generated code.
930 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
932 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
933 to produce the end of execution leak report in a xtree callgrind format
936 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
937 the leak report in an xtree file.
941 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
943 - For some workloads (typically, for big applications), Massif memory
944 consumption and CPU consumption has decreased significantly.
948 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
950 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
951 for Ada gnat compiled applications.
953 * ==================== OTHER CHANGES ====================
955 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
956 append the inner guest stacktrace to the inner host stacktrace. This helps
957 to investigate the errors reported by the outer, when they are caused by the
958 inner guest program (such as an inner regtest). See README_DEVELOPERS for
961 * To allow fast detection of callgrind files by desktop environments and file
962 managers, the format was extended to have an optional first line that
963 uniquely identifies the format ("# callgrind format"). Callgrind creates
964 this line now, as does the new xtree functionality.
966 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
967 have a new %n format letter that is replaced by a sequence number.
969 * "--version -v" now shows the SVN revision numbers from which Valgrind was
972 * ==================== FIXED BUGS ====================
974 The following bugs have been fixed or resolved. Note that "n-i-bz"
975 stands for "not in bugzilla" -- that is, a bug that was reported to us
976 but never got a bugzilla entry. We encourage you to file bugs in
977 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
978 than mailing the developers (or mailing lists) directly -- bugs that
979 are not entered into bugzilla tend to get forgotten about or ignored.
981 To see details of a given bug, visit
982 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
983 where XXXXXX is the bug number as listed below.
985 162848 --log-file output isn't split when a program forks
986 340777 Illegal instruction on mips (ar71xx)
987 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
988 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
989 to a different stack.
990 344139 x86 stack-seg overrides, needed by the Wine people
991 344524 store conditional of guest applications always fail - observed on
993 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
994 352395 Please provide SVN revision info in --version -v
995 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
996 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
997 INVALID_POSIX_THREADID' failed
998 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
999 barrier implementation
1000 358697 valgrind.h: Some code remains even when defining NVALGRIND
1001 359202 Add musl libc configure/compile
1002 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
1003 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
1004 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
1005 362223 assertion failed when .valgrindrc is a directory instead of a file
1006 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
1007 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
1008 368507 can't malloc chunks larger than about 34GB
1009 368529 Android arm target link error, missing atexit and pthread_atfork
1010 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
1011 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
1012 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
1013 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
1014 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
1015 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
1016 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
1017 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
1018 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
1019 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
1020 370028 Reduce the number of compiler warnings on MIPS platforms
1021 370635 arm64 missing syscall getcpu
1022 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
1023 371227 Clean AArch64 syscall table
1024 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
1025 371471 Valgrind complains about non legit memory leaks on placement new (C++)
1026 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
1027 371503 disInstr(arm64): unhandled instruction 0xF89F0000
1028 371869 support '%' in symbol Z-encoding
1029 371916 execution tree xtree concept
1030 372120 c++ demangler demangles symbols which are not c++
1031 372185 Support of valgrind on ARMv8 with 32 bit executable
1032 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1033 372195 Power PC, xxsel instruction is not always recognized.
1034 372504 Hanging on exit_group
1035 372600 process loops forever when fatal signals are arriving quickly
1036 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1037 373046 Stacks registered by core are never deregistered
1038 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1039 373086 Implement additional Xen hypercalls
1040 373192 Calling posix_spawn in glibc 2.24 completely broken
1041 373488 Support for fanotify API on ARM64 architecture
1042 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1043 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1044 373938 const IRExpr arguments for matchIRExpr()
1045 374719 some spelling fixes
1046 374963 increase valgrind's load address to prevent mmap failure
1047 375514 valgrind_get_tls_addr() does not work in case of static TLS
1048 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1049 for ML_(find_rx_mapping)()
1050 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1051 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1052 == 377159 "vex: the `impossible' happened" still present
1053 == 375150 Assertion 'tres.status == VexTransOK' failed
1054 == 378068 valgrind crashes on AVX2 function in FFmpeg
1055 376142 Segfaults on MIPS Cavium Octeon boards
1056 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1057 376455 Solaris: unhandled syscall lgrpsys(180)
1058 376518 Solaris: unhandled fast trap getlgrp(6)
1059 376611 ppc64 and arm64 don't know about prlimit64 syscall
1060 376729 PPC64, remove R2 from the clobber list
1062 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1063 to be wrongly marked as addressable
1064 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1065 PIE enabled by default
1066 377376 memcheck/tests/linux/getregset fails with glibc2.24
1067 377427 PPC64, lxv instruction failing on odd destination register
1068 377478 PPC64: ISA 3.0 setup fixes
1069 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1070 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1071 and 2 args for FUTEX_TRYLOCK_PI
1072 377717 Fix massive space leak when reading compressed debuginfo sections
1073 377891 Update Xen 4.6 domctl wrappers
1074 377930 fcntl syscall wrapper is missing flock structure check
1075 378524 libvexmultiarch_test regression on s390x and ppc64
1076 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1077 378673 Update libiberty demangler
1078 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1079 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1080 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1081 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1082 (task_register_dyld_image_infos)
1083 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1084 (task_register_dyld_shared_cache_image_info)
1085 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1086 379473 MIPS: add support for rdhwr cycle counter register
1087 379504 remove TileGX/Linux port
1088 379525 Support more x86 nop opcodes
1089 379838 disAMode(x86): not an addr!
1090 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1091 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1092 379895 clock_gettime does not execute POST syscall wrapper
1093 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1094 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1095 380200 xtree generated callgrind files refer to files without directory name
1096 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1097 380397 s390x: __GI_strcspn() replacement needed
1098 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1100 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1101 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1102 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1106 Release 3.12.0 (20 October 2016)
1107 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1109 3.12.0 is a feature release with many improvements and the usual
1110 collection of bug fixes.
1112 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1113 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1114 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1115 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1116 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1117 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1119 * ================== PLATFORM CHANGES =================
1121 * POWER: Support for ISA 3.0 has been added
1123 * mips: support for O32 FPXX ABI has been added.
1124 * mips: improved recognition of different processors
1125 * mips: determination of page size now done at run time
1127 * amd64: Partial support for AMD FMA4 instructions.
1129 * arm, arm64: Support for v8 crypto and CRC instructions.
1131 * Improvements and robustification of the Solaris port.
1133 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1135 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1136 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1137 where possible. Valgrind's support for 32-bit x86 has stagnated in
1138 recent years and has fallen far behind that for 64-bit x86
1139 instructions. By contrast 64-bit x86 is well supported, up to and
1142 * ==================== TOOL CHANGES ====================
1146 - Added meta mempool support for describing a custom allocator which:
1147 - Auto-frees all chunks assuming that destroying a pool destroys all
1149 - Uses itself to allocate other memory blocks
1151 - New flag --ignore-range-below-sp to ignore memory accesses below
1152 the stack pointer, if you really have to. The related flag
1153 --workaround-gcc296-bugs=yes is now deprecated. Use
1154 --ignore-range-below-sp=1024-1 as a replacement.
1158 - Improved thread startup time significantly on non-Linux platforms.
1162 - Added collection of the metric "tot-blocks-allocd"
1164 * ==================== OTHER CHANGES ====================
1166 * Replacement/wrapping of malloc/new related functions is now done not just
1167 for system libraries by default, but for any globally defined malloc/new
1168 related function (both in shared libraries and statically linked alternative
1169 malloc implementations). The dynamic (runtime) linker is excluded, though.
1170 To only intercept malloc/new related functions in
1171 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1172 "nouserintercepts" can be any non-existing library name).
1173 This new functionality is not implemented for MacOS X.
1175 * The maximum number of callers in a suppression entry is now equal to
1176 the maximum size for --num-callers (500).
1177 Note that --gen-suppressions=yes|all similarly generates suppressions
1178 containing up to --num-callers frames.
1180 * New and modified GDB server monitor features:
1182 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1183 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1186 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1187 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1190 * Valgrind is able to read compressed debuginfo sections in two formats:
1191 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1192 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1194 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1195 for the most common use case (x86_64-linux, Memcheck) has been
1198 * Improved performance for programs that do a lot of discarding of
1199 instruction address ranges of 8KB or less.
1201 * The C++ symbol demangler has been updated.
1203 * More robustness against invalid syscall parameters on Linux.
1205 * ==================== FIXED BUGS ====================
1207 The following bugs have been fixed or resolved. Note that "n-i-bz"
1208 stands for "not in bugzilla" -- that is, a bug that was reported to us
1209 but never got a bugzilla entry. We encourage you to file bugs in
1210 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1211 than mailing the developers (or mailing lists) directly -- bugs that
1212 are not entered into bugzilla tend to get forgotten about or ignored.
1214 To see details of a given bug, visit
1215 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1216 where XXXXXX is the bug number as listed below.
1218 191069 Exiting due to signal not reported in XML output
1219 199468 Suppressions: stack size limited to 25
1220 while --num-callers allows more frames
1221 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1222 278744 cvtps2pd with redundant RexW
1223 303877 valgrind doesn't support compressed debuginfo sections.
1224 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1225 348345 Assertion fails for negative lineno
1226 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1227 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1228 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1229 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1230 352197 mips: mmap2() not wrapped correctly for page size > 4096
1231 353083 arm64 doesn't implement various xattr system calls
1232 353084 arm64 doesn't support sigpending system call
1233 353137 www: update info for Supported Platforms
1234 353138 www: update "The Valgrind Developers" page
1235 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1238 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1239 353398 WARNING: unhandled amd64-solaris syscall: 207
1240 353660 XML in auxwhat tag not escaping reserved symbols properly
1241 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1242 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1243 353802 ELF debug info reader confused with multiple .rodata sections
1244 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1245 353917 unhandled amd64-solaris syscall fchdir(120)
1246 353920 unhandled amd64-solaris syscall: 170
1247 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1248 354392 unhandled amd64-solaris syscall: 171
1249 354797 Vbit test does not include Iops for Power 8 instruction support
1250 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1254 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1255 355188 valgrind should intercept all malloc related global functions
1256 355454 do not intercept malloc related symbols from the runtime linker
1257 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1258 356044 Dwarf line info reader misinterprets is_stmt register
1259 356112 mips: replace addi with addiu
1260 356393 valgrind (vex) crashes because isZeroU happened
1263 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1264 356678 arm64-linux: unhandled syscall 232 (mincore)
1265 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1266 356823 Unsupported ARM instruction: stlex
1267 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1268 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1269 357673 crash if I try to run valgrind with a binary link with libcurl
1270 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1271 357871 pthread_spin_destroy not properly wrapped
1272 357887 Calls to VG_(fclose) do not close the file descriptor
1273 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1274 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1275 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1276 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1277 359181 Buffer Overflow during Demangling
1278 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1279 359289 s390x: popcnt (B9E1) not implemented
1280 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1281 359503 Add missing syscalls for aarch64 (arm64)
1282 359645 "You need libc6-dbg" help message could be more helpful
1283 359703 s390: wire up separate socketcalls system calls
1284 359724 getsockname might crash - deref_UInt should call safe_to_deref
1285 359733 amd64 implement ld.so strchr/index override like x86
1286 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1287 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1289 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1290 359871 Incorrect mask handling in ppoll
1291 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1292 360008 Contents of Power vr registers contents is not printed correctly when
1293 the --vgdb-shadow-registers=yes option is used
1294 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1295 non-zero shadow bits
1296 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1297 360425 arm64 unsupported instruction ldpsw
1299 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1300 360571 Error about the Android Runtime reading below the stack pointer on ARM
1301 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1302 360749 kludge for multiple .rodata sections on Solaris no longer needed
1303 360752 raise the number of reserved fds in m_main.c from 10 to 12
1304 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1305 361226 s390x: risbgn (EC59) not implemented
1306 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1307 361354 ppc64[le]: wire up separate socketcalls system calls
1308 361615 Inconsistent termination for multithreaded process terminated by signal
1309 361926 Unhandled Solaris syscall: sysfs(84)
1310 362009 V dumps core on unimplemented functionality before threads are created
1311 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1312 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1313 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1314 362953 Request for an update to the Valgrind Developers page
1315 363680 add renameat2() support
1316 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1317 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1318 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1319 364058 clarify in manual limitations of array overruns detections
1320 364413 pselect sycallwrapper mishandles NULL sigmask
1321 364728 Power PC, missing support for several HW registers in
1322 get_otrack_shadow_offset_wrk()
1323 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1324 365273 Invalid write to stack location reported after signal handler runs
1325 365912 ppc64BE segfault during jm-insns test (RELRO)
1326 366079 FPXX Support for MIPS32 Valgrind
1327 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1328 366344 Multiple unhandled instruction for Aarch64
1329 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1330 367995 Integration of memcheck with custom memory allocator
1331 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1332 368412 False positive result for altivec capability check
1333 368416 Add tc06_two_races_xml.exp output for ppc64
1334 368419 Perf Events ioctls not implemented
1335 368461 mmapunmap test fails on ppc64
1336 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1337 369000 AMD64 fma4 instructions unsupported.
1338 369169 ppc64 fails jm_int_isa_2_07 test
1339 369175 jm_vec_isa_2_07 test crashes on ppc64
1340 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1341 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1342 369359 msghdr_foreachfield can crash when handling bad iovec
1343 369360 Bad sigprocmask old or new sets can crash valgrind
1344 369361 vmsplice syscall wrapper crashes on bad iovec
1345 369362 Bad sigaction arguments crash valgrind
1346 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1347 369402 Bad set/get_thread_area pointer crashes valgrind
1348 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1349 369446 valgrind crashes on unknown fcntl command
1350 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1351 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1352 370265 ISA 3.0 HW cap stuff needs updating
1353 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1354 372195 Power PC, xxsel instruction is not always recognized
1356 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1357 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1358 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1359 n-i-bz Fix ppoll_alarm exclusion on OS X
1360 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1361 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1362 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1363 n-i-bz amd64: memcheck false positive with shr %edx
1364 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1365 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1366 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1367 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1369 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1370 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1371 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1375 Release 3.11.0 (22 September 2015)
1376 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1378 3.11.0 is a feature release with many improvements and the usual
1379 collection of bug fixes.
1381 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1382 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1383 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1384 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1385 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1386 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1388 * ================== PLATFORM CHANGES =================
1390 * Support for Solaris/x86 and Solaris/amd64 has been added.
1392 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1394 * Preliminary support for the Tilera TileGX architecture has been added.
1396 * s390x: It is now required for the host to have the "long displacement"
1397 facility. The oldest supported machine model is z990.
1399 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1400 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1402 * The JIT's register allocator is significantly faster, making the JIT
1403 as a whole somewhat faster, so JIT-intensive activities, for example
1404 program startup, are modestly faster, around 5%.
1406 * There have been changes to the default settings of several command
1407 line flags, as detailed below.
1409 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1410 capable hosts, the simulated CPUID will now indicate AVX2 support.
1412 * ==================== TOOL CHANGES ====================
1416 - The default value for --leak-check-heuristics has been changed from
1417 "none" to "all". This helps to reduce the number of possibly
1418 lost blocks, in particular for C++ applications.
1420 - The default value for --keep-stacktraces has been changed from
1421 "malloc-then-free" to "malloc-and-free". This has a small cost in
1422 memory (one word per malloc-ed block) but allows Memcheck to show the
1423 3 stacktraces of a dangling reference: where the block was allocated,
1424 where it was freed, and where it is acccessed after being freed.
