1 Release 3.??.? (?? 202?)
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 =================
21 * ==================== TOOL CHANGES ===================
33 * =================== OTHER CHANGES ==================
35 * ==================== FIXED BUGS ====================
37 The following bugs have been fixed or resolved. Note that "n-i-bz"
38 stands for "not in bugzilla" -- that is, a bug that was reported to us
39 but never got a bugzilla entry. We encourage you to file bugs in
40 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
41 than mailing the developers (or mailing lists) directly -- bugs that
42 are not entered into bugzilla tend to get forgotten about or ignored.
44 444242 s390x: Valgrind crashes on EXRL with negative offset
45 444495 dhat/tests/copy fails on s390x
46 444571 PPC, fix the lxsibzx and lxsihzx so they only load their respective
48 444836 PPC, pstq instruction for R=1 is not storing to the correct address.
49 445032 valgrind/memcheck crash with SIGSEGV when SIGVTALRM timer used and
52 To see details of a given bug, visit
53 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
54 where XXXXXX is the bug number as listed below.
57 Release 3.18.0 (15 Oct 2021)
58 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
60 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
61 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
62 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
63 X86/Solaris, AMD64/Solaris, AMD64/MacOSX 10.12, X86/FreeBSD and
64 AMD64/FreeBSD. There is also preliminary support for X86/macOS 10.13,
65 AMD64/macOS 10.13 and nanoMIPS/Linux.
67 * ==================== CORE CHANGES ===================
69 * The libiberty demangler has been updated, which brings support for
70 Rust v0 name demangling
72 * __libc_freeres isn't called anymore after the program recieves a
73 fatal signal. Causing some internal glibc resources to hang around,
74 but preventing any crashes after the program has ended.
76 * The DWARF reader is now very much faster at startup when just
77 --read-inline-info=yes (the default in most cases) is given.
79 * glibc 2.34, which moved various functions from libpthread.so into
80 libc.so, is now supported.
82 * ================== PLATFORM CHANGES =================
86 - v8.2 scalar and vector FABD, FACGE, FACGT and FADD.
87 - v8.2 FP compare & conditional compare instructions.
88 - Zero variants of v8.2 FP compare instructions.
92 - Support the miscellaneous-instruction-extensions facility 3 and
93 the vector-enhancements facility 2. This enables programs
94 compiled with "-march=arch13" or "-march=z15" to be executed
99 - ISA 3.1 support is now complete
100 - ISA 3.0 support for the darn instruction added.
101 - ISA 3.0 support for the vector system call instruction scv added.
102 - ISA 3.0 support for the copy, paste and cpabort instructions added.
104 * Support for X86/FreeBSD and AMD64/FreeBSD has been added.
106 * ==================== OTHER CHANGES ====================
108 * Memcheck on amd64: minor fixes to remove some false positive
111 * ==================== FIXED BUGS ====================
113 The following bugs have been fixed or resolved. Note that "n-i-bz"
114 stands for "not in bugzilla" -- that is, a bug that was reported to us
115 but never got a bugzilla entry. We encourage you to file bugs in
116 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
117 than mailing the developers (or mailing lists) directly -- bugs that
118 are not entered into bugzilla tend to get forgotten about or ignored.
120 208531 [PATCH]: FreeBSD support for valgrind
121 368960 WARNING: unhandled amd64-linux syscall: 163 (acct)
122 407589 [Linux] Add support for C11 aligned_alloc() and GNU reallocarray()
123 423963 Error in child thread when CLONE_PIDFD is used
124 426148 crash with "impossible happened" when running BPF CO-RE programs
125 429375 PPC ISA 3.1 support is missing, part 9
126 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
127 431306 Update demangler to support Rust v0 name mangling
128 432387 s390x: z15 instructions support
129 433437 FreeBSD support, part 1
130 433438 FreeBSD support, part 2
131 433439 FreeBSD support, part 3
132 433469 FreeBSD support, part 4
133 433473 FreeBSD support, part 5
134 433477 FreeBSD support, part 6
135 433479 FreeBSD support, part 7
136 433504 FreeBSD support, part 8
137 433506 FreeBSD support, part 9
138 433507 FreeBSD support, part 10
139 433508 FreeBSD support, part 11
140 433510 FreeBSD support, part 12
141 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
142 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
143 434296 s390x: False-positive memcheck diagnostics from vector string
145 434840 PPC64 darn instruction not supported
146 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
147 435908 valgrind tries to fetch from deubginfod for files which already
148 have debug information
149 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2
150 439046 valgrind is unusably large when linked with lld
151 439090 Implement close_range(2)
152 439326 Valgrind 3.17.0 won't compile with Intel 2021 oneAPI compilers
153 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
154 440670 unhandled ppc64le-linux syscall: 252 statfs64 and 253 fstatfs64
155 440906 Fix impossible constraint issue in P10 testcase.
156 441512 Remove a unneeded / unnecessary prefix check.
157 441534 Update the expected output for test_isa_3_1_VRT.
158 442061 very slow execution under Fedora 34 (readdwarf3)
159 443031 Gcc -many change requires explicit .machine directives
160 443033 Add support for the ISA 3.0 mcrxrx instruction
161 443034 Sraw, srawi, srad, sradi, mfs
162 443178 Powerpc, test jm-mfspr expected output needs to be updated.
163 443179 Need new test for the lxvx and stxvx instructions on ISA 2.07 and
165 443180 The subnormal test and the ISA 3.0 test generate compiler warnings
166 443314 In the latest GIT version, Valgrind with "--trace-flags" crashes
168 443605 Don't call final_tidyup (__libc_freeres) on FatalSignal
170 To see details of a given bug, visit
171 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
172 where XXXXXX is the bug number as listed below.
174 (3.18.0.RC1: 12 Oct 2021)
175 (3.18.0: 15 Oct 2021)
179 Release 3.17.0 (19 Mar 2021)
180 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
182 3.17.0 fixes a number of bugs and adds some functional changes: support for
183 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
184 some new instructions for Arm64, S390 and POWER. There are also some tool
187 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
188 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
189 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
190 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
191 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
193 * ==================== CORE CHANGES ===================
195 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
198 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
199 debugging information. When a debuginfo file cannot be found locally,
200 Valgrind is able to query debuginfod servers for the file using its
201 build-id. See the user manual for more information about debuginfod support.
203 * ================== PLATFORM CHANGES =================
207 - Inaccuracies resulting from double-rounding in the simulation of
208 floating-point multiply-add/subtract instructions have been fixed. These
209 should now behave exactly as the hardware does.
211 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
212 ongoing. Support for the half-word variants of at least the following
213 instructions has been added:
215 FABS <Vd>.<T>, <Vn>.<T>
217 FNEG <Vd>.<T>, <Vn>.<T>
219 FSQRT <Vd>.<T>, <Vn>.<T>
224 - Implement the new instructions/features that were added to z/Architecture
225 with the vector-enhancements facility 1. Also cover the instructions from
226 the vector-packed-decimal facility that are defined outside the chapter
227 "Vector Decimal Instructions", but not the ones from that chapter itself.
229 For a detailed list of newly supported instructions see the updates to
230 `docs/internals/s390-opcodes.csv'.
232 Since the miscellaneous instruction extensions facility 2 was already
233 added in Valgrind 3.16.0, this completes the support necessary to run
234 general programs built with `--march=z14' under Valgrind. The
235 vector-packed-decimal facility is currently not exploited by the standard
236 toolchain and libraries.
240 - Various bug fixes. Fix for the sync field to limit setting just two of
241 the two bits in the L-field. Fix the write size for the stxsibx and
242 stxsihx instructions. Fix the modsw and modsd instructions.
244 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
245 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
246 conversion instructions are still missing.
248 * ==================== TOOL CHANGES ====================
250 * General tool changes
252 - All the tools and their vgpreload libraries are now installed under
253 libexec because they cannot be executed directly and should be run through
254 the valgrind executable. This should be an internal, not user visible,
255 change, but might impact valgrind packagers.
257 - The --track-fds option now respects -q, --quiet and won't output anything
258 if no file descriptors are leaked. It also won't report the standard stdin
259 (0), stdout (1) or stderr (2) descriptors as being leaked with
260 --trace-fds=yes anymore. To track whether the standard file descriptors
261 are still open at the end of the program run use --trace-fds=all.
265 - DHAT has been extended, with two new modes of operation. The new
266 --mode=copy flag triggers copy profiling, which records calls to memcpy,
267 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
268 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
269 the new dhat/dhat.h file. This is useful for learning more about hot code
270 paths. See the user manual for more information about the new modes.
272 - Because of these changes, DHAT's file format has changed. DHAT output
273 files produced with earlier versions of DHAT will not work with this
274 version of DHAT's viewer, and DHAT output files produced with this version
275 of DHAT will not work with earlier versions of DHAT's viewer.
277 * ==================== FIXED BUGS ====================
279 The following bugs have been fixed or resolved. Note that "n-i-bz"
280 stands for "not in bugzilla" -- that is, a bug that was reported to us
281 but never got a bugzilla entry. We encourage you to file bugs in
282 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
283 than mailing the developers (or mailing lists) directly -- bugs that
284 are not entered into bugzilla tend to get forgotten about or ignored.
286 To see details of a given bug, visit
287 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
288 where XXXXXX is the bug number as listed below.
290 140178 open("/proc/self/exe", ...); doesn't quite work
291 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
292 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
293 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
294 345077 linux syscall execveat support (linux 3.19)
295 361770 Missing F_ADD_SEALS
296 369029 handle linux syscalls sched_getattr and sched_setattr
297 384729 __libc_freeres inhibits cross-platform valgrind
298 388787 Support for C++17 new/delete
299 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
300 396656 Warnings while reading debug info
301 397605 ioctl FICLONE mishandled
302 401416 Compile failure with openmpi 4.0
303 408663 Suppression file for musl libc
304 404076 s390x: z14 vector instructions not implemented
305 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
306 (actually affected all x86 and nanomips regardless of host bitness)
307 413547 regression test does not check for Arm 64 features.