1426 - The default value for --partial-loads-ok has been changed from "no" to
1427 "yes", so as to avoid false positive errors resulting from some kinds
1428 of vectorised loops.
1430 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1431 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1432 than get_vbits when you need to associate byte data value with
1433 their corresponding validity bits.
1435 - The 'block_list' monitor command has been enhanced:
1436 o it can print a range of loss records
1437 o it now accepts an optional argument 'limited <max_blocks>'
1438 to control the number of blocks printed.
1439 o if a block has been found using a heuristic, then
1440 'block_list' now shows the heuristic after the block size.
1441 o the loss records/blocks to print can be limited to the blocks
1442 found via specified heuristics.
1444 - The C helper functions used to instrument loads on
1445 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1446 replaced by handwritten assembly sequences. This gives speedups
1447 in the region of 0% to 7% for those targets only.
1449 - A new command line option, --expensive-definedness-checks=yes|no,
1450 has been added. This is useful for avoiding occasional invalid
1451 uninitialised-value errors in optimised code. Watch out for
1452 runtime degradation, as this can be up to 25%. As always, though,
1453 the slowdown is highly application specific. The default setting
1458 - A new monitor command 'all_snapshots <filename>' dumps all
1459 snapshots taken so far.
1463 - Significant memory reduction and moderate speedups for
1464 --history-level=full for applications accessing a lot of memory
1465 with many different stacktraces.
1467 - The default value for --conflict-cache-size=N has been doubled to
1468 2000000. Users that were not using the default value should
1469 preferably also double the value they give.
1471 The default was changed due to the changes in the "full history"
1472 implementation. Doubling the value gives on average a slightly more
1473 complete history and uses similar memory (or significantly less memory
1474 in the worst case) than the previous implementation.
1476 - The Helgrind monitor command 'info locks' now accepts an optional
1477 argument 'lock_addr', which shows information about the lock at the
1480 - When using --history-level=full, the new Helgrind monitor command
1481 'accesshistory <addr> [<len>]' will show the recorded accesses for
1482 <len> (or 1) bytes at <addr>.
1484 * ==================== OTHER CHANGES ====================
1486 * The default value for the --smc-check option has been changed from
1487 "stack" to "all-non-file" on targets that provide automatic D-I
1488 cache coherence (x86, amd64 and s390x). The result is to provide,
1489 by default, transparent support for JIT generated and self-modifying
1490 code on all targets.
1492 * Mac OS X only: the default value for the --dsymutil option has been
1493 changed from "no" to "yes", since any serious usage on Mac OS X
1494 always required it to be "yes".
1496 * The command line options --db-attach and --db-command have been removed.
1497 They were deprecated in 3.10.0.
1499 * When a process dies due to a signal, Valgrind now shows the signal
1500 and the stacktrace at default verbosity (i.e. verbosity 1).
1502 * The address description logic used by Memcheck and Helgrind now
1503 describes addresses in anonymous segments, file mmap-ed segments,
1504 shared memory segments and the brk data segment.
1506 * The new option --error-markers=<begin>,<end> can be used to mark the
1507 begin/end of errors in textual output mode, to facilitate
1508 searching/extracting errors in output files that mix valgrind errors
1509 with program output.
1511 * The new option --max-threads=<number> can be used to change the number
1512 of threads valgrind can handle. The default is 500 threads which
1513 should be more than enough for most applications.
1515 * The new option --valgrind-stacksize=<number> can be used to change the
1516 size of the private thread stacks used by Valgrind. This is useful
1517 for reducing memory use or increasing the stack size if Valgrind
1518 segfaults due to stack overflow.
1520 * The new option --avg-transtab-entry-size=<number> can be used to specify
1521 the expected instrumented block size, either to reduce memory use or
1522 to avoid excessive retranslation.
1524 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1526 * New and modified GDB server monitor features:
1528 - When a signal is reported in GDB, you can now use the GDB convenience
1529 variable $_siginfo to examine detailed signal information.
1531 - Valgrind's gdbserver now allows the user to change the signal
1532 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1533 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1534 continue without passing the signal to the process.
1536 - With GDB >= 7.10, the command 'target remote'
1537 will automatically load the executable file of the process running
1538 under Valgrind. This means you do not need to specify the executable
1539 file yourself, GDB will discover it itself. See GDB documentation about
1540 'qXfer:exec-file:read' packet for more info.
1542 * ==================== FIXED BUGS ====================
1544 The following bugs have been fixed or resolved. Note that "n-i-bz"
1545 stands for "not in bugzilla" -- that is, a bug that was reported to us
1546 but never got a bugzilla entry. We encourage you to file bugs in
1547 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1548 than mailing the developers (or mailing lists) directly -- bugs that
1549 are not entered into bugzilla tend to get forgotten about or ignored.
1551 To see details of a given bug, visit
1552 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1553 where XXXXXX is the bug number as listed below.
1555 116002 VG_(printf): Problems with justification of strings and integers
1556 155125 avoid cutting away file:lineno after long function name
1557 197259 Unsupported arch_prtctl PR_SET_GS option
1558 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
1559 201216 Fix Valgrind does not support pthread_sigmask() on OS X
1560 201435 Fix Darwin: -v does not show kernel version
1561 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
1562 211256 Fixed an outdated comment regarding the default platform.
1563 211529 Incomplete call stacks for code compiled by newer versions of MSVC
1564 211926 Avoid compilation warnings in valgrind.h with -pedantic
1565 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
1567 226609 Crediting upstream authors in man page
1568 231257 Valgrind omits path when executing script from shebang line
1569 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
1570 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
1571 269360 s390x: Fix addressing mode selection for compare-and-swap
1572 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
1574 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
1575 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
1576 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
1577 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
1578 330147 libmpiwrap PMPI_Get_count returns undefined value
1579 333051 mmap of huge pages fails due to incorrect alignment
1581 334802 valgrind does not always explain why a given option is bad
1582 335618 mov.w rN, pc/sp (ARM32)
1583 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
1588 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
1589 338602 AVX2 bit in CPUID missing
1590 338606 Strange message for scripts with invalid interpreter
1591 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
1592 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
1593 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
1595 339156 gdbsrv not called for fatal signal
1596 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
1597 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
1598 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
1599 339442 Fix testsuite build failure on OS X 10.9
1600 339542 Enable compilation with Intel's ICC compiler
1601 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
1602 339688 Mac-specific ASM does not support .version directive (cpuid,
1603 tronical and pushfpopf tests)
1604 339745 Valgrind crash when check Marmalade app (partial fix)
1605 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
1606 339778 Linux/TileGx platform support for Valgrind
1607 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
1608 339789 Fix none/tests/execve test on Mac OS X 10.9
1609 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
1610 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
1611 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
1612 340392 Allow user to select more accurate definedness checking in memcheck
1613 to avoid invalid complaints on optimised code
1614 340430 Fix some grammatical weirdness in the manual.
1615 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
1616 341419 Signal handler ucontext_t not filled out correctly on OS X
1617 341539 VG_(describe_addr) should not describe address as belonging to client
1618 segment if it is past the heap end
1619 341613 Enable building of manythreads and thread-exits tests on Mac OS X
1620 341615 Fix none/tests/darwin/access_extended test on Mac OS X
1621 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
1622 341789 aarch64: shmat fails with valgrind on ARMv8
1623 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
1624 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
1625 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
1626 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
1627 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
1628 342221 socket connect false positive uninit memory for unknown af family
1629 342353 Allow dumping full massif output while valgrind is still running
1630 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
1634 342603 Add I2C_SMBUS ioctl support
1635 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
1636 342683 Mark memory past the initial brk limit as unaddressable
1637 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
1638 342795 Internal glibc __GI_mempcpy call should be intercepted
1639 342841 s390x: Support instructions fiebr(a) and fidbr(a)
1640 343012 Unhandled syscall 319 (memfd_create)
1641 343069 Patch updating v4l2 API support
1642 343173 helgrind crash during stack unwind
1643 343219 fix GET_STARTREGS for arm
1644 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
1645 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
1646 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
1647 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
1648 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
1649 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
1650 343597 ppc64le: incorrect use of offseof macro
1651 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
1652 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
1653 343732 Unhandled syscall 144 (setgid) on aarch64
1654 343733 Unhandled syscall 187 (msgctl and related) on aarch64
1655 343802 s390x: False positive "conditional jump or move depends on [..]
1656 343902 --vgdb=yes doesn't break when --xml=yes is used
1657 343967 Don't warn about setuid/setgid/setcap executable for directories
1658 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
1659 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
1660 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
1661 344054 www - update info for Solaris/illumos
1662 344416 'make regtest' does not work cleanly on OS X
1663 344235 Remove duplicate include of pub_core_aspacemgr.h
1664 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
1665 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
1666 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
1667 344314 callgrind_annotate ... warnings about commands containing newlines
1668 344318 socketcall should wrap recvmmsg and sendmmsg
1669 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
1670 344416 Fix 'make regtest' does not work cleanly on OS X
1671 344499 Fix compilation for Linux kernel >= 4.0.0
1672 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
1673 unix:349 (__pthread_fchdir)
1674 344559 Garbage collection of unused segment names in address space manager
1675 344560 Fix stack traces missing penultimate frame on OS X
1676 344621 Fix memcheck/tests/err_disable4 test on OS X
1677 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
1678 344702 Fix missing libobjc suppressions on OS X 10.10
1680 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
1681 344939 Fix memcheck/tests/xml1 on OS X 10.10
1682 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
1683 345079 Fix build problems in VEX/useful/test_main.c
1684 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
1685 345177 arm64: prfm (reg) not implemented
1686 345215 Performance improvements for the register allocator
1687 345248 add support for Solaris OS in valgrind
1688 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
1689 345394 Fix memcheck/tests/strchr on OS X
1690 345637 Fix memcheck/tests/sendmsg on OS X
1691 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
1692 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
1693 345887 Fix an assertion in the address space manager
1694 345928 amd64: callstack only contains current function for small stacks
1695 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1696 345987 MIPS64: Implement cavium LHX instruction
1697 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1698 346185 Fix typo saving altivec register v24
1699 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1700 and LibVEX_GuestPPC64_get_CR()
1701 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1702 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1703 346307 fuse filesystem syscall deadlocks
1704 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1705 346411 MIPS: SysRes::_valEx handling is incorrect
1706 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1707 346474 PPC64 Power 8, spr TEXASRU register not supported
1708 346487 Compiler generates "note" about a future ABI change for PPC64
1709 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1710 and causing spurious "invalid read of size 8" warnings
1711 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1712 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1713 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1714 347322 Power PC regression test cleanup
1715 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1717 347389 unhandled syscall: 373 (Linux ARM syncfs)
1718 347686 Patch set to cleanup PPC64 regtests
1719 347978 Remove bash dependencies where not needed
1720 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1721 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1723 348102 Patch updating v4l2 API support
1724 348247 amd64 front end: jno jumps wrongly when overflow is not set
1725 348269 Improve mmap MAP_HUGETLB support.
1726 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1727 348345 Assertion fails for negative lineno
1728 348377 Unsupported ARM instruction: yield
1729 348565 Fix detection of command line option availability for clang
1730 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1731 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1732 348748 Fix redundant condition
1733 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1734 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1735 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1736 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1737 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1738 349626 Implemented additional Xen hypercalls
1739 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1740 349790 Clean up of the hardware capability checking utilities.
1741 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1742 349874 Fix typos in source code
1743 349879 memcheck: add handwritten assembly for helperc_LOADV*
1744 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1745 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1746 350202 Add limited param to 'monitor block_list'
1747 350290 s390x: Support instructions fixbr(a)
1748 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1749 350809 Fix none/tests/async-sigs for Solaris
1750 350811 Remove reference to --db-attach which has been removed.
1751 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1752 350854 hard-to-understand code in VG_(load_ELF)()
1753 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1754 351386 Solaris: Cannot run ld.so.1 under Valgrind
1755 351474 Fix VG_(iseqsigset) as obvious
1756 351531 Typo in /include/vki/vki-xen-physdev.h header guard
1757 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
1758 351858 ldsoexec support on Solaris
1759 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
1760 352130 helgrind reports false races for printfs using mempcpy on FILE* state
1761 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
1762 352320 arm64 crash on none/tests/nestedfs
1763 352765 Vbit test fails on Power 6
1764 352768 The mbar instruction is missing from the Power PC support
1765 352769 Power PC program priority register (PPR) is not supported
1766 n-i-bz Provide implementations of certain compiler builtins to support
1767 compilers that may not provide those
1768 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
1769 n-i-bz Fix compilation on distros with glibc < 2.5
1770 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
1771 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
1772 n-i-bz mremap did not work properly on shared memory
1773 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
1774 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
1776 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
1777 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
1778 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
1782 Release 3.10.1 (25 November 2014)
1783 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1784 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
1785 and backports fixes for all reported missing AArch64 ARMv8 instructions
1786 and syscalls from the trunk. If you package or deliver 3.10.0 for others
1787 to use, you might want to consider upgrading to 3.10.1 instead.
1789 The following bugs have been fixed or resolved. Note that "n-i-bz"
1790 stands for "not in bugzilla" -- that is, a bug that was reported to us
1791 but never got a bugzilla entry. We encourage you to file bugs in
1792 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1793 than mailing the developers (or mailing lists) directly -- bugs that
1794 are not entered into bugzilla tend to get forgotten about or ignored.
1796 To see details of a given bug, visit
1797 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1798 where XXXXXX is the bug number as listed below.
1800 335440 arm64: ld1 (single structure) is not implemented
1801 335713 arm64: unhanded instruction: prfm (immediate)
1802 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
1803 339182 ppc64: AvSplat ought to load destination vector register with [..]
1804 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
1805 339433 ppc64 lxvw4x instruction uses four 32-byte loads
1806 339645 Use correct tag names in sys_getdents/64 wrappers
1807 339706 Fix false positive for ioctl(TIOCSIG) on linux
1808 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
1809 339853 arm64 times syscall unknown
1810 339855 arm64 unhandled getsid/setsid syscalls
1811 339858 arm64 dmb sy not implemented
1812 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
1813 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
1814 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
1816 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
1817 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
1818 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
1819 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
1820 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
1821 340509 arm64: unhandled instruction fcvtas
1822 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
1823 340632 arm64: unhandled instruction fcvtas
1824 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
1825 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
1826 340788 warning: unhandled syscall: 318 (getrandom)
1827 340807 disInstr(arm): unhandled instruction: 0xEE989B20
1828 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
1829 340922 arm64: unhandled getgroups/setgroups syscalls
1830 350251 Fix typo in VEX utility program (test_main.c).
1831 350407 arm64: unhandled instruction ucvtf (vector, integer)
1832 350809 none/tests/async-sigs breaks when run under cron on Solaris
1833 350811 update README.solaris after r15445
1834 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
1835 350854 strange code in VG_(load_ELF)()
1836 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1837 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
1838 n-i-bz Add missing ]] to terminate CDATA.
1839 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
1840 n-i-bz Enable sys_fadvise64_64 on arm32.
1841 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
1842 n-i-bz Add test cases for all known arm64 load/store instructions.
1843 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
1844 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
1845 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
1846 n-i-bz arm64: Implement "fcvtpu w, s".