308 414268 Enable AArch64 feature detection and decoding for v8.x instructions
309 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
310 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
311 422261 platform selection fails for unqualified client name
312 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
313 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
314 423195 PPC ISA 3.1 support is missing, part 1
315 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
316 424012 crash with readv/writev having invalid but not NULL arg2 iovec
317 424298 amd64: Implement RDSEED
318 425232 PPC ISA 3.1 support is missing, part 2
319 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
320 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
321 426123 PPC ISA 3.1 support is missing, part 3
322 426144 Fix "condition variable has not been initialized" on Fedora 33.
323 427400 PPC ISA 3.1 support is missing, part 4
324 427401 PPC ISA 3.1 support is missing, part 5
325 427404 PPC ISA 3.1 support is missing, part 6
326 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
327 427787 Support new faccessat2 linux syscall (439)
328 427969 debuginfo section duplicates a section in the main ELF file
329 428035 drd: Unbreak the musl build
330 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
331 428716 cppcheck detects potential leak in VEX/useful/smchash.c
332 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
333 429352 PPC ISA 3.1 support is missing, part 7
334 429354 PPC ISA 3.1 support is missing, part 8
335 429692 unhandled ppc64le-linux syscall: 147 (getsid)
336 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
338 429952 Errors when building regtest with clang
339 430354 ppc stxsibx and stxsihx instructions write too much data
340 430429 valgrind.h doesn't compile on s390x with clang
341 430485 expr_is_guardable doesn't handle Iex_Qop
342 431556 Complete arm64 FADDP v8.2 instruction support
343 432102 Add support for DWARF5 as produced by GCC11
344 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
345 432381 drd: Process STACK_REGISTER client requests
346 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
347 432672 vg_regtest: test-specific environment variables not reset between tests
348 432809 VEX should support REX.W + POPF
349 432861 PPC modsw and modsd give incorrect results for 1 mod 12
350 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
351 432215 Add debuginfod functionality
352 433323 Use pkglibexecdir as vglibdir
353 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
354 433629 valgrind/README has type "abd" instead of "and"
355 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
356 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
357 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
358 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
359 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
361 (3.17.0.RC1: 13 Mar 2021)
362 (3.17.0.RC2: 17 Mar 2021)
363 (3.17.0: 19 Mar 2021)
367 Release 3.16.1 (22 June 2020)
368 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
370 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
371 fixes character encoding problems in the documentation HTML.
373 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
374 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
376 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
380 Release 3.16.0 (27 May 2020)
381 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
383 3.16.0 is a feature release with many improvements and the usual collection of
386 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
387 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
388 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
389 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
390 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
392 * ==================== CORE CHANGES ===================
394 * It is now possible to dynamically change the value of many command line
395 options while your program (or its children) are running under Valgrind.
397 To see the list of dynamically changeable options, run
398 "valgrind --help-dyn-options".
400 You can change the options from the shell by using vgdb to launch
401 the monitor command "v.clo <clo option>...".
402 The same monitor command can be used from a gdb connected
403 to the valgrind gdbserver.
404 Your program can also change the dynamically changeable options using
405 the client request VALGRIND_CLO_CHANGE(option).
407 * ================== PLATFORM CHANGES =================
409 * MIPS: preliminary support for nanoMIPS instruction set has been added.
411 * ==================== TOOL CHANGES ====================
415 - The implicit memcpy done by each call to realloc now counts towards the
416 read and write counts of resized heap blocks, making those counts higher
421 - cg_annotate's --auto and --show-percs options now default to 'yes', because
422 they are usually wanted.
426 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
427 because they are usually wanted.
429 - The command option --collect-systime has been enhanced to specify
430 the unit used to record the elapsed time spent during system calls.
431 The command option now accepts the values no|yes|msec|usec|nsec,
432 where yes is a synonym of msec. When giving the value nsec, the
433 system cpu time of system calls is also recorded.
437 - Several memcheck options are now dynamically changeable.
438 Use valgrind --help-dyn-options to list them.
440 - The release 3.15 introduced a backward incompatible change for
441 some suppression entries related to preadv and pwritev syscalls.
442 When reading a suppression entry using the unsupported 3.14 format,
443 valgrind will now produce a warning to say the suppression entry will not
444 work, and suggest the needed change.
446 - Significantly fewer false positive errors on optimised code generated by
447 Clang and GCC. In particular, Memcheck now deals better with the
448 situation where the compiler will transform C-level "A && B" into "B && A"
449 under certain circumstances (in which the transformation is valid).
450 Handling of integer equality/non-equality checks on partially defined
451 values is also improved on some architectures.