1847 n-i-bz arm64: implement ADDP and various others
1848 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
1849 n-i-bz arm64: enable FCVT{A,N}S X,S.
1851 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
1855 Release 3.10.0 (10 September 2014)
1856 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1858 3.10.0 is a feature release with many improvements and the usual
1859 collection of bug fixes.
1861 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1862 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1863 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
1864 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
1865 significantly improved relative to the 3.9.0 release.
1867 * ================== PLATFORM CHANGES =================
1869 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
1870 is mostly complete, and is usable, but some SIMD instructions are as
1873 * Support for little-endian variant of the 64-bit POWER architecture.
1875 * Support for Android on MIPS32.
1877 * Support for 64bit FPU on MIPS32 platforms.
1879 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
1881 * Configuration for and running on Android targets has changed.
1882 See README.android in the source tree for details.
1884 * ================== DEPRECATED FEATURES =================
1886 * --db-attach is now deprecated and will be removed in the next
1887 valgrind feature release. The built-in GDB server capabilities are
1888 superior and should be used instead. Learn more here:
1889 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
1891 * ==================== TOOL CHANGES ====================
1895 - Client code can now selectively disable and re-enable reporting of
1896 invalid address errors in specific ranges using the new client
1897 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
1898 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
1900 - Leak checker: there is a new leak check heuristic called
1901 "length64". This is used to detect interior pointers pointing 8
1902 bytes inside a block, on the assumption that the first 8 bytes
1903 holds the value "block size - 8". This is used by
1904 sqlite3MemMalloc, for example.
1906 - Checking of system call parameters: if a syscall parameter
1907 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
1908 several fields not initialised, an error is now reported for each
1909 field. Previously, an error was reported only for the first
1910 uninitialised field.
1912 - Mismatched alloc/free checking: a new flag
1913 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
1914 such checks if necessary.
1918 - Improvements to error messages:
1920 o Race condition error message involving heap allocated blocks also
1921 show the thread number that allocated the raced-on block.
1923 o All locks referenced by an error message are now announced.
1924 Previously, some error messages only showed the lock addresses.
1926 o The message indicating where a lock was first observed now also
1927 describes the address/location of the lock.
1929 - Helgrind now understands the Ada task termination rules and
1930 creates a happens-before relationship between a terminated task
1931 and its master. This avoids some false positives and avoids a big
1932 memory leak when a lot of Ada tasks are created and terminated.
1933 The interceptions are only activated with forthcoming releases of
1934 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
1936 - A new GDB server monitor command "info locks" giving the list of
1937 locks, their location, and their status.
1941 - callgrind_control now supports the --vgdb-prefix argument,
1942 which is needed if valgrind was started with this same argument.
1944 * ==================== OTHER CHANGES ====================
1946 * Unwinding through inlined function calls. Stack unwinding can now
1947 make use of Dwarf3 inlined-unwind information if it is available.
1948 The practical effect is that inlined calls become visible in stack
1949 traces. The suppression matching machinery has been adjusted
1950 accordingly. This is controlled by the new option
1951 --read-inline-info=yes|no. Currently this is enabled by default
1952 only on Linux and Android targets and only for the tools Memcheck,
1955 * Valgrind can now read EXIDX unwind information on 32-bit ARM
1956 targets. If an object contains both CFI and EXIDX unwind
1957 information, Valgrind will prefer the CFI over the EXIDX. This
1958 facilitates unwinding through system libraries on arm-android
1961 * Address description logic has been improved and is now common
1962 between Memcheck and Helgrind, resulting in better address
1963 descriptions for some kinds of error messages.
1965 * Error messages about dubious arguments (eg, to malloc or calloc) are
1966 output like other errors. This means that they can be suppressed
1967 and they have a stack trace.
1969 * The C++ demangler has been updated for better C++11 support.
1971 * New and modified GDB server monitor features:
1973 - Thread local variables/storage (__thread) can now be displayed.
1975 - The GDB server monitor command "v.info location <address>"
1976 displays information about an address. The information produced
1977 depends on the tool and on the options given to valgrind.
1978 Possibly, the following are described: global variables, local
1979 (stack) variables, allocated or freed blocks, ...
1981 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
1982 ask the GDB server to stop at the start of program execution, at
1983 the end of the program execution and on Valgrind internal errors.
1985 - A new monitor command "v.info stats" shows various Valgrind core
1986 and tool statistics.
1988 - A new monitor command "v.set hostvisibility" allows the GDB server
1989 to provide access to Valgrind internal host status/memory.
1991 * A new option "--aspace-minaddr=<address>" can in some situations
1992 allow the use of more memory by decreasing the address above which
1993 Valgrind maps memory. It can also be used to solve address
1994 conflicts with system libraries by increasing the default value.
1995 See user manual for details.
1997 * The amount of memory used by Valgrind to store debug info (unwind
1998 info, line number information and symbol data) has been
1999 significantly reduced, even though Valgrind now reads more
2000 information in order to support unwinding of inlined function calls.
2002 * Dwarf3 handling with --read-var-info=yes has been improved:
2004 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
2006 - Code compiled with
2007 -ffunction-sections -fdata-sections -Wl,--gc-sections
2008 no longer causes assertion failures.
2010 * Improved checking for the --sim-hints= and --kernel-variant=
2011 options. Unknown strings are now detected and reported to the user
2014 * The semantics of stack start/end boundaries in the valgrind.h
2015 VALGRIND_STACK_REGISTER client request has been clarified and
2016 documented. The convention is that start and end are respectively
2017 the lowest and highest addressable bytes of the stack.
2019 * ==================== FIXED BUGS ====================
2021 The following bugs have been fixed or resolved. Note that "n-i-bz"
2022 stands for "not in bugzilla" -- that is, a bug that was reported to us
2023 but never got a bugzilla entry. We encourage you to file bugs in
2024 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2025 than mailing the developers (or mailing lists) directly -- bugs that
2026 are not entered into bugzilla tend to get forgotten about or ignored.
2028 To see details of a given bug, visit
2029 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2030 where XXXXXX is the bug number as listed below.
2032 175819 Support for ipv6 socket reporting with --track-fds
2033 232510 make distcheck fails
2034 249435 Analyzing wine programs with callgrind triggers a crash
2035 278972 support for inlined function calls in stacktraces and suppression
2037 291310 FXSAVE instruction marks memory as undefined on amd64
2038 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2039 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2040 315199 vgcore file for threaded app does not show which thread crashed
2041 315952 tun/tap ioctls are not supported
2042 323178 Unhandled instruction: PLDW register (ARM)
2043 323179 Unhandled instruction: PLDW immediate (ARM)
2044 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2045 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2046 325124 [MIPSEL] Compilation error
2047 325477 Phase 4 support for IBM Power ISA 2.07
2048 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2049 325628 Phase 5 support for IBM Power ISA 2.07
2050 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2051 325751 Missing the two privileged Power PC Transactional Memory Instructions
2052 325816 Phase 6 support for IBM Power ISA 2.07
2053 325856 Make SGCheck fail gracefully on unsupported platforms
2054 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2055 326436 DRD: False positive in libstdc++ std::list::push_back
2056 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2057 326462 Refactor vgdb to isolate invoker stuff into separate module
2058 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2059 326623 DRD: false positive conflict report in a field assignment
2060 326724 Valgrind does not compile on OSX 1.9 Mavericks
2061 326816 Intercept for __strncpy_sse2_unaligned missing?
2062 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2063 326983 Clear direction flag after tests on amd64.
2064 327212 Do not prepend the current directory to absolute path names.
2065 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2066 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2067 327284 s390x: Fix translation of the risbg instruction
2068 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2069 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2070 327916 DW_TAG_typedef may have no name
2071 327943 s390x: add a redirection for the 'index' function
2072 328100 XABORT not implemented
2073 328205 Implement additional Xen hypercalls
2074 328454 add support Backtraces with ARM unwind tables (EXIDX)
2075 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2076 328711 valgrind.1 manpage "memcheck options" section is badly generated
2077 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2078 329612 Incorrect handling of AT_BASE for image execution
2079 329694 clang warns about using uninitialized variable
2080 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2081 330228 mmap must align to VKI_SHMLBA on mips32
2082 330257 LLVM does not support `-mno-dynamic-no-pic` option
2083 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2084 330459 --track-fds=yes doesn't track eventfds
2085 330469 Add clock_adjtime syscall support
2086 330594 Missing sysalls on PowerPC / uClibc
2087 330622 Add test to regression suite for POWER instruction: dcbzl
2088 330939 Support for AMD's syscall instruction on x86
2090 330941 Typo in PRE(poll) syscall wrapper
2091 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2092 331254 Fix expected output for memcheck/tests/dw4
2093 331255 Fix race condition in test none/tests/coolo_sigaction
2094 331257 Fix type of jump buffer in test none/tests/faultstatus
2095 331305 configure uses bash specific syntax
2096 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2097 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2098 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2099 331829 Unexpected ioctl opcode sign extension
2100 331830 ppc64: WARNING: unhandled syscall: 96/97
2101 331839 drd/tests/sem_open specifies invalid semaphore name
2102 331847 outcome of drd/tests/thread_name is nondeterministic
2103 332037 Valgrind cannot handle Thumb "add pc, reg"
2104 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2105 consistency checks enabled
2106 332263 intercepts for pthread_rwlock_timedrdlock and
2107 pthread_rwlock_timedwrlock are incorrect
2108 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2110 332276 Implement additional Xen hypercalls
2111 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2112 332765 Fix ms_print to create temporary files in a proper directory
2113 333072 drd: Add semaphore annotations
2114 333145 Tests for missaligned PC+#imm access for arm
2115 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2116 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2117 333248 WARNING: unhandled syscall: unix:443
2118 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2119 333501 cachegrind: assertion: Cache set count is not a power of two.
2122 333666 Recognize MPX instructions and bnd prefix.
2123 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2124 333817 Valgrind reports the memory areas written to by the SG_IO
2126 334049 lzcnt fails silently (x86_32)
2127 334384 Valgrind does not have support Little Endian support for
2129 334585 recvmmsg unhandled (+patch) (arm)
2130 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2131 334727 Build fails with -Werror=format-security
2132 334788 clarify doc about --log-file initial program directory
2133 334834 PPC64 Little Endian support, patch 2
2134 334836 PPC64 Little Endian support, patch 3 testcase fixes
2135 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2136 335034 Unhandled ioctl: HCIGETDEVLIST
2137 335155 vgdb, fix error print statement.
2138 335262 arm64: movi 8bit version is not supported
2139 335263 arm64: dmb instruction is not implemented
2140 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2141 335496 arm64: sbc/abc instructions are not implemented
2142 335554 arm64: unhandled instruction: abs
2143 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2144 335735 arm64: unhandled instruction: cnt
2145 335736 arm64: unhandled instruction: uaddlv
2146 335848 arm64: unhandled instruction: {s,u}cvtf
2147 335902 arm64: unhandled instruction: sli
2148 335903 arm64: unhandled instruction: umull (vector)
2149 336055 arm64: unhandled instruction: mov (element)
2150 336062 arm64: unhandled instruction: shrn{,2}
2151 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2152 336189 arm64: unhandled Instruction: mvn
2153 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2154 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2155 336772 Make moans about unknown ioctls more informative
2156 336957 Add a section about the Solaris/illumos port on the webpage
2157 337094 ifunc wrapper is broken on ppc64
2158 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2159 337528 leak check heuristic for block prefixed by length as 64bit number
2160 337740 Implement additional Xen hypercalls
2161 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2162 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2163 337871 deprecate --db-attach
2164 338023 Add support for all V4L2/media ioctls
2165 338024 inlined functions are not shown if DW_AT_ranges is used
2166 338106 Add support for 'kcmp' syscall
2167 338115 DRD: computed conflict set differs from actual after fork
2168 338160 implement display of thread local storage in gdbsrv
2169 338205 configure.ac and check for -Wno-tautological-compare
2170 338300 coredumps are missing one byte of every segment
2171 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2172 338499 --sim-hints parsing broken due to wrong order in tokens
2173 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2174 338681 Unable to unwind through clone thread created on i386-linux
2175 338698 race condition between gdbsrv and vgdb on startup
2176 338703 helgrind on arm-linux gets false positives in dynamic loader
2177 338791 alt dwz files can be relative of debug/main file
2178 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2179 338932 build V-trunk with gcc-trunk
2180 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2181 345079 Fix build problems in VEX/useful/test_main.c
2182 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2183 n-i-bz s390x: Fix memory corruption for multithreaded applications
2184 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2185 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2186 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2187 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2188 n-i-bz Add command-line option --default-suppressions
2189 n-i-bz Add support for BLKDISCARDZEROES ioctl
2190 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2191 n-i-bz Add support for sys_pivot_root and sys_unshare
2193 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2194 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2195 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2199 Release 3.9.0 (31 October 2013)
2200 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2201 3.9.0 is a feature release with many improvements and the usual
2202 collection of bug fixes.
2204 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2205 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2206 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2207 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2209 * ================== PLATFORM CHANGES =================
2211 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2212 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2214 * Support for MIPS DSP ASE on MIPS32 platforms.
2216 * Support for s390x Decimal Floating Point instructions on hosts that
2217 have the DFP facility installed.
2219 * Support for POWER8 (Power ISA 2.07) instructions
2221 * Support for Intel AVX2 instructions. This is available only on 64
2224 * Initial support for Intel Transactional Synchronization Extensions,
2227 * Initial support for Hardware Transactional Memory on POWER.
2229 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2230 run large GUI apps tolerably well.
2232 * ==================== TOOL CHANGES ====================
2236 - Improvements in handling of vectorised code, leading to
2237 significantly fewer false error reports. You need to use the flag
2238 --partial-loads-ok=yes to get the benefits of these changes.
2240 - Better control over the leak checker. It is now possible to
2241 specify which leak kinds (definite/indirect/possible/reachable)
2242 should be displayed, which should be regarded as errors, and which
2243 should be suppressed by a given leak suppression. This is done
2244 using the options --show-leak-kinds=kind1,kind2,..,
2245 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2246 "match-leak-kinds:" line in suppression entries, respectively.
2248 Note that generated leak suppressions contain this new line and
2249 are therefore more specific than in previous releases. To get the
2250 same behaviour as previous releases, remove the "match-leak-kinds:"
2251 line from generated suppressions before using them.
2253 - Reduced "possible leak" reports from the leak checker by the use
2254 of better heuristics. The available heuristics provide detection
2255 of valid interior pointers to std::stdstring, to new[] allocated
2256 arrays with elements having destructors and to interior pointers
2257 pointing to an inner part of a C++ object using multiple
2258 inheritance. They can be selected individually using the
2259 option --leak-check-heuristics=heur1,heur2,...
2261 - Better control of stacktrace acquisition for heap-allocated
2262 blocks. Using the --keep-stacktraces option, it is possible to
2263 control independently whether a stack trace is acquired for each
2264 allocation and deallocation. This can be used to create better
2265 "use after free" errors or to decrease Valgrind's resource
2266 consumption by recording less information.
2268 - Better reporting of leak suppression usage. The list of used
2269 suppressions (shown when the -v option is given) now shows, for
2270 each leak suppressions, how many blocks and bytes it suppressed
2271 during the last leak search.
2275 - False errors resulting from the use of statically initialised
2276 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2279 - False errors resulting from the use of pthread_cond_waits that
2280 timeout, have been removed.