455 - The exprimental Stack and Global Array Checking tool has been removed.
456 It only ever worked on x86 and amd64, and even on those it had a
457 high false positive rate and was slow. An alternative for detecting
458 stack and global array overruns is using the AddressSanitizer (ASAN)
459 facility of the GCC and Clang compilers, which require you to rebuild
460 your code with -fsanitize=address.
462 * ==================== OTHER CHANGES ====================
464 * New and modified GDB server monitor features:
466 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
468 - The gdbserver monitor commands that require an address and an optional
469 length argument now accepts the alternate 'C like' syntax "address[length]".
470 For example, the memcheck command "monitor who_points_at 0x12345678 120"
471 can now also be given as "monitor who_points_at 0x12345678[120]".
473 * ==================== FIXED BUGS ====================
475 The following bugs have been fixed or resolved. Note that "n-i-bz"
476 stands for "not in bugzilla" -- that is, a bug that was reported to us
477 but never got a bugzilla entry. We encourage you to file bugs in
478 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
479 than mailing the developers (or mailing lists) directly -- bugs that
480 are not entered into bugzilla tend to get forgotten about or ignored.
482 To see details of a given bug, visit
483 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
484 where XXXXXX is the bug number as listed below.
486 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
487 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
488 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
489 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
490 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
491 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
492 400872 Add nanoMIPS support to Valgrind
493 403212 drd/tests/trylock hangs on FreeBSD
494 404406 s390x: z14 miscellaneous instructions not implemented
495 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
496 406561 mcinfcallWSRU gdbserver_test fails on ppc64
497 406824 Unsupported baseline
498 407218 Add support for the copy_file_range syscall
499 407307 Intercept stpcpy also in ld.so for arm64
500 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
502 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
503 408009 Expose rdrand and f16c even on avx if host cpu supports them
504 408091 Missing pkey syscalls
505 408414 Add support for missing for preadv2 and pwritev2 syscalls
506 409141 Valgrind hangs when SIGKILLed
507 409206 Support for Linux PPS and PTP ioctls
508 409367 exit_group() after signal to thread waiting in futex() causes hangs
509 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
510 409780 References to non-existent configure.in
511 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
512 410599 Non-deterministic behaviour of pth_self_kill_15_other test
513 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
514 411134 Allow the user to change a set of command line options during execution
515 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
516 412344 Problem setting mips flags with specific paths
517 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
518 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
519 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
520 413603 callgrind_annotate/cg_annotate truncate function names at '#'
521 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
522 415136 ARMv8.1 Compare-and-Swap instructions are not supported
523 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
524 416239 valgrind crashes when handling clock_adjtime
525 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
526 416286 DRD reports "conflicting load" error on std::mutex::lock()
527 416301 s390x: "compare and signal" not supported
528 416387 finit_module and bpf syscalls are unhandled on arm64
529 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
530 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
531 416753 new 32bit time syscalls for 2038+
532 417075 pwritev(vector[...]) suppression ignored
533 417075 is not fixed, but incompatible supp entries are detected
534 and a warning is produced for these.
535 417187 [MIPS] Conditional branch problem since 'grail' changes
536 417238 Test memcheck/tests/vbit-test fails on mips64 BE
537 417266 Make memcheck/tests/linux/sigqueue usable with musl
538 417281 s390x: /bin/true segfaults with "grail" enabled
539 417427 commit to fix vki_siginfo_t definition created numerous regression
541 417452 s390_insn_store_emit: dst->tag for HRcVec128
542 417578 Add suppressions for glibc DTV leaks
543 417906 clone with CLONE_VFORK and no CLONE_VM fails
544 418004 Grail code additions break ppc64.
545 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
546 418997 s390x: Support Iex_ITE for float and vector types
547 419503 s390x: Avoid modifying registers returned from isel functions
548 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
549 421570 std_mutex fails on Arm v8.1 h/w
550 434035 vgdb might crash if valgrind is killed
551 n-i-bz Fix minor one time leaks in dhat.
552 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
553 n-i-bz Add support for the Linux io_uring system calls
554 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
555 n-i-bz Fix non-glibc build of test suite with s390x_features
556 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
557 423195 PPC ISA 3.1 support is missing, part 1
559 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
560 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
561 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
565 Release 3.15.0 (12 April 2019)
566 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
568 3.15.0 is a feature release with many improvements and the usual collection of
571 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
572 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
573 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
574 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
575 support for X86/macOS 10.13 and AMD64/macOS 10.13.
577 * ==================== CORE CHANGES ===================
579 * The XTree Massif output format now makes use of the information obtained
580 when specifying --read-inline-info=yes.
582 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
584 * ==================== TOOL CHANGES ====================
588 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
589 it has been promoted from an experimental tool to a regular tool. Run it
590 with --tool=dhat instead of --tool=exp-dhat.