2282 * ==================== OTHER CHANGES ====================
2284 * Some attempt to tune Valgrind's space requirements to the expected
2285 capabilities of the target:
2287 - The default size of the translation cache has been reduced from 8
2288 sectors to 6 on Android platforms, since each sector occupies
2289 about 40MB when using Memcheck.
2291 - The default size of the translation cache has been increased to 16
2292 sectors on all other platforms, reflecting the fact that large
2293 applications require instrumentation and storage of huge amounts
2294 of code. For similar reasons, the number of memory mapped
2295 segments that can be tracked has been increased by a factor of 6.
2297 - In all cases, the maximum number of sectors in the translation
2298 cache can be controlled by the new flag --num-transtab-sectors.
2300 * Changes in how debug info (line numbers, etc) is read:
2302 - Valgrind no longer temporarily mmaps the entire object to read
2303 from it. Instead, reading is done through a small fixed sized
2304 buffer. This avoids virtual memory usage spikes when Valgrind
2305 reads debuginfo from large shared objects.
2307 - A new experimental remote debug info server. Valgrind can read
2308 debug info from a different machine (typically, a build host)
2309 where debuginfo objects are stored. This can save a lot of time
2310 and hassle when running Valgrind on resource-constrained targets
2311 (phones, tablets) when the full debuginfo objects are stored
2312 somewhere else. This is enabled by the --debuginfo-server=
2315 - Consistency checking between main and debug objects can be
2316 disabled using the --allow-mismatched-debuginfo option.
2318 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2319 scanning can recover stack traces in some cases when the normal
2320 unwind mechanisms fail. Stack scanning is best described as "a
2321 nasty, dangerous and misleading hack" and so is disabled by default.
2322 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2325 * Detection and merging of recursive stack frame cycles. When your
2326 program has recursive algorithms, this limits the memory used by
2327 Valgrind for recorded stack traces and avoids recording
2328 uninteresting repeated calls. This is controlled by the command
2329 line option --merge-recursive-frame and by the monitor command
2330 "v.set merge-recursive-frames".
2332 * File name and line numbers for used suppressions. The list of used
2333 suppressions (shown when the -v option is given) now shows, for each
2334 used suppression, the file name and line number where the suppression
2337 * New and modified GDB server monitor features:
2339 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2340 that can be used to execute gdbserver monitor commands from the
2343 - A new monitor command, "v.info open_fds", that gives the list of
2344 open file descriptors and additional details.
2346 - An optional message in the "v.info n_errs_found" monitor command,
2347 for example "v.info n_errs_found test 1234 finished", allowing a
2348 comment string to be added to the process output, perhaps for the
2349 purpose of separating errors of different tests or test phases.
2351 - A new monitor command "v.info execontext" that shows information
2352 about the stack traces recorded by Valgrind.
2354 - A new monitor command "v.do expensive_sanity_check_general" to run
2355 some internal consistency checks.
2357 * New flag --sigill-diagnostics to control whether a diagnostic
2358 message is printed when the JIT encounters an instruction it can't
2359 translate. The actual behavior -- delivery of SIGILL to the
2360 application -- is unchanged.
2362 * The maximum amount of memory that Valgrind can use on 64 bit targets
2363 has been increased from 32GB to 64GB. This should make it possible
2364 to run applications on Memcheck that natively require up to about 35GB.
2366 * ==================== FIXED BUGS ====================
2368 The following bugs have been fixed or resolved. Note that "n-i-bz"
2369 stands for "not in bugzilla" -- that is, a bug that was reported to us
2370 but never got a bugzilla entry. We encourage you to file bugs in
2371 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2372 than mailing the developers (or mailing lists) directly -- bugs that
2373 are not entered into bugzilla tend to get forgotten about or ignored.
2375 To see details of a given bug, visit
2376 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2377 where XXXXXX is the bug number as listed below.
2379 123837 system call: 4th argument is optional, depending on cmd
2380 135425 memcheck should tell you where Freed blocks were Mallocd
2381 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2382 207815 Adds some of the drm ioctls to syswrap-linux.c
2383 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2384 252955 Impossible to compile with ccache
2385 253519 Memcheck reports auxv pointer accesses as invalid reads.
2386 263034 Crash when loading some PPC64 binaries
2387 269599 Increase deepest backtrace
2388 274695 s390x: Support "compare to/from logical" instructions (z196)
2389 275800 s390x: Autodetect cache info (part 2)
2390 280271 Valgrind reports possible memory leaks on still-reachable std::string
2391 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2392 289578 Backtraces with ARM unwind tables (stack scan flags)
2393 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2394 304832 ppc32: build failure
2395 305431 Use find_buildid shdr fallback for separate .debug files
2396 305728 Add support for AVX2 instructions
2397 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2398 306035 s390x: Fix IR generation for LAAG and friends
2399 306054 s390x: Condition code computation for convert-to-int/logical
2400 306098 s390x: alternate opcode form for convert to/from fixed
2401 306587 Fix cache line detection from auxiliary vector for PPC.
2402 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2403 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2404 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2405 307101 sys_capget second argument can be NULL
2406 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2407 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2408 307113 s390x: DFP support
2409 307141 valgrind does't work in mips-linux system
2410 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2411 307285 x86_amd64 feature test for avx in test suite is wrong
2412 307290 memcheck overlap testcase needs memcpy version filter
2413 307463 Please add "&limit=0" to the "all open bugs" link
2414 307465 --show-possibly-lost=no should reduce the error count / exit code
2415 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2416 307729 pkgconfig support broken valgrind.pc
2417 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2418 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2419 308089 Unhandled syscall on ppc64: prctl
2420 308135 PPC32 MPC8xx has 16 bytes cache size
2421 308321 testsuite memcheck filter interferes with gdb_filter
2423 308341 vgdb should report process exit (or fatal signal)
2424 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2425 308495 Remove build dependency on installed Xen headers
2426 308573 Internal error on 64-bit instruction executed in 32-bit mode
2428 308627 pmovmskb validity bit propagation is imprecise
2429 308644 vgdb command for having the info for the track-fds option
2430 308711 give more info about aspacemgr and arenas in out_of_memory
2431 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2432 308718 ARM implement SMLALBB family of instructions
2433 308886 Missing support for PTRACE_SET/GETREGSET
2434 308930 syscall name_to_handle_at (303 on amd64) not handled
2435 309229 V-bit tester does not report number of tests generated
2436 309323 print unrecognized instuction on MIPS
2437 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2438 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2439 309430 Self hosting ppc64 encounters a vassert error on operand type
2440 309600 valgrind is a bit confused about 0-sized sections
2441 309823 Generate errors for still reachable blocks
2442 309921 PCMPISTRI validity bit propagation is imprecise
2443 309922 none/tests/ppc64/test_dfp5 sometimes fails
2444 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2445 310424 --read-var-info does not properly describe static variables
2446 310792 search additional path for debug symbols
2447 310931 s390x: Message-security assist (MSA) instruction extension [..]
2448 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2449 311318 ARM: "128-bit constant is not implemented" error message
2450 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2451 311690 V crashes because it redirects branches inside of a redirected function
2452 311880 x86_64: make regtest hangs at shell_valid1
2453 311922 WARNING: unhandled syscall: 170
2455 312171 ppc: insn selection for DFP
2456 312571 Rounding mode call wrong for the DFP Iops [..]
2457 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2458 312913 Dangling pointers error should also report the alloc stack trace
2459 312980 Building on Mountain Lion generates some compiler warnings
2460 313267 Adding MIPS64/Linux port to Valgrind
2463 313811 Buffer overflow in assert_fail
2464 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2465 314269 ppc: dead code in insn selection
2466 314718 ARM: implement integer divide instruction (sdiv and udiv)
2467 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2468 315441 sendmsg syscall should ignore unset msghdr msg_flags
2469 315534 msgrcv inside a thread causes valgrind to hang (block)
2470 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2471 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2472 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2473 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2474 316144 valgrind.1 manpage contains unknown ??? strings [..]
2475 316145 callgrind command line options in manpage reference (unknown) [..]
2476 316145 callgrind command line options in manpage reference [..]
2477 316181 drd: Fixed a 4x slowdown for certain applications
2478 316503 Valgrind does not support SSE4 "movntdqa" instruction
2479 316535 Use of |signed int| instead of |size_t| in valgrind messages
2480 316696 fluidanimate program of parsec 2.1 stuck
2481 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2482 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2483 317186 "Impossible happens" when occurs VCVT instruction on ARM
2484 317318 Support for Threading Building Blocks "scalable_malloc"
2485 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2486 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2487 317463 bmi testcase IR SANITY CHECK FAILURE
2488 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2489 318050 libmpiwrap fails to compile with out-of-source build
2490 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2491 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2492 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2493 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2494 318932 Add missing PPC64 and PPC32 system call support
2495 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2496 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2497 319494 VEX Makefile-gcc standalone build update after r2702
2498 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2499 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2500 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2501 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2502 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2503 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2504 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2505 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2506 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2507 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2508 320895 add fanotify support (patch included)
2509 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2510 321065 Valgrind updates for Xen 4.3
2511 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2512 321363 Unhandled instruction: SSAX (ARM + Thumb)
2513 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2514 321466 Unhandled instruction: SHASX (ARM + Thumb)
2515 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2516 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2517 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2518 321620 Unhandled instruction: UASX (ARM + Thumb)
2519 321621 Unhandled instruction: USAX (ARM + Thumb)
2520 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2521 321693 Unhandled instruction: LDRSBT (Thumb)
2522 321694 Unhandled instruction: UQASX (ARM + Thumb)
2523 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2524 321697 Unhandled instruction: UHASX (ARM + Thumb)
2525 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2526 321704 Unhandled instruction: REVSH (ARM + Thumb)
2527 321730 Add cg_diff and cg_merge man pages
2528 321738 Add vgdb and valgrind-listener man pages
2530 321891 Unhandled instruction: LDRHT (Thumb)
2531 321960 pthread_create() then alloca() causing invalid stack write errors
2532 321969 ppc32 and ppc64 don't support [lf]setxattr
2533 322254 Show threadname together with tid if set by application
2534 322294 Add initial support for IBM Power ISA 2.07
2535 322368 Assertion failure in wqthread_hijack under OS X 10.8
2536 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2537 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2538 322851 0bXXX binary literal syntax is not standard
2539 323035 Unhandled instruction: LDRSHT(Thumb)
2540 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2541 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2542 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2543 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2544 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2545 323437 Phase 2 support for IBM Power ISA 2.07
2546 323713 Support mmxext (integer sse) subset on i386 (athlon)
2547 323803 Transactional memory instructions are not supported for Power
2548 323893 SSE3 not available on amd cpus in valgrind
2549 323905 Probable false positive from Valgrind/drd on close()
2550 323912 valgrind.h header isn't compatible for mingw64
2551 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
2552 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
2553 324181 mmap does not handle MAP_32BIT
2554 324227 memcheck false positive leak when a thread calls exit+block [..]
2555 324421 Support for fanotify API on ARM architecture
2556 324514 gdbserver monitor cmd output behaviour consistency [..]
2557 324518 ppc64: Emulation of dcbt instructions does not handle [..]
2558 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
2559 324582 When access is made to freed memory, report both allocation [..]
2560 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
2561 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
2562 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
2563 324834 Unhandled instructions in Microsoft C run-time for x86_64
2564 324894 Phase 3 support for IBM Power ISA 2.07
2565 326091 drd: Avoid false race reports from optimized strlen() impls
2566 326113 valgrind libvex hwcaps error on AMD64
2567 n-i-bz Some wrong command line options could be ignored
2568 n-i-bz patch to allow fair-sched on android
2569 n-i-bz report error for vgdb snapshot requested before execution
2570 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
2572 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
2576 Release 3.8.1 (19 September 2012)
2577 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2578 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
2579 that occur moderately frequently in real use cases, adds support for
2580 some missing instructions on ARM, and fixes a deadlock condition on
2581 MacOSX. If you package or deliver 3.8.0 for others to use, you might
2582 want to consider upgrading to 3.8.1 instead.
2584 The following bugs have been fixed or resolved. Note that "n-i-bz"
2585 stands for "not in bugzilla" -- that is, a bug that was reported to us
2586 but never got a bugzilla entry. We encourage you to file bugs in
2587 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2588 than mailing the developers (or mailing lists) directly -- bugs that
2589 are not entered into bugzilla tend to get forgotten about or ignored.
2591 To see details of a given bug, visit
2592 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2593 where XXXXXX is the bug number as listed below.
2596 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
2597 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
2598 298281 wcslen causes false(?) uninitialised value warnings
2599 301281 valgrind hangs on OS X when the process calls system()
2600 304035 disInstr(arm): unhandled instruction 0xE1023053
2601 304867 implement MOVBE instruction in x86 mode
2602 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
2603 305042 amd64: implement 0F 7F encoding of movq between two registers
2604 305199 ARM: implement QDADD and QDSUB
2605 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
2606 305513 killed by fatal signal: SIGSEGV
2607 305690 DRD reporting invalid semaphore when sem_trywait fails
2608 305926 Invalid alignment checks for some AVX instructions
2609 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
2610 306310 3.8.0 release tarball missing some files
2611 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
2612 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
2613 n-i-bz shmat of a segment > 4Gb does not work
2614 n-i-bz simulate_control_c script wrong USR1 signal number on mips
2615 n-i-bz vgdb ptrace calls wrong on mips [...]
2616 n-i-bz Fixes for more MPI false positives
2617 n-i-bz exp-sgcheck's memcpy causes programs to segfault
2618 n-i-bz OSX build w/ clang: asserts at startup
2619 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
2620 n-i-bz fix a couple of union tag-vs-field mixups
2621 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
2623 The following bugs were fixed in 3.8.0 but not listed in this NEWS
2626 254088 Valgrind should know about UD2 instruction
2629 304754 NEWS blows TeX's little mind
2631 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
2635 Release 3.8.0 (10 August 2012)
2636 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2637 3.8.0 is a feature release with many improvements and the usual
2638 collection of bug fixes.
2640 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2641 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
2642 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
2643 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
2644 There is initial support for MacOSX 10.8, but it is not usable for
2645 serious work at present.
2647 * ================== PLATFORM CHANGES =================
2649 * Support for MIPS32 platforms running Linux. Valgrind has been
2650 tested on MIPS32 and MIPS32r2 platforms running different Debian
2651 Squeeze and MeeGo distributions. Both little-endian and big-endian
2652 cores are supported. The tools Memcheck, Massif and Lackey have
2653 been tested and are known to work. See README.mips for more details.
2655 * Preliminary support for Android running on x86.
2657 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
2659 * Support for Intel AVX instructions and for AES instructions. This
2660 support is available only for 64 bit code.
2662 * Support for POWER Decimal Floating Point instructions.
2664 * ==================== TOOL CHANGES ====================
2666 * Non-libc malloc implementations are now supported. This is useful
2667 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
2668 Using the new option --soname-synonyms, such tools can be informed
2669 that the malloc implementation is either linked statically into the
2670 executable, or is present in some other shared library different
2671 from libc.so. This makes it possible to process statically linked
2672 programs, and programs using other malloc libraries, for example
2673 TCMalloc or JEMalloc.
2675 * For tools that provide their own replacement for malloc et al, the
2676 option --redzone-size=<number> allows users to specify the size of
2677 the padding blocks (redzones) added before and after each client
2678 allocated block. Smaller redzones decrease the memory needed by
2679 Valgrind. Bigger redzones increase the chance to detect blocks
2680 overrun or underrun. Prior to this change, the redzone size was
2681 hardwired to 16 bytes in Memcheck.
2685 - The leak_check GDB server monitor command now can
2686 control the maximum nr of loss records to output.
2688 - Reduction of memory use for applications allocating
2689 many blocks and/or having many partially defined bytes.
2691 - Addition of GDB server monitor command 'block_list' that lists
2692 the addresses/sizes of the blocks of a leak search loss record.
2694 - Addition of GDB server monitor command 'who_points_at' that lists
2695 the locations pointing at a block.
2697 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2698 detect an invalid access of these redzones, by marking them
2699 noaccess. Similarly, if a redzone size is given for a memory
2700 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2701 This still allows to find some bugs if the user has forgotten to
2702 mark the pool superblock noaccess.
2704 - Performance of memory leak check has been improved, especially in
2705 cases where there are many leaked blocks and/or many suppression
2706 rules used to suppress leak reports.
2708 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2709 more precise analysis, which is important for LLVM/Clang
2710 generated code. This is at the cost of somewhat reduced
2711 performance. Note there is no change to analysis precision or
2712 costs on Linux targets.
2716 - Added even more facilities that can help finding the cause of a data
2717 race, namely the command-line option --ptrace-addr and the macro
2718 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2720 - Fixed a subtle bug that could cause false positive data race reports.
2722 * ==================== OTHER CHANGES ====================
2724 * The C++ demangler has been updated so as to work well with C++
2725 compiled by up to at least g++ 4.6.
2727 * Tool developers can make replacement/wrapping more flexible thanks
2728 to the new option --soname-synonyms. This was reported above, but
2729 in fact is very general and applies to all function
2730 replacement/wrapping, not just to malloc-family functions.
2732 * Round-robin scheduling of threads can be selected, using the new
2733 option --fair-sched= yes. Prior to this change, the pipe-based
2734 thread serialisation mechanism (which is still the default) could
2735 give very unfair scheduling. --fair-sched=yes improves
2736 responsiveness of interactive multithreaded applications, and
2737 improves repeatability of results from the thread checkers Helgrind
2740 * For tool developers: support to run Valgrind on Valgrind has been
2741 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2743 * gdbserver now shows the float shadow registers as integer
2744 rather than float values, as the shadow values are mostly
2745 used as bit patterns.
2747 * Increased limit for the --num-callers command line flag to 500.
2749 * Performance improvements for error matching when there are many
2750 suppression records in use.
2752 * Improved support for DWARF4 debugging information (bug 284184).
2754 * Initial support for DWZ compressed Dwarf debug info.
2756 * Improved control over the IR optimiser's handling of the tradeoff
2757 between performance and precision of exceptions. Specifically,
2758 --vex-iropt-precise-memory-exns has been removed and replaced by
2759 --vex-iropt-register-updates, with extended functionality. This
2760 allows the Valgrind gdbserver to always show up to date register
2763 * Modest performance gains through the use of translation chaining for
2766 * ==================== FIXED BUGS ====================
2768 The following bugs have been fixed or resolved. Note that "n-i-bz"
2769 stands for "not in bugzilla" -- that is, a bug that was reported to us
2770 but never got a bugzilla entry. We encourage you to file bugs in
2771 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2772 than mailing the developers (or mailing lists) directly -- bugs that
2773 are not entered into bugzilla tend to get forgotten about or ignored.
2775 To see details of a given bug, visit
2776 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2777 where XXXXXX is the bug number as listed below.
2779 197914 Building valgrind from svn now requires automake-1.10
2780 203877 increase to 16Mb maximum allowed alignment for memalign et al
2781 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
2782 247386 make perf does not run all performance tests
2783 270006 Valgrind scheduler unfair
2784 270777 Adding MIPS/Linux port to Valgrind
2785 270796 s390x: Removed broken support for the TS insn
2786 271438 Fix configure for proper SSE4.2 detection
2787 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
2788 273475 Add support for AVX instructions
2789 274078 improved configure logic for mpicc
2790 276993 fix mremap 'no thrash checks'
2791 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
2792 281482 memcheck incorrect byte allocation count in realloc() for silly argument
2793 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
2794 283413 Fix wrong sanity check
2795 283671 Robustize alignment computation in LibVEX_Alloc
2796 283961 Adding support for some HCI IOCTLs
2797 284124 parse_type_DIE: confused by: DWARF 4
2798 284864 == 273475 (Add support for AVX instructions)
2799 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
2800 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
2801 285725 == 273475 (Add support for AVX instructions)
2802 286261 add wrapper for linux I2C_RDWR ioctl
2803 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
2804 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
2805 286384 configure fails "checking for a supported version of gcc"
2806 286497 == 273475 (Add support for AVX instructions)
2807 286596 == 273475 (Add support for AVX instructions)
2808 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
2809 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
2810 287260 Incorrect conditional jump or move depends on uninitialised value(s)
2811 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
2812 287307 == 273475 (Add support for AVX instructions)
2813 287858 VG_(strerror): unknown error
2814 288298 (MacOSX) unhandled syscall shm_unlink
2815 288995 == 273475 (Add support for AVX instructions)
2816 289470 Loading of large Mach-O thin binaries fails.
2817 289656 == 273475 (Add support for AVX instructions)
2818 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
2819 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
2820 289839 s390x: Provide support for unicode conversion instructions
2821 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
2822 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
2823 290655 Add support for AESKEYGENASSIST instruction
2824 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
2825 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
2826 291253 ES register not initialised in valgrind simulation
2827 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
2828 291865 s390x: Support the "Compare Double and Swap" family of instructions
2829 292300 == 273475 (Add support for AVX instructions)
2830 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
2831 292493 == 273475 (Add support for AVX instructions)
2832 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
2833 292627 Missing support for some SCSI ioctls
2834 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
2835 292841 == 273475 (Add support for AVX instructions)
2836 292993 implement the getcpu syscall on amd64-linux
2837 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
2838 293088 Add some VEX sanity checks for ppc64 unhandled instructions
2839 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
2840 293754 PCMPxSTRx not implemented for 16-bit characters
2841 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
2842 293808 CLFLUSH not supported by latest VEX for amd64
2843 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
2844 294048 MPSADBW instruction not implemented
2845 294055 regtest none/tests/shell fails when locale is not set to C
2846 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
2847 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
2848 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
2849 294260 disInstr_AMD64: disInstr miscalculated next %rip
2850 294523 --partial-loads-ok=yes causes false negatives
2851 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
2852 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
2853 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
2854 295089 can not annotate source for both helgrind and drd
2855 295221 POWER Processor decimal floating point instruction support missing
2856 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
2857 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
2858 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
2859 295617 ARM - Add some missing syscalls
2860 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
2861 296229 Linux user input device ioctls missing wrappers
2862 296318 ELF Debug info improvements (more than one rx/rw mapping)
2863 296422 Add translation chaining support
2864 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
2865 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
2866 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
2867 297078 gdbserver signal handling problems [..]
2868 297147 drd false positives on newly allocated memory
2869 297329 disallow decoding of IBM Power DFP insns on some machines
2870 297497 POWER Processor decimal floating point instruction support missing
2871 297701 Another alias for strncasecmp_l in libc-2.13.so
2872 297911 'invalid write' not reported when using APIs for custom mem allocators.
2873 297976 s390x: revisit EX implementation
2874 297991 Valgrind interferes with mmap()+ftell()
2875 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
2876 297993 Fix compilation of valgrind with gcc -g3.
2877 298080 POWER Processor DFP support missing, part 3
2878 298227 == 273475 (Add support for AVX instructions)
2879 298335 == 273475 (Add support for AVX instructions)
2880 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
2881 298394 s390x: Don't bail out on an unknown machine model. [..]
2882 298421 accept4() syscall (366) support is missing for ARM
2883 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
2884 298732 valgrind installation problem in ubuntu with kernel version 3.x
2885 298862 POWER Processor DFP instruction support missing, part 4
2886 298864 DWARF reader mis-parses DW_FORM_ref_addr
2887 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
2888 299053 Support DWARF4 DW_AT_high_pc constant form
2889 299104 == 273475 (Add support for AVX instructions)
2890 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
2891 299629 dup3() syscall (358) support is missing for ARM
2892 299694 POWER Processor DFP instruction support missing, part 5
2893 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
2894 299803 == 273475 (Add support for AVX instructions)
2895 299804 == 273475 (Add support for AVX instructions)
2896 299805 == 273475 (Add support for AVX instructions)
2897 300140 ARM - Missing (T1) SMMUL
2898 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
2899 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
2900 300414 FCOM and FCOMP unimplemented for amd64 guest
2901 301204 infinite loop in canonicaliseSymtab with ifunc symbol
2902 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
2903 301265 add x86 support to Android build
2904 301984 configure script doesn't detect certain versions of clang
2905 302205 Fix compiler warnings for POWER VEX code and POWER test cases
2906 302287 Unhandled movbe instruction on Atom processors
2907 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
2908 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
2909 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
2910 302656 == 273475 (Add support for AVX instructions)
2911 302709 valgrind for ARM needs extra tls support for android emulator [..]
2912 302827 add wrapper for CDROM_GET_CAPABILITY
2913 302901 Valgrind crashes with dwz optimized debuginfo
2914 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
2915 303116 Add support for the POWER instruction popcntb
2916 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
2917 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
2918 303466 == 273475 (Add support for AVX instructions)
2919 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
2920 303963 strstr() function produces wrong results under valgrind callgrind
2921 304054 CALL_FN_xx macros need to enforce stack alignment
2922 304561 tee system call not supported
2923 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
2924 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
2925 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
2926 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
2927 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
2928 n-i-bz s390x: Shadow registers can now be examined using vgdb
2930 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
2931 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
2935 Release 3.7.0 (5 November 2011)
2936 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2937 3.7.0 is a feature release with many significant improvements and the
2938 usual collection of bug fixes.
2940 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2941 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
2942 Support for recent distros and toolchain components (glibc 2.14, gcc
2943 4.6, MacOSX 10.7) has been added.
2945 * ================== PLATFORM CHANGES =================
2947 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
2948 analyse 64-bit programs running on z/Architecture. Most user space
2949 instructions up to and including z10 are supported. Valgrind has
2950 been tested extensively on z9, z10, and z196 machines running SLES
2951 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
2952 known to work well. Callgrind, Helgrind, and DRD work reasonably
2953 well on z9 and later models. See README.s390 for more details.
2955 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
2956 64-bit processes are supported. Some complex threaded applications
2957 (Firefox) are observed to hang when run as 32 bit applications,
2958 whereas 64-bit versions run OK. The cause is unknown. Memcheck
2959 will likely report some false errors. In general, expect some rough
2960 spots. This release also supports MacOSX 10.6, but drops support
2963 * Preliminary support for Android (on ARM). Valgrind can now run
2964 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
2965 README.android for more details, plus instructions on how to get
2968 * Support for the IBM Power ISA 2.06 (Power7 instructions)
2970 * General correctness and performance improvements for ARM/Linux, and,
2971 by extension, ARM/Android.
2973 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
2974 instruction set support is under development but is not available in
2977 * Support for AIX5 has been removed.
2979 * ==================== TOOL CHANGES ====================
2981 * Memcheck: some incremental changes:
2983 - reduction of memory use in some circumstances
2985 - improved handling of freed memory, which in some circumstances
2986 can cause detection of use-after-free that would previously have
2989 - fix of a longstanding bug that could cause false negatives (missed
2990 errors) in programs doing vector saturated narrowing instructions.
2992 * Helgrind: performance improvements and major memory use reductions,
2993 particularly for large, long running applications which perform many
2994 synchronisation (lock, unlock, etc) events. Plus many smaller
2997 - display of locksets for both threads involved in a race
2999 - general improvements in formatting/clarity of error messages
3001 - addition of facilities and documentation regarding annotation
3002 of thread safe reference counted C++ classes
3004 - new flag --check-stack-refs=no|yes [yes], to disable race checking
3005 on thread stacks (a performance hack)
3007 - new flag --free-is-write=no|yes [no], to enable detection of races
3008 where one thread accesses heap memory but another one frees it,
3009 without any coordinating synchronisation event
3011 * DRD: enabled XML output; added support for delayed thread deletion
3012 in order to detect races that occur close to the end of a thread
3013 (--join-list-vol); fixed a memory leak triggered by repeated client
3014 memory allocatation and deallocation; improved Darwin support.
3016 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
3018 * exp-sgcheck: this tool has been reduced in scope so as to improve
3019 performance and remove checking that Memcheck does better.
3020 Specifically, the ability to check for overruns for stack and global
3021 arrays is unchanged, but the ability to check for overruns of heap
3022 blocks has been removed. The tool has accordingly been renamed to
3023 exp-sgcheck ("Stack and Global Array Checking").
3025 * ==================== OTHER CHANGES ====================
3027 * GDB server: Valgrind now has an embedded GDB server. That means it
3028 is possible to control a Valgrind run from GDB, doing all the usual
3029 things that GDB can do (single stepping, breakpoints, examining
3030 data, etc). Tool-specific functionality is also available. For
3031 example, it is possible to query the definedness state of variables
3032 or memory from within GDB when running Memcheck; arbitrarily large
3033 memory watchpoints are supported, etc. To use the GDB server, start
3034 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3037 * Improved support for unfriendly self-modifying code: a new option
3038 --smc-check=all-non-file is available. This adds the relevant
3039 consistency checks only to code that originates in non-file-backed
3040 mappings. In effect this confines the consistency checking only to
3041 code that is or might be JIT generated, and avoids checks on code
3042 that must have been compiled ahead of time. This significantly
3043 improves performance on applications that generate code at run time.
3045 * It is now possible to build a working Valgrind using Clang-2.9 on
3048 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3049 These enable and disable error reporting on a per-thread, and
3050 nestable, basis. This is useful for hiding errors in particularly
3051 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3052 now uses this facility.
3054 * Added the --mod-funcname option to cg_diff.
3056 * ==================== FIXED BUGS ====================
3058 The following bugs have been fixed or resolved. Note that "n-i-bz"
3059 stands for "not in bugzilla" -- that is, a bug that was reported to us
3060 but never got a bugzilla entry. We encourage you to file bugs in
3061 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3062 mailing the developers (or mailing lists) directly -- bugs that are
3063 not entered into bugzilla tend to get forgotten about or ignored.
3065 To see details of a given bug, visit
3066 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3067 where XXXXXX is the bug number as listed below.
3069 79311 malloc silly arg warning does not give stack trace
3070 210935 port valgrind.h (not valgrind) to win32 to support client requests
3071 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3072 243404 Port to zSeries
3073 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3074 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3075 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3076 253206 Some fixes for the faultstatus testcase
3077 255223 capget testcase fails when running as root
3078 256703 xlc_dbl_u32.c testcase broken
3079 256726 Helgrind tests have broken inline asm
3080 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3081 264800 testcase compile failure on zseries
3082 265762 make public VEX headers compilable by G++ 3.x
3083 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3084 266753 configure script does not give the user the option to not use QtCore
3085 266931 gen_insn_test.pl is broken
3086 266961 ld-linux.so.2 i?86-linux strlen issues
3087 266990 setns instruction causes false positive
3088 267020 Make directory for temporary files configurable at run-time.