592 - DHAT now prints only minimal data when the program ends, instead writing
593 the bulk of the profiling data to a file. As a result, the --show-top-n
594 and --sort-by options have been removed.
596 - Profile results can be viewed with the new viewer, dh_view.html. When
597 a run ends, a short message is printed, explaining how to view the result.
599 - See the documentation for more details.
603 - cg_annotate has a new option, --show-percs, which prints percentages next
608 - callgrind_annotate has a new option, --show-percs, which prints percentages
609 next to all event counts.
611 - callgrind_annotate now inserts commas in call counts, and
612 sort the caller/callee lists in the call tree.
616 - The default value for --read-inline-info is now "yes" on
617 Linux/Android/Solaris. It is still "no" on other OS.
621 - The option --xtree-leak=yes (to output leak result in xtree format)
622 automatically activates the option --show-leak-kinds=all, as xtree
623 visualisation tools such as kcachegrind can in any case select what kind
624 of leak to visualise.
626 - There has been further work to avoid false positives. In particular,
627 integer equality on partially defined inputs (C == and !=) is now handled
630 * ==================== OTHER CHANGES ====================
632 * The new option --show-error-list=no|yes displays, at the end of the run, the
633 list of detected errors and the used suppressions. Prior to this change,
634 showing this information could only be done by specifying "-v -v", but that
635 also produced a lot of other possibly-non-useful messages. The option -s is
636 equivalent to --show-error-list=yes.
638 * ==================== FIXED BUGS ====================
640 The following bugs have been fixed or resolved. Note that "n-i-bz"
641 stands for "not in bugzilla" -- that is, a bug that was reported to us
642 but never got a bugzilla entry. We encourage you to file bugs in
643 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
644 than mailing the developers (or mailing lists) directly -- bugs that
645 are not entered into bugzilla tend to get forgotten about or ignored.
647 To see details of a given bug, visit
648 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
649 where XXXXXX is the bug number as listed below.
651 385411 s390x: z13 vector floating-point instructions not implemented
652 397187 z13 vector register support for vgdb gdbserver
653 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
654 398870 Please add support for instruction vcvtps2ph
655 399287 amd64 front end: Illegal Instruction vcmptrueps
656 399301 Use inlined frames in Massif XTree output.
657 399322 Improve callgrind_annotate output
658 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
659 400164 helgrind test encounters mips x-compiler warnings and assembler error
660 400490 s390x: VRs allocated as if separate from FPRs
661 400491 s390x: Operand of LOCH treated as unsigned integer
662 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
663 options, which specify a mips64 processor
664 401112 LLVM 5.0 generates comparison against partially initialized data
665 401277 More bugs in z13 support
666 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
667 401578 drd: crashes sometimes on fork()
668 401627 memcheck errors with glibc avx2 optimized wcsncmp
669 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
670 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
671 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
673 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
674 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
675 402123 invalid assembler opcodes for mips32r2
676 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
677 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
678 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
679 402351 mips64 libvexmultiarch_test fails on s390x
681 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
682 402480 Do not use %rsp in clobber list
683 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
684 402515 Implement new option --show-error-list=no|yes / -s
685 402519 POWER 3.0 addex instruction incorrectly implemented
686 402781 Redo the cache used to process indirect branch targets
687 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
688 403552 s390x: wrong facility bit checked for vector facility
689 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
690 404638 Add VG_(replaceIndexXA)
691 404843 s390x: backtrace sometimes ends prematurely
692 404888 autotools cleanup series
693 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
694 405182 Valgrind fails to build with Clang
695 405205 filter_libc: remove the line holding the futex syscall error entirely
696 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
697 the upper and lower 32-bits of the 64-bit result
698 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
699 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
700 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
702 405403 s390x disassembler cannot be used on x86
703 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
704 405458 MIPS mkFormVEC arguments swapped?
705 405716 drd: Fix an integer overflow in the stack margin calculation
706 405722 Support arm64 core dump
707 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
708 of the 64-bit destination field.
709 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
710 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
711 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
713 406256 PPC64, vector floating point instructions don't handle subnormal
714 according to VSCR[NJ] bit setting.
715 406352 cachegrind/callgrind fails ann tests because of missing a.c
716 406354 dhat is broken on x86 (32bit)
717 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
718 406357 gdbserver_tests fails because of gdb output change
719 406360 memcheck/tests/libstdc++.supp needs more supression variants
720 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
721 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
722 407340 PPC64, does not support the vlogefp, vexptefp instructions.
723 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
724 n-i-bz Fix callgrind_annotate non deterministic order for equal total
725 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
726 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
727 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
729 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
730 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
731 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
735 Release 3.14.0 (9 October 2018)
736 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
738 3.14.0 is a feature release with many improvements and the usual collection of
741 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
742 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
743 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
744 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
745 support for X86/macOS 10.13, AMD64/macOS 10.13.