3089 267342 == 267997 (segmentation fault on Mac OS 10.6)
3090 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3091 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3092 267488 regtest: darwin support for 64-bit build
3093 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3094 267630 Add support for IBM Power ISA 2.06 -- stage 1
3095 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3096 267819 Add client request for informing the core about reallocation
3097 267925 laog data structure quadratic for a single sequence of lock
3098 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3099 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3100 268513 missed optimizations in fold_Expr
3101 268619 s390x: fpr - gpr transfer facility
3102 268620 s390x: reconsider "long displacement" requirement
3103 268621 s390x: improve IR generation for XC
3104 268715 s390x: FLOGR is not universally available
3105 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3106 268930 s390x: MHY is not universally available
3107 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3108 269079 Support ptrace system call on ARM
3109 269144 missing "Bad option" error message
3110 269209 conditional load and store facility (z196)
3111 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3112 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3113 269736 s390x: minor code generation tweaks
3114 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3115 269863 s390x: remove unused function parameters
3116 269864 s390x: tweak s390_emit_load_cc
3117 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3118 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3119 270115 s390x: rewrite some testcases
3120 270309 == 267997 (valgrind crash on startup)
3121 270320 add support for Linux FIOQSIZE ioctl() call
3122 270326 segfault while trying to sanitize the environment passed to execle
3123 270794 IBM POWER7 support patch causes regression in none/tests
3124 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3125 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3126 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3127 270959 s390x: invalid use of R0 as base register
3128 271042 VSX configure check fails when it should not
3129 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3130 271259 s390x: fix code confusion
3131 271337 == 267997 (Valgrind segfaults on MacOS X)
3132 271385 s390x: Implement Ist_MBE
3133 271501 s390x: misc cleanups
3134 271504 s390x: promote likely and unlikely
3135 271579 ppc: using wrong enum type
3136 271615 unhandled instruction "popcnt" (arch=amd10h)
3137 271730 Fix bug when checking ioctls: duplicate check
3138 271776 s390x: provide STFLE instruction support
3139 271779 s390x: provide clock instructions like STCK
3140 271799 Darwin: ioctls without an arg report a memory error
3141 271820 arm: fix type confusion
3142 271917 pthread_cond_timedwait failure leads to not-locked false positive
3143 272067 s390x: fix DISP20 macro
3144 272615 A typo in debug output in mc_leakcheck.c
3145 272661 callgrind_annotate chokes when run from paths containing regex chars
3146 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3147 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3148 272967 make documentation build-system more robust
3149 272986 Fix gcc-4.6 warnings with valgrind.h
3150 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3151 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3152 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3153 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3154 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3155 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3156 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3157 273778 exp-ptrcheck: unhandled sysno == 259
3158 274089 exp-ptrcheck: unhandled sysno == 208
3159 274378 s390x: Various dispatcher tweaks
3160 274447 WARNING: unhandled syscall: 340
3161 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3162 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3163 274926 valgrind does not build against linux-3
3164 275148 configure FAIL with glibc-2.14
3165 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3166 275168 Make Valgrind work for MacOSX 10.7 Lion
3167 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3168 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3169 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3170 275308 Fix implementation for ppc64 fres instruc
3171 275339 s390x: fix testcase compile warnings
3172 275517 s390x: Provide support for CKSM instruction
3173 275710 s390x: get rid of redundant address mode calculation
3174 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3175 275852 == 250101 (valgrind uses all swap space and is killed)
3176 276784 Add support for IBM Power ISA 2.06 -- stage 3
3177 276987 gdbsrv: fix tests following recent commits
3178 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3179 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3180 277471 Unhandled syscall: 340
3181 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3182 277653 ARM: support Thumb2 PLD instruction
3183 277663 ARM: NEON float VMUL by scalar incorrect
3184 277689 ARM: tests for VSTn with register post-index are broken
3185 277694 ARM: BLX LR instruction broken in ARM mode
3186 277780 ARM: VMOV.F32 (immediate) instruction is broken
3187 278057 fuse filesystem syscall deadlocks
3188 278078 Unimplemented syscall 280 on ppc32
3189 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3190 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3191 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3192 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3193 279027 Support for MVCL and CLCL instruction
3194 279027 s390x: Provide support for CLCL and MVCL instructions
3195 279062 Remove a redundant check in the insn selector for ppc.
3196 279071 JDK creates PTEST with redundant REX.W prefix
3197 279212 gdbsrv: add monitor cmd v.info scheduler.
3198 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3199 279698 memcheck discards valid-bits for packuswb
3200 279795 memcheck reports uninitialised values for mincore on amd64
3201 279994 Add support for IBM Power ISA 2.06 -- stage 3
3202 280083 mempolicy syscall check errors
3203 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3204 280710 s390x: config files for nightly builds
3205 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3206 280965 Valgrind breaks fcntl locks when program does mmap
3207 281138 WARNING: unhandled syscall: 340
3208 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3209 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3210 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3211 281468 s390x: handle do_clone and gcc clones in call traces
3212 281488 ARM: VFP register corruption
3213 281828 == 275284 (false memmove warning: "Source and destination overlap")
3214 281883 s390x: Fix system call wrapper for "clone".
3215 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3216 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3217 282238 SLES10: make check fails
3218 282979 strcasestr needs replacement with recent(>=2.12) glibc
3219 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3220 283243 Regression in ppc64 memcheck tests
3221 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3222 283427 re-connect epoll_pwait syscall on ARM linux
3223 283600 gdbsrv: android: port vgdb.c
3224 283709 none/tests/faultstatus needs to account for page size
3225 284305 filter_gdb needs enhancement to work on ppc64
3226 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3227 284472 Thumb2 ROR.W encoding T2 not implemented
3228 284621 XML-escape process command line in XML output
3229 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3230 that have non-power-of-2 sizes (also AMDs)
3231 n-i-bz don't be spooked by libraries mashed by elfhack
3232 n-i-bz don't be spooked by libxul.so linked with gold
3233 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3235 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3236 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3237 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3241 Release 3.6.1 (16 February 2011)
3242 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3243 3.6.1 is a bug fix release. It adds support for some SSE4
3244 instructions that were omitted in 3.6.0 due to lack of time. Initial
3245 support for glibc-2.13 has been added. A number of bugs causing
3246 crashing or assertion failures have been fixed.
3248 The following bugs have been fixed or resolved. Note that "n-i-bz"
3249 stands for "not in bugzilla" -- that is, a bug that was reported to us
3250 but never got a bugzilla entry. We encourage you to file bugs in
3251 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3252 mailing the developers (or mailing lists) directly -- bugs that are
3253 not entered into bugzilla tend to get forgotten about or ignored.
3255 To see details of a given bug, visit
3256 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3257 where XXXXXX is the bug number as listed below.
3259 188572 Valgrind on Mac should suppress setenv() mem leak
3260 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3261 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3262 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3263 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3264 254420 memory pool tracking broken
3265 254957 Test code failing to compile due to changes in memcheck.h
3266 255009 helgrind/drd: crash on chmod with invalid parameter
3267 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3268 255355 helgrind/drd: crash on threaded programs doing fork
3270 255418 (SSE4.x) rint call compiled with ICC
3271 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3272 255888 closing valgrindoutput tag outputted to log-stream on error
3273 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3274 255966 Slowness when using mempool annotations
3275 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3276 256600 super-optimized strcasecmp() false positive
3277 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3278 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3279 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3280 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3281 257276 Missing case in memcheck --track-origins=yes
3282 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3283 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3284 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3285 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3286 263099 callgrind_annotate counts Ir improperly [...]
3287 263877 undefined coprocessor instruction on ARMv7
3288 265964 configure FAIL with glibc-2.13
3289 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3290 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3291 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3292 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3293 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3295 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3299 Release 3.6.0 (21 October 2010)
3300 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3301 3.6.0 is a feature release with many significant improvements and the
3302 usual collection of bug fixes.
3304 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3305 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3306 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3308 -------------------------
3310 Here are some highlights. Details are shown further down:
3312 * Support for ARM/Linux.
3314 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3316 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3318 * Support for the SSE4.2 instruction set.
3320 * Enhancements to the Callgrind profiler, including the ability to
3321 handle CPUs with three levels of cache.
3323 * A new experimental heap profiler, DHAT.
3325 * A huge number of bug fixes and small enhancements.
3327 -------------------------
3329 Here are details of the above changes, together with descriptions of
3330 many other changes, and a list of fixed bugs.
3332 * ================== PLATFORM CHANGES =================
3334 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3335 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3336 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3338 This requires a CPU capable of running the ARMv7-A instruction set
3339 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3340 of the user space instruction set, including ARM and Thumb integer
3341 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3342 Cachegrind and Massif tools work properly; other tools work to
3345 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3346 with support for recent releases of the underlying toolchain
3347 components, notably gcc-4.5 and glibc-2.12.
3349 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3350 support also works much better on OS X 10.5, and is as solid as
3353 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3354 64-bit mode. In 32-bit mode, support is only available up to and
3355 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3356 supported in 64-bit mode, and 32-bit mode does in fact support the
3357 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3358 10.6 on 32-bit targets.
3360 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3361 and including version 2.05 is supported.
3363 * ==================== TOOL CHANGES ====================
3365 * Cachegrind has a new processing script, cg_diff, which finds the
3366 difference between two profiles. It's very useful for evaluating
3367 the performance effects of a change in a program.
3369 Related to this change, the meaning of cg_annotate's (rarely-used)
3370 --threshold option has changed; this is unlikely to affect many
3371 people, if you do use it please see the user manual for details.
3373 * Callgrind now can do branch prediction simulation, similar to
3374 Cachegrind. In addition, it optionally can count the number of
3375 executed global bus events. Both can be used for a better
3376 approximation of a "Cycle Estimation" as derived event (you need to
3377 update the event formula in KCachegrind yourself).
3379 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3380 rather than the L2 cache. This is to accommodate machines with
3381 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3382 cache configuration of such a machine it will run the simulation as
3383 if the L2 cache isn't present. This means the results are less
3384 likely to match the true result for the machine, but
3385 Cachegrind/Callgrind's results are already only approximate, and
3386 should not be considered authoritative. The results are still
3387 useful for giving a general idea about a program's locality.
3389 * Massif has a new option, --pages-as-heap, which is disabled by
3390 default. When enabled, instead of tracking allocations at the level
3391 of heap blocks (as allocated with malloc/new/new[]), it instead
3392 tracks memory allocations at the level of memory pages (as mapped by
3393 mmap, brk, etc). Each mapped page is treated as its own block.
3394 Interpreting the page-level output is harder than the heap-level
3395 output, but this option is useful if you want to account for every
3396 byte of memory used by a program.
3398 * DRD has two new command-line options: --free-is-write and
3399 --trace-alloc. The former allows to detect reading from already freed
3400 memory, and the latter allows tracing of all memory allocations and
3403 * DRD has several new annotations. Custom barrier implementations can
3404 now be annotated, as well as benign races on static variables.
3406 * DRD's happens before / happens after annotations have been made more
3407 powerful, so that they can now also be used to annotate e.g. a smart
3408 pointer implementation.
3410 * Helgrind's annotation set has also been drastically improved, so as
3411 to provide to users a general set of annotations to describe locks,
3412 semaphores, barriers and condition variables. Annotations to
3413 describe thread-safe reference counted heap objects have also been
3416 * Memcheck has a new command-line option, --show-possibly-lost, which
3417 is enabled by default. When disabled, the leak detector will not
3418 show possibly-lost blocks.
3420 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3421 has been added. DHAT keeps track of allocated heap blocks, and also
3422 inspects every memory reference to see which block (if any) is being
3423 accessed. This gives a lot of insight into block lifetimes,
3424 utilisation, turnover, liveness, and the location of hot and cold
3425 fields. You can use DHAT to do hot-field profiling.
3427 * ==================== OTHER CHANGES ====================
3429 * Improved support for unfriendly self-modifying code: the extra
3430 overhead incurred by --smc-check=all has been reduced by
3431 approximately a factor of 5 as compared with 3.5.0.
3433 * Ability to show directory names for source files in error messages.
3434 This is combined with a flexible mechanism for specifying which
3435 parts of the paths should be shown. This is enabled by the new flag
3438 * A new flag, --require-text-symbol, which will stop the run if a
3439 specified symbol is not found it a given shared object when it is
3440 loaded into the process. This makes advanced working with function
3441 intercepting and wrapping safer and more reliable.
3443 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3444 and control of Valgrind is now available for the tools Memcheck and
3445 Helgrind. XML output from Valgrind is available for Memcheck,
3446 Helgrind and exp-Ptrcheck.
3448 * More reliable stack unwinding on amd64-linux, particularly in the
3449 presence of function wrappers, and with gcc-4.5 compiled code.
3451 * Modest scalability (performance improvements) for massive
3452 long-running applications, particularly for those with huge amounts
3455 * Support for analyzing programs running under Wine with has been
3456 improved. The header files <valgrind/valgrind.h>,
3457 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3458 Windows-programs compiled with MinGW or one of the Microsoft Visual
3461 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3462 The 32-bit simulator was not affected. This did not occur often,
3463 but when it did would usually crash the program under test.
3466 * A large number of bugs were fixed. These are shown below.
3468 * A number of bugs were investigated, and were candidates for fixing,
3469 but are not fixed in 3.6.0, due to lack of developer time. They may
3470 get fixed in later releases. They are:
3472 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3473 212419 false positive "lock order violated" (A+B vs A)
3474 213685 Undefined value propagates past dependency breaking instruction
3475 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3476 237920 valgrind segfault on fork failure
3477 242137 support for code compiled by LLVM-2.8
3478 242423 Another unknown Intel cache config value
3479 243232 Inconsistent Lock Orderings report with trylock
3480 243483 ppc: callgrind triggers VEX assertion failure
3481 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3482 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3484 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3485 249435 Analyzing wine programs with callgrind triggers a crash
3486 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3487 250065 Handling large allocations
3488 250101 huge "free" memory usage due to m_mallocfree.c
3489 "superblocks fragmentation"
3490 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3491 252091 Callgrind on ARM does not detect function returns correctly
3492 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3493 254420 memory pool tracking broken
3494 n-i-bz support for adding symbols for JIT generated code
3497 The following bugs have been fixed or resolved. Note that "n-i-bz"
3498 stands for "not in bugzilla" -- that is, a bug that was reported to us
3499 but never got a bugzilla entry. We encourage you to file bugs in
3500 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3501 mailing the developers (or mailing lists) directly -- bugs that are
3502 not entered into bugzilla tend to get forgotten about or ignored.
3504 To see details of a given bug, visit
3505 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3506 where XXXXXX is the bug number as listed below.
3508 135264 dcbzl instruction missing
3510 153699 Valgrind should report unaligned reads with movdqa
3512 190429 Valgrind reports lost of errors in ld.so
3513 with x86_64 2.9.90 glibc
3514 197266 valgrind appears to choke on the xmms instruction
3516 197988 Crash when demangling very large symbol names
3517 202315 unhandled syscall: 332 (inotify_init1)
3518 203256 Add page-level profiling to Massif
3519 205093 dsymutil=yes needs quotes, locking (partial fix)
3520 205241 Snow Leopard 10.6 support (partial fix)
3521 206600 Leak checker fails to upgrade indirect blocks when their
3522 parent becomes reachable
3523 210935 port valgrind.h (not valgrind) to win32 so apps run under
3524 wine can make client requests
3525 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3526 within Linux ip-stack checksum functions
3527 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3529 213685 Undefined value propagates past dependency breaking instruction
3531 215914 Valgrind inserts bogus empty environment variable
3533 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3534 222545 shmat fails under valgind on some arm targets
3535 222560 ARM NEON support
3538 232509 Docs build fails with formatting inside <title></title> elements
3540 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3541 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3542 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3543 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3544 237485 symlink (syscall 57) is not supported on Mac OS
3545 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3547 238208 is_just_below_ESP doesn't take into account red-zone
3548 238345 valgrind passes wrong $0 when executing a shell script
3549 238679 mq_timedreceive syscall doesn't flag the reception buffer
3551 238696 fcntl command F_DUPFD_CLOEXEC not supported
3552 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3553 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3554 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
3556 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
3562 242606 unhandled syscall: setegid (in Ptrcheck)
3563 242814 Helgrind "Impossible has happened" during
3564 QApplication::initInstance();
3565 243064 Valgrind attempting to read debug information from iso
3566 243270 Make stack unwinding in Valgrind wrappers more reliable
3567 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
3568 sysno = 277 (mq_open)
3569 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
3570 244493 ARM VFP d16-d31 registers support
3571 244670 add support for audit_session_self syscall on Mac OS 10.6
3572 244921 The xml report of helgrind tool is not well format
3573 244923 In the xml report file, the <preamble> not escape the
3574 xml char, eg '<','&','>'
3575 245535 print full path names in plain text reports
3576 245925 x86-64 red zone handling problem
3577 246258 Valgrind not catching integer underruns + new [] s
3578 246311 reg/reg cmpxchg doesn't work on amd64
3579 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
3580 246888 Improve Makefile.vex.am
3581 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
3582 to [f]chmod_extended
3583 247526 IBM POWER6 (ISA 2.05) support is incomplete
3584 247561 Some leak testcases fails due to reachable addresses in
3586 247875 sizeofIRType to handle Ity_I128
3587 247894 [PATCH] unhandled syscall sys_readahead
3588 247980 Doesn't honor CFLAGS passed to configure
3589 248373 darwin10.supp is empty in the trunk
3590 248822 Linux FIBMAP ioctl has int parameter instead of long
3591 248893 [PATCH] make readdwarf.c big endianess safe to enable
3592 unwinding on big endian systems
3593 249224 Syscall 336 not supported (SYS_proc_info)
3595 249775 Incorrect scheme for detecting NEON capabilities of host CPU
3596 249943 jni JVM init fails when using valgrind
3597 249991 Valgrind incorrectly declares AESKEYGENASSIST support
3599 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
3600 250799 frexp$fenv_access_off function generates SIGILL
3601 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
3602 251251 support pclmulqdq insn
3603 251362 valgrind: ARM: attach to debugger either fails or provokes
3605 251674 Unhandled syscall 294
3608 254257 Add support for debugfiles found by build-id
3609 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
3610 254646 Wrapped functions cause stack misalignment on OS X
3611 (and possibly Linux)
3612 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
3614 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
3618 Release 3.5.0 (19 August 2009)
3619 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3620 3.5.0 is a feature release with many significant improvements and the
3621 usual collection of bug fixes. The main improvement is that Valgrind
3622 now works on Mac OS X.
3624 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
3625 and X86/Darwin. Support for recent distros and toolchain components
3626 (glibc 2.10, gcc 4.5) has been added.
3628 -------------------------
3630 Here is a short summary of the changes. Details are shown further
3633 * Support for Mac OS X (10.5.x).
3635 * Improvements and simplifications to Memcheck's leak checker.
3637 * Clarification and simplifications in various aspects of Valgrind's
3640 * XML output for Helgrind and Ptrcheck.
3642 * Performance and stability improvements for Helgrind and DRD.
3644 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
3646 * A new experimental tool, BBV, useful for computer architecture
3649 * Improved Wine support, including ability to read Windows PDB
3652 -------------------------
3654 Here are details of the above changes, followed by descriptions of
3655 many other minor changes, and a list of fixed bugs.
3658 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
3659 called "Darwin" because that is the name of the OS core, which is the
3660 level that Valgrind works at.)
3664 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
3665 because it would require work and 10.4 is only becoming less common.
3667 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
3668 fairly well. For 10.5.x, 32-bit programs are the default even on
3669 64-bit machines, so it handles most current programs.
3671 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
3672 officially supported, but simple programs at least will probably work.
3673 However, start-up is slow.
3675 - PowerPC machines are not supported.
3677 Things that don't work:
3679 - The Ptrcheck tool.
3681 - Objective-C garbage collection.
3685 - If you have Rogue Amoeba's "Instant Hijack" program installed,
3686 Valgrind will fail with a SIGTRAP at start-up. See
3687 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
3692 - You will likely find --dsymutil=yes a useful option, as error
3693 messages may be imprecise without it.
3695 - Mac OS X support is new and therefore will be less robust than the
3696 Linux support. Please report any bugs you find.
3698 - Threaded programs may run more slowly than on Linux.
3700 Many thanks to Greg Parker for developing this port over several years.
3703 * Memcheck's leak checker has been improved.
3705 - The results for --leak-check=summary now match the summary results
3706 for --leak-check=full. Previously they could differ because
3707 --leak-check=summary counted "indirectly lost" blocks and
3708 "suppressed" blocks as "definitely lost".
3710 - Blocks that are only reachable via at least one interior-pointer,
3711 but are directly pointed to by a start-pointer, were previously
3712 marked as "still reachable". They are now correctly marked as
3715 - The default value for the --leak-resolution option has been
3716 changed from "low" to "high". In general, this means that more
3717 leak reports will be produced, but each leak report will describe
3718 fewer leaked blocks.
3720 - With --leak-check=full, "definitely lost" and "possibly lost"
3721 leaks are now considered as proper errors, ie. they are counted
3722 for the "ERROR SUMMARY" and affect the behaviour of
3723 --error-exitcode. These leaks are not counted as errors if
3724 --leak-check=summary is specified, however.
3726 - Documentation for the leak checker has been improved.
3729 * Various aspects of Valgrind's text output have changed.
3731 - Valgrind's start-up message has changed. It is shorter but also
3732 includes the command being run, which makes it easier to use
3733 --trace-children=yes. An example:
3735 - Valgrind's shut-down messages have also changed. This is most
3736 noticeable with Memcheck, where the leak summary now occurs before
3737 the error summary. This change was necessary to allow leaks to be
3738 counted as proper errors (see the description of the leak checker
3739 changes above for more details). This was also necessary to fix a
3740 longstanding bug in which uses of suppressions against leaks were
3741 not "counted", leading to difficulties in maintaining suppression
3742 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3744 - Behavior of -v has changed. In previous versions, -v printed out
3745 a mixture of marginally-user-useful information, and tool/core
3746 statistics. The statistics printing has now been moved to its own
3747 flag, --stats=yes. This means -v is less verbose and more likely
3748 to convey useful end-user information.
3750 - The format of some (non-XML) stack trace entries has changed a
3751 little. Previously there were six possible forms:
3753 0x80483BF: really (a.c:20)
3754 0x80483BF: really (in /foo/a.out)
3756 0x80483BF: (within /foo/a.out)
3757 0x80483BF: ??? (a.c:20)
3760 The third and fourth of these forms have been made more consistent
3761 with the others. The six possible forms are now:
3763 0x80483BF: really (a.c:20)
3764 0x80483BF: really (in /foo/a.out)
3765 0x80483BF: really (in ???)
3766 0x80483BF: ??? (in /foo/a.out)
3767 0x80483BF: ??? (a.c:20)
3770 Stack traces produced when --xml=yes is specified are different
3774 * Helgrind and Ptrcheck now support XML output, so they can be used
3775 from GUI tools. Also, the XML output mechanism has been
3778 - The XML format has been overhauled and generalised, so it is more
3779 suitable for error reporting tools in general. The Memcheck
3780 specific aspects of it have been removed. The new format, which
3781 is an evolution of the old format, is described in
3782 docs/internals/xml-output-protocol4.txt.
3784 - Memcheck has been updated to use the new format.
3786 - Helgrind and Ptrcheck are now able to emit output in this format.
3788 - The XML output mechanism has been overhauled. XML is now output
3789 to its own file descriptor, which means that:
3791 * Valgrind can output text and XML independently.
3793 * The longstanding problem of XML output being corrupted by
3794 unexpected un-tagged text messages is solved.
3796 As before, the destination for text output is specified using
3797 --log-file=, --log-fd= or --log-socket=.
3799 As before, XML output for a tool is enabled using --xml=yes.
3801 Because there's a new XML output channel, the XML output
3802 destination is now specified by --xml-file=, --xml-fd= or
3805 Initial feedback has shown this causes some confusion. To
3806 clarify, the two envisaged usage scenarios are:
3808 (1) Normal text output. In this case, do not specify --xml=yes
3809 nor any of --xml-file=, --xml-fd= or --xml-socket=.
3811 (2) XML output. In this case, specify --xml=yes, and one of
3812 --xml-file=, --xml-fd= or --xml-socket= to select the XML
3813 destination, one of --log-file=, --log-fd= or --log-socket=
3814 to select the destination for any remaining text messages,
3815 and, importantly, -q.
3817 -q makes Valgrind completely silent on the text channel,
3818 except in the case of critical failures, such as Valgrind
3819 itself segfaulting, or failing to read debugging information.
3820 Hence, in this scenario, it suffices to check whether or not
3821 any output appeared on the text channel. If yes, then it is
3822 likely to be a critical error which should be brought to the
3823 attention of the user. If no (the text channel produced no
3824 output) then it can be assumed that the run was successful.
3826 This allows GUIs to make the critical distinction they need to
3827 make (did the run fail or not?) without having to search or
3828 filter the text output channel in any way.
3830 It is also recommended to use --child-silent-after-fork=yes in
3834 * Improvements and changes in Helgrind:
3836 - XML output, as described above
3838 - Checks for consistent association between pthread condition
3839 variables and their associated mutexes are now performed.
3841 - pthread_spinlock functions are supported.
3843 - Modest performance improvements.
3845 - Initial (skeletal) support for describing the behaviour of
3846 non-POSIX synchronisation objects through ThreadSanitizer
3847 compatible ANNOTATE_* macros.
3849 - More controllable tradeoffs between performance and the level of
3850 detail of "previous" accesses in a race. There are now three
3853 * --history-level=full. This is the default, and was also the
3854 default in 3.4.x. It shows both stacks involved in a race, but
3855 requires a lot of memory and can be very slow in programs that
3856 do many inter-thread synchronisation events.
3858 * --history-level=none. This only shows the later stack involved
3859 in a race. This can be much faster than --history-level=full,
3860 but makes it much more difficult to find the other access
3861 involved in the race.
3863 The new intermediate setting is
3865 * --history-level=approx
3867 For the earlier (other) access, two stacks are presented. The
3868 earlier access is guaranteed to be somewhere in between the two
3869 program points denoted by those stacks. This is not as useful
3870 as showing the exact stack for the previous access (as per
3871 --history-level=full), but it is better than nothing, and it's
3872 almost as fast as --history-level=none.
3875 * New features and improvements in DRD:
3877 - The error messages printed by DRD are now easier to interpret.
3878 Instead of using two different numbers to identify each thread
3879 (Valgrind thread ID and DRD thread ID), DRD does now identify
3880 threads via a single number (the DRD thread ID). Furthermore
3881 "first observed at" information is now printed for all error
3882 messages related to synchronization objects.
3884 - Added support for named semaphores (sem_open() and sem_close()).
3886 - Race conditions between pthread_barrier_wait() and
3887 pthread_barrier_destroy() calls are now reported.
3889 - Added support for custom allocators through the macros
3890 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
3891 in <valgrind/valgrind.h>). An alternative for these two macros is
3892 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
3895 - Added support for annotating non-POSIX synchronization objects
3896 through several new ANNOTATE_*() macros.
3898 - OpenMP: added support for the OpenMP runtime (libgomp) included
3899 with gcc versions 4.3.0 and 4.4.0.
3903 - Added two new command-line options (--first-race-only and
3904 --segment-merging-interval).
3907 * Genuinely atomic support for x86/amd64/ppc atomic instructions
3909 Valgrind will now preserve (memory-access) atomicity of LOCK-
3910 prefixed x86/amd64 instructions, and any others implying a global
3911 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
3913 This means that Valgrinded processes will "play nicely" in
3914 situations where communication with other processes, or the kernel,
3915 is done through shared memory and coordinated with such atomic
3916 instructions. Prior to this change, such arrangements usually
3917 resulted in hangs, races or other synchronisation failures, because
3918 Valgrind did not honour atomicity of such instructions.
3921 * A new experimental tool, BBV, has been added. BBV generates basic
3922 block vectors for use with the SimPoint analysis tool, which allows
3923 a program's overall behaviour to be approximated by running only a
3924 fraction of it. This is useful for computer architecture
3925 researchers. You can run BBV by specifying --tool=exp-bbv (the
3926 "exp-" prefix is short for "experimental"). BBV was written by
3930 * Modestly improved support for running Windows applications under
3931 Wine. In particular, initial support for reading Windows .PDB debug
3932 information has been added.
3935 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
3936 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
3940 * The Valgrind client requests VALGRIND_PRINTF and
3941 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
3942 the string was always printed immediately on its own line. Now, the
3943 string will be added to a buffer but not printed until a newline is
3944 encountered, or other Valgrind output is printed (note that for
3945 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
3946 "other Valgrind output"). This allows you to use multiple
3947 VALGRIND_PRINTF calls to build up a single output line, and also to
3948 print multiple output lines with a single request (by embedding
3949 multiple newlines in the string).
3952 * The graphs drawn by Massif's ms_print program have changed slightly:
3954 - The half-height chars '.' and ',' are no longer drawn, because
3955 they are confusing. The --y option can be used if the default
3956 y-resolution is not high enough.
3958 - Horizontal lines are now drawn after the top of a snapshot if
3959 there is a gap until the next snapshot. This makes it clear that
3960 the memory usage has not dropped to zero between snapshots.
3963 * Something that happened in 3.4.0, but wasn't clearly announced: the
3964 option --read-var-info=yes can be used by some tools (Memcheck,
3965 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
3966 variable type and location information. This makes those tools
3967 start up more slowly and increases memory consumption, but
3968 descriptions of data addresses in error messages become more
3972 * exp-Omega, an experimental instantaneous leak-detecting tool, was
3973 disabled in 3.4.0 due to a lack of interest and maintenance,
3974 although the source code was still in the distribution. The source
3975 code has now been removed from the distribution. For anyone
3976 interested, the removal occurred in SVN revision r10247.
3979 * Some changes have been made to the build system.
3981 - VEX/ is now integrated properly into the build system. This means
3982 that dependency tracking within VEX/ now works properly, "make
3983 install" will work without requiring "make" before it, and
3984 parallel builds (ie. 'make -j') now work (previously a
3985 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
3986 was effectively ignored).
3988 - The --with-vex configure option has been removed. It was of
3989 little use and removing it simplified the build system.