747 * ==================== CORE CHANGES ===================
749 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
750 debug info for unloaded code. This allows saved stack traces (e.g. for
751 memory leaks) to include file/line info for code that has been dlclose'd (or
752 similar). See the user manual for more information and known limitations.
754 * Ability to specify suppressions based on source file name and line number.
756 * Majorly overhauled register allocator. No end-user changes, but the JIT
757 generates code a bit more quickly now.
759 * ================== PLATFORM CHANGES =================
761 * Preliminary support for macOS 10.13 has been added.
763 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
765 * mips: support for MIPS SIMD architecture (MSA) has been added.
767 * mips: support for MIPS N32 ABI has been added.
769 * s390: partial support for vector instructions (integer and string) has been
772 * ==================== TOOL CHANGES ====================
774 * Helgrind: Addition of a flag
775 --delta-stacktrace=no|yes [yes on linux amd64/x86]
776 which specifies how full history stack traces should be computed.
777 Setting this to =yes can speed up Helgrind by 25% when using
778 --history-level=full.
780 * Memcheck: reduced false positive rate for optimised code created by Clang 6
781 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
782 blocks more carefully to determine where it can avoid expensive definedness
783 checks without loss of precision. This is controlled by the flag
784 --expensive-definedness-checks=no|auto|yes [auto].
786 * ==================== OTHER CHANGES ====================
788 * Valgrind is now buildable with link-time optimisation (LTO). A new
789 configure option --enable-lto=yes allows building Valgrind with LTO. If the
790 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
791 Note that if you are doing Valgrind development, --enable-lto=yes massively
792 slows down the build process.
794 * ==================== FIXED BUGS ====================
796 The following bugs have been fixed or resolved. Note that "n-i-bz"
797 stands for "not in bugzilla" -- that is, a bug that was reported to us
798 but never got a bugzilla entry. We encourage you to file bugs in
799 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
800 than mailing the developers (or mailing lists) directly -- bugs that
801 are not entered into bugzilla tend to get forgotten about or ignored.
803 To see details of a given bug, visit
804 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
805 where XXXXXX is the bug number as listed below.
807 79362 Debug info is lost for .so files when they are dlclose'd
808 208052 strlcpy error when n = 0
809 255603 exp-sgcheck Assertion '!already_present' failed
810 338252 building valgrind with -flto (link time optimisation) fails
811 345763 MIPS N32 ABI support
812 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
813 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
814 372347 Replacement problem of the additional c++14/c++17 new/delete operators
815 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
816 376257 helgrind history full speed up using a cached stack
817 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
819 379748 Fix missing pselect syscall (OS X 10.11)
820 379754 Fix missing syscall ulock_wait (OS X 10.12)
821 380397 s390x: __GI_strcspn() replacemenet needed
822 381162 possible array overrun in VEX register allocator
823 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
824 381274 powerpc too chatty even with --sigill-diagnostics=no
825 381289 epoll_pwait can have a NULL sigmask
826 381553 VEX register allocator v3
827 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
828 381769 Use ucontext_t instead of struct ucontext
829 381805 arm32 needs ld.so index hardwire for new glibc security fixes
830 382256 gz compiler flag test doesn't work for gold
831 382407 vg_perf needs "--terse" command line option
832 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
833 382563 MIPS MSA ASE support
834 382998 xml-socket doesn't work
835 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
836 383723 Fix missing kevent_qos syscall (macOS 10.11)
837 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
838 384096 Mention AddrCheck at Memcheck's command line option [..]
839 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
840 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
841 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
842 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
843 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
844 384337 performance improvements to VEX register allocator v2 and v3
845 384526 reduce number of spill insns generated by VEX register allocator v3
846 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
847 384631 Sanitise client args as printed with -v
848 384633 Add a simple progress-reporting facility
849 384987 VEX regalloc: allocate caller-save registers for short lived vregs
850 385055 PPC VEX temporary storage exhausted
851 385182 PPC64 is missing support for the DSCR
852 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
853 385207 PPC64, generate_store_FPRF() generates too many Iops
854 385208 PPC64, xxperm instruction exhausts temporary memory
855 385210 PPC64, vpermr instruction could exhaust temporary memory
856 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
857 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
858 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
859 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
860 385408 s390x: z13 vector "support" instructions not implemented
861 385409 s390x: z13 vector integer instructions not implemented
862 385410 s390x: z13 vector string instructions not implemented
863 385412 s390x: new non-vector z13 instructions not implemented
864 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
865 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
866 385939 Optionally exit on the first error
867 386318 valgrind.org/info/tools.html is missing SGCheck
868 386425 running valgrind + wine on armv7l gives illegal opcode
869 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
870 387410 MIPSr6 support
871 387664 Memcheck: make expensive-definedness-checks be the default
872 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
873 387766 asm shifts cause false positive "Conditional jump or move depends
874 on uninitialised value"
875 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
876 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
877 388786 Support bpf syscall in amd64 Linux
878 388862 Add replacements for wmemchr and wcsnlen on Linux
879 389065 valgrind meets gcc flag -Wlogical-op
880 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
881 390471 suppression by specification of source-file line number
882 390723 make xtree dump files world wide readable, similar to log files
883 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
884 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
885 392118 unhandled amd64-linux syscall: 332 (statx)
886 392449 callgrind not clearing the number of calls properly
887 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
888 lxssp, stxssp and stxvl instructions.
889 393023 callgrind_control risks using the wrong vgdb
890 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
891 393099 posix_memalign() invalid write if alignment == 0
892 393146 failing assert "is_DebugInfo_active(di)"
893 395709 PPC64 is missing support for the xvnegsp instruction
894 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
896 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
897 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
898 396839 s390x: Trap instructions not implemented
899 396887 arch_prctl should return EINVAL on unknown option
900 == 397286 crash before launching binary (Unsupported arch_prctl option)
901 == 397393 valgrind: the 'impossible' happened: (Archlinux)
902 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
903 396906 compile tests failure on mips32-linux: broken inline asm in tests on
905 397012 glibc ld.so uses arch_prctl on i386
906 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
907 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
908 397424 glibc 2.27 and gdb_server tests
909 398028 Assertion `cfsi_fits` failing in simple C program
910 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
912 n-i-bz Fix missing workq_ops operations (macOS)
913 n-i-bz fix bug in strspn replacement
914 n-i-bz Add support for the Linux BLKFLSBUF ioctl
915 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
916 n-i-bz Fix possible stack trashing by semctl syscall wrapping
917 n-i-bz Add support for the Linux membarrier() system call
918 n-i-bz x86 front end: recognise and handle UD2 correctly
919 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
920 correctly aligned before entering the handler.
922 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
923 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
924 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
928 Release 3.13.0 (15 June 2017)
929 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
931 3.13.0 is a feature release with many improvements and the usual collection of
934 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
935 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
936 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
937 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
939 * ==================== CORE CHANGES ===================
941 * The translation cache size has been increased to keep up with the demands of
942 large applications. The maximum number of sectors has increased from 24 to
943 48. The default number of sectors has increased from 16 to 32 on all
944 targets except Android, where the increase is from 6 to 12.
946 * The amount of memory that Valgrind can use has been increased from 64GB to
947 128GB. In particular this means your application can allocate up to about
948 60GB when running on Memcheck.
950 * Valgrind's default load address has been changed from 0x3800'0000 to
951 0x5800'0000, so as to make it possible to load larger executables. This
952 should make it possible to load executables of size at least 1200MB.
954 * A massive spaceleak caused by reading compressed debuginfo files has been
955 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
958 * The C++ demangler has been updated.
960 * Support for demangling Rust symbols has been added.
962 * A new representation of stack traces, the "XTree", has been added. An XTree
963 is a tree of stacktraces with data associated with the stacktraces. This is
964 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
965 consumption of your program. Reporting is controlled by the new options
966 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
968 A report can also be produced on demand using the gdbserver monitor command
969 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
970 format' and 'massif format. The existing visualisers for these formats (e.g.
971 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
972 analyse these reports.
974 Memcheck can also produce XTree leak reports using the Callgrind file
975 format. For more details, see the user manual.
977 * ================== PLATFORM CHANGES =================
979 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
981 * amd64: fixes for JIT failure problems on long AVX2 code blocks
983 * amd64 and x86: support for CET prefixes has been added
985 * arm32: a few missing ARMv8 instructions have been implemented
987 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
988 Store-Conditional instructions has been added. This is to deal with
989 processor implementations that implement the LL/SC specifications strictly
990 and as a result cause Valgrind to hang in certain situations. The
991 alternative implementation is automatically enabled at startup, as required.
992 You can use the option --sim-hints=fallback-llsc to force-enable it if you
995 * Support for OSX 10.12 has been improved.
997 * On Linux, clone handling has been improved to honour CLONE_VFORK that
998 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
999 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
1000 CLONE_VM exact semantics will (still) not work.
1002 * The TileGX/Linux port has been removed because it appears to be both unused
1005 * ==================== TOOL CHANGES ====================
1009 - Memcheck should give fewer false positives when running optimised
1010 Clang/LLVM generated code.
1012 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1014 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
1015 to produce the end of execution leak report in a xtree callgrind format
1018 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
1019 the leak report in an xtree file.
1023 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1025 - For some workloads (typically, for big applications), Massif memory
1026 consumption and CPU consumption has decreased significantly.
1030 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
1032 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
1033 for Ada gnat compiled applications.
1035 * ==================== OTHER CHANGES ====================
1037 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
1038 append the inner guest stacktrace to the inner host stacktrace. This helps
1039 to investigate the errors reported by the outer, when they are caused by the
1040 inner guest program (such as an inner regtest). See README_DEVELOPERS for
1043 * To allow fast detection of callgrind files by desktop environments and file
1044 managers, the format was extended to have an optional first line that
1045 uniquely identifies the format ("# callgrind format"). Callgrind creates
1046 this line now, as does the new xtree functionality.