3991 - The location of some install files has changed. This should not
3992 affect most users. Those who might be affected:
3994 * For people who use Valgrind with MPI programs, the installed
3995 libmpiwrap.so library has moved from
3996 $(INSTALL)/<platform>/libmpiwrap.so to
3997 $(INSTALL)/libmpiwrap-<platform>.so.
3999 * For people who distribute standalone Valgrind tools, the
4000 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
4001 have moved to $(INSTALL)/libcoregrind-<platform>.a.
4003 These changes simplify the build system.
4005 - Previously, all the distributed suppression (*.supp) files were
4006 installed. Now, only default.supp is installed. This should not
4007 affect users as the other installed suppression files were not
4008 read; the fact that they were installed was a mistake.
4011 * KNOWN LIMITATIONS:
4013 - Memcheck is unusable with the Intel compiler suite version 11.1,
4014 when it generates code for SSE2-and-above capable targets. This
4015 is because of icc's use of highly optimised inlined strlen
4016 implementations. It causes Memcheck to report huge numbers of
4017 false errors even in simple programs. Helgrind and DRD may also
4020 Versions 11.0 and earlier may be OK, but this has not been
4024 The following bugs have been fixed or resolved. Note that "n-i-bz"
4025 stands for "not in bugzilla" -- that is, a bug that was reported to us
4026 but never got a bugzilla entry. We encourage you to file bugs in
4027 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4028 mailing the developers (or mailing lists) directly -- bugs that are
4029 not entered into bugzilla tend to get forgotten about or ignored.
4031 To see details of a given bug, visit
4032 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4033 where XXXXXX is the bug number as listed below.
4035 84303 How about a LockCheck tool?
4036 91633 dereference of null ptr in vgPlain_st_basetype
4037 97452 Valgrind doesn't report any pthreads problems
4038 100628 leak-check gets assertion failure when using
4039 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4040 108528 NPTL pthread cleanup handlers not called
4041 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4042 110128 mallinfo is not implemented...
4043 110770 VEX: Generated files not always updated when making valgrind
4044 111102 Memcheck: problems with large (memory footprint) applications
4045 115673 Vex's decoder should never assert
4046 117564 False positive: Syscall param clone(child_tidptr) contains
4047 uninitialised byte(s)
4048 119404 executing ssh from inside valgrind fails
4049 133679 Callgrind does not write path names to sources with dwarf debug
4051 135847 configure.in problem with non gnu compilers (and possible fix)
4052 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4053 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4054 136230 memcheck reports "possibly lost", should be "still reachable"
4055 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4056 137904 Valgrind reports a memory leak when using POSIX threads,
4058 139076 valgrind VT_GETSTATE error
4059 142228 complaint of elf_dynamic_do_rela in trivial usage
4060 145347 spurious warning with USBDEVFS_REAPURB
4061 148441 (wine) can't find memory leak in Wine, win32 binary
4063 148742 Leak-check fails assert on exit
4064 149878 add (proper) check for calloc integer overflow
4065 150606 Call graph is broken when using callgrind control
4066 152393 leak errors produce an exit code of 0. I need some way to
4067 cause leak errors to result in a nonzero exit code.
4068 157154 documentation (leak-resolution doc speaks about num-callers
4069 def=4) + what is a loss record
4070 159501 incorrect handling of ALSA ioctls
4071 162020 Valgrinding an empty/zero-byte file crashes valgrind
4072 162482 ppc: Valgrind crashes while reading stabs information
4073 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4074 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4075 163560 VEX/test_main.c is missing from valgrind-3.3.1
4076 164353 malloc_usable_size() doesn't return a usable size
4077 165468 Inconsistent formatting in memcheck manual -- please fix
4078 169505 main.c:286 (endOfInstr):
4079 Assertion 'ii->cost_offset == *cost_offset' failed
4080 177206 Generate default.supp during compile instead of configure
4081 177209 Configure valt_load_address based on arch+os
4082 177305 eventfd / syscall 323 patch lost
4083 179731 Tests fail to build because of inlining of non-local asm labels
4084 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4085 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4086 181594 Bogus warning for empty text segment
4087 181707 dwarf doesn't require enumerations to have name
4088 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4089 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4090 Assertion '!already_present' failed.
4091 185359 exp-ptrcheck: unhandled syscall getresuid()
4092 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4093 185816 Valgrind is unable to handle debug info for files with split
4094 debug info that are prelinked afterwards
4095 185980 [darwin] unhandled syscall: sem_open
4096 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4097 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4098 186790 Suppression pattern used for leaks are not reported
4099 186796 Symbols with length>200 in suppression files are ignored
4100 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4101 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4102 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4103 188046 bashisms in the configure script
4104 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4105 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4106 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4107 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4108 assertion fail "!lock->heldBy"
4109 188427 Add support for epoll_create1 (with patch)
4110 188530 Support for SIOCGSTAMPNS
4111 188560 Include valgrind.spec in the tarball
4112 188572 Valgrind on Mac should suppress setenv() mem leak
4113 189054 Valgrind fails to build because of duplicate non-local asm labels
4114 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4115 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4116 189763 drd assertion failure: s_threadinfo[tid].is_recording
4117 190219 unhandled syscall: 328 (x86-linux)
4118 190391 dup of 181394; see above
4119 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4120 190820 No debug information on powerpc-linux
4121 191095 PATCH: Improve usbdevfs ioctl handling
4122 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4124 191189 --xml=yes should obey --gen-suppressions=all
4125 191192 syslog() needs a suppression on macosx
4126 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4127 191761 getrlimit on MacOSX
4128 191992 multiple --fn-skip only works sometimes; dependent on order
4129 192634 V. reports "aspacem sync_check_mapping_callback:
4130 segment mismatch" on Darwin
4131 192954 __extension__ missing on 2 client requests
4132 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4133 194474 "INSTALL" file has different build instructions than "README"
4134 194671 Unhandled syscall (sem_wait?) from mac valgrind
4135 195069 memcheck: reports leak (memory still reachable) for
4137 195169 drd: (vgDrd_barrier_post_wait):
4138 Assertion 'r->sg[p->post_iteration]' failed.
4139 195268 valgrind --log-file doesn't accept ~/...
4140 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4141 195860 WARNING: unhandled syscall: unix:223
4142 196528 need a error suppression for pthread_rwlock_init under os x?
4143 197227 Support aio_* syscalls on Darwin
4144 197456 valgrind should reject --suppressions=(directory)
4145 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4146 197591 unhandled syscall 27 (mincore)
4147 197793 Merge DCAS branch to the trunk == 85756, 142103
4148 197794 Avoid duplicate filenames in Vex
4149 197898 make check fails on current SVN
4150 197901 make check fails also under exp-ptrcheck in current SVN
4151 197929 Make --leak-resolution=high the default
4152 197930 Reduce spacing between leak reports
4153 197933 Print command line of client at start-up, and shorten preamble
4154 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4155 198395 add BBV to the distribution as an experimental tool
4156 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4157 198649 callgrind_annotate doesn't cumulate counters
4158 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4159 199977 Valgrind complains about an unrecognized instruction in the
4160 atomic_incs test program
4161 200029 valgrind isn't able to read Fedora 12 debuginfo
4162 200760 darwin unhandled syscall: unix:284
4163 200827 DRD doesn't work on Mac OS X
4164 200990 VG_(read_millisecond_timer)() does not work correctly
4165 201016 Valgrind does not support pthread_kill() on Mac OS
4166 201169 Document --read-var-info
4167 201323 Pre-3.5.0 performance sanity checking
4168 201384 Review user manual for the 3.5.0 release
4169 201585 mfpvr not implemented on ppc
4170 201708 tests failing because x86 direction flag is left set
4171 201757 Valgrind doesn't handle any recent sys_futex additions
4172 204377 64-bit valgrind can not start a shell script
4173 (with #!/path/to/shell) if the shell is a 32-bit executable
4174 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4175 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4176 about memory allocation events with memory access tracing enabled
4177 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4179 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4183 Release 3.4.1 (28 February 2009)
4184 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4185 3.4.1 is a bug-fix release that fixes some regressions and assertion
4186 failures in debug info reading in 3.4.0, most notably incorrect stack
4187 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4188 other debug info problems are also fixed. A number of bugs in the
4189 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4191 In view of the fact that 3.4.0 contains user-visible regressions
4192 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4193 encouraged to ship 3.4.1 in preference to 3.4.0.
4195 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4196 bugzilla" -- that is, a bug that was reported to us but never got a
4197 bugzilla entry. We encourage you to file bugs in bugzilla
4198 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4199 developers (or mailing lists) directly -- bugs that are not entered
4200 into bugzilla tend to get forgotten about or ignored.
4202 n-i-bz Fix various bugs reading icc-11 generated debug info
4203 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4204 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4205 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4206 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4207 179618 exp-ptrcheck crashed / exit prematurely
4208 179624 helgrind: false positive races with pthread_create and
4209 recv/open/close/read
4210 134207 pkg-config output contains @VG_PLATFORM@
4211 176926 floating point exception at valgrind startup with PPC 440EPX
4212 181594 Bogus warning for empty text segment
4213 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4214 181707 Dwarf3 doesn't require enumerations to have name
4215 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4216 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4217 Assertion '!already_present' failed.
4218 185359 exp-ptrcheck unhandled syscall getresuid()
4220 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4221 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4225 Release 3.4.0 (2 January 2009)
4226 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4227 3.4.0 is a feature release with many significant improvements and the
4228 usual collection of bug fixes. This release supports X86/Linux,
4229 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4230 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4232 3.4.0 brings some significant tool improvements. Memcheck can now
4233 report the origin of uninitialised values, the thread checkers
4234 Helgrind and DRD are much improved, and we have a new experimental
4235 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4236 global arrays. In detail:
4238 * Memcheck is now able to track the origin of uninitialised values.
4239 When it reports an uninitialised value error, it will try to show
4240 the origin of the value, as either a heap or stack allocation.
4241 Origin tracking is expensive and so is not enabled by default. To
4242 use it, specify --track-origins=yes. Memcheck's speed will be
4243 essentially halved, and memory usage will be significantly
4244 increased. Nevertheless it can drastically reduce the effort
4245 required to identify the root cause of uninitialised value errors,
4246 and so is often a programmer productivity win, despite running more
4249 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4250 3.4.0, will be released shortly.
4252 * Helgrind's race detection algorithm has been completely redesigned
4253 and reimplemented, to address usability and scalability concerns:
4255 - The new algorithm has a lower false-error rate: it is much less
4256 likely to report races that do not really exist.
4258 - Helgrind will display full call stacks for both accesses involved
4259 in a race. This makes it easier to identify the root causes of
4262 - Limitations on the size of program that can run have been removed.
4264 - Performance has been modestly improved, although that is very
4267 - Direct support for Qt4 threading has been added.
4269 - pthread_barriers are now directly supported.
4271 - Helgrind works well on all supported Linux targets.
4273 * The DRD thread debugging tool has seen major improvements:
4275 - Greatly improved performance and significantly reduced memory
4278 - Support for several major threading libraries (Boost.Thread, Qt4,
4279 glib, OpenMP) has been added.
4281 - Support for atomic instructions, POSIX semaphores, barriers and
4282 reader-writer locks has been added.
4284 - Works now on PowerPC CPUs too.
4286 - Added support for printing thread stack usage at thread exit time.
4288 - Added support for debugging lock contention.
4290 - Added a manual for Drd.
4292 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4293 checks for misuses of pointers. In that sense it is a bit like
4294 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4295 detect overruns of stack and global arrays, it can detect
4296 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4297 detect accesses heap blocks that have been freed a very long time
4298 ago (millions of blocks in the past).
4300 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4301 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4302 of the main Valgrind documentation. As this is an experimental
4303 tool, we would be particularly interested in hearing about your
4304 experiences with it.
4306 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4307 longer built by default, although the code remains in the repository
4308 and the tarball. This is due to three factors: a perceived lack of
4309 users, a lack of maintenance, and concerns that it may not be
4310 possible to achieve reliable operation using the existing design.
4312 * As usual, support for the latest Linux distros and toolchain
4313 components has been added. It should work well on Fedora Core 10,
4314 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4315 state) is supported, as is glibc-2.9. The C++ demangler has been
4316 updated so as to work well with C++ compiled by even the most recent
4319 * You can now use frame-level wildcards in suppressions. This was a
4320 frequently-requested enhancement. A line "..." in a suppression now
4321 matches zero or more frames. This makes it easier to write
4322 suppressions which are precise yet insensitive to changes in
4325 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4327 * Very basic support for IBM Power6 has been added (64-bit processes only).
4329 * Valgrind is now cross-compilable. For example, it is possible to
4330 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4331 on a ppc32/64-linux target.
4333 * You can set the main thread's stack size at startup using the
4334 new --main-stacksize= flag (subject of course to ulimit settings).
4335 This is useful for running apps that need a lot of stack space.
4337 * The limitation that you can't use --trace-children=yes together
4338 with --db-attach=yes has been removed.
4340 * The following bugs have been fixed. Note that "n-i-bz" stands for
4341 "not in bugzilla" -- that is, a bug that was reported to us but
4342 never got a bugzilla entry. We encourage you to file bugs in
4343 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4344 mailing the developers (or mailing lists) directly.
4346 n-i-bz Make return types for some client requests 64-bit clean
4347 n-i-bz glibc 2.9 support
4348 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4349 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4350 n-i-bz Building in an env without gdb gives bogus gdb attach
4351 92456 Tracing the origin of uninitialised memory
4352 106497 Valgrind does not demangle some C++ template symbols
4354 151612 Suppression with "..." (frame-level wildcards in .supp files)
4355 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4356 159285 unhandled syscall:25 (stime, on x86-linux)
4357 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4358 160954 ppc build of valgrind crashes with illegal instruction (isel)
4359 160956 mallinfo implementation, w/ patch
4360 162092 Valgrind fails to start gnome-system-monitor
4361 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4362 163794 assertion failure with "--track-origins=yes"
4363 163933 sigcontext.err and .trapno must be set together
4364 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4365 164476 Missing kernel module loading system calls
4366 164669 SVN regression: mmap() drops posix file locks
4367 166581 Callgrind output corruption when program forks
4368 167288 Patch file for missing system calls on Cell BE
4369 168943 unsupported scas instruction pentium
4370 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4371 172417 x86->IR: 0x82 ...
4372 172563 amd64->IR: 0xD9 0xF5 - fprem1
4373 173099 .lds linker script generation error
4374 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4375 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4377 174908 --log-file value not expanded correctly for core file
4378 175044 Add lookup_dcookie for amd64
4379 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4381 Developer-visible changes:
4383 * Valgrind's debug-info reading machinery has been majorly overhauled.
4384 It can now correctly establish the addresses for ELF data symbols,
4385 which is something that has never worked properly before now.
4387 Also, Valgrind can now read DWARF3 type and location information for
4388 stack and global variables. This makes it possible to use the
4389 framework to build tools that rely on knowing the type and locations
4390 of stack and global variables, for example exp-Ptrcheck.
4392 Reading of such information is disabled by default, because most
4393 tools don't need it, and because it is expensive in space and time.
4394 However, you can force Valgrind to read it, using the
4395 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4396 make use of such information, if present, to provide source-level
4397 descriptions of data addresses in the error messages they create.
4399 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4400 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).