1048 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
1049 have a new %n format letter that is replaced by a sequence number.
1051 * "--version -v" now shows the SVN revision numbers from which Valgrind was
1054 * ==================== FIXED BUGS ====================
1056 The following bugs have been fixed or resolved. Note that "n-i-bz"
1057 stands for "not in bugzilla" -- that is, a bug that was reported to us
1058 but never got a bugzilla entry. We encourage you to file bugs in
1059 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1060 than mailing the developers (or mailing lists) directly -- bugs that
1061 are not entered into bugzilla tend to get forgotten about or ignored.
1063 To see details of a given bug, visit
1064 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1065 where XXXXXX is the bug number as listed below.
1067 162848 --log-file output isn't split when a program forks
1068 340777 Illegal instruction on mips (ar71xx)
1069 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
1070 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
1071 to a different stack.
1072 344139 x86 stack-seg overrides, needed by the Wine people
1073 344524 store conditional of guest applications always fail - observed on
1075 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
1076 352395 Please provide SVN revision info in --version -v
1077 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
1078 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
1079 INVALID_POSIX_THREADID' failed
1080 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
1081 barrier implementation
1082 358697 valgrind.h: Some code remains even when defining NVALGRIND
1083 359202 Add musl libc configure/compile
1084 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
1085 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
1086 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
1087 362223 assertion failed when .valgrindrc is a directory instead of a file
1088 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
1089 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
1090 368507 can't malloc chunks larger than about 34GB
1091 368529 Android arm target link error, missing atexit and pthread_atfork
1092 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
1093 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
1094 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
1095 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
1096 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
1097 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
1098 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
1099 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
1100 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
1101 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
1102 370028 Reduce the number of compiler warnings on MIPS platforms
1103 370635 arm64 missing syscall getcpu
1104 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
1105 371227 Clean AArch64 syscall table
1106 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
1107 371471 Valgrind complains about non legit memory leaks on placement new (C++)
1108 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
1109 371503 disInstr(arm64): unhandled instruction 0xF89F0000
1110 371869 support '%' in symbol Z-encoding
1111 371916 execution tree xtree concept
1112 372120 c++ demangler demangles symbols which are not c++
1113 372185 Support of valgrind on ARMv8 with 32 bit executable
1114 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1115 372195 Power PC, xxsel instruction is not always recognized.
1116 372504 Hanging on exit_group
1117 372600 process loops forever when fatal signals are arriving quickly
1118 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1119 373046 Stacks registered by core are never deregistered
1120 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1121 373086 Implement additional Xen hypercalls
1122 373192 Calling posix_spawn in glibc 2.24 completely broken
1123 373488 Support for fanotify API on ARM64 architecture
1124 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1125 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1126 373938 const IRExpr arguments for matchIRExpr()
1127 374719 some spelling fixes
1128 374963 increase valgrind's load address to prevent mmap failure
1129 375514 valgrind_get_tls_addr() does not work in case of static TLS
1130 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1131 for ML_(find_rx_mapping)()
1132 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1133 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1134 == 377159 "vex: the `impossible' happened" still present
1135 == 375150 Assertion 'tres.status == VexTransOK' failed
1136 == 378068 valgrind crashes on AVX2 function in FFmpeg
1137 376142 Segfaults on MIPS Cavium Octeon boards
1138 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1139 376455 Solaris: unhandled syscall lgrpsys(180)
1140 376518 Solaris: unhandled fast trap getlgrp(6)
1141 376611 ppc64 and arm64 don't know about prlimit64 syscall
1142 376729 PPC64, remove R2 from the clobber list
1144 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1145 to be wrongly marked as addressable
1146 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1147 PIE enabled by default
1148 377376 memcheck/tests/linux/getregset fails with glibc2.24
1149 377427 PPC64, lxv instruction failing on odd destination register
1150 377478 PPC64: ISA 3.0 setup fixes
1151 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1152 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1153 and 2 args for FUTEX_TRYLOCK_PI
1154 377717 Fix massive space leak when reading compressed debuginfo sections
1155 377891 Update Xen 4.6 domctl wrappers
1156 377930 fcntl syscall wrapper is missing flock structure check
1157 378524 libvexmultiarch_test regression on s390x and ppc64
1158 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1159 378673 Update libiberty demangler
1160 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1161 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1162 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1163 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1164 (task_register_dyld_image_infos)
1165 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1166 (task_register_dyld_shared_cache_image_info)
1167 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1168 379473 MIPS: add support for rdhwr cycle counter register
1169 379504 remove TileGX/Linux port
1170 379525 Support more x86 nop opcodes
1171 379838 disAMode(x86): not an addr!
1172 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1173 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1174 379895 clock_gettime does not execute POST syscall wrapper
1175 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1176 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1177 380200 xtree generated callgrind files refer to files without directory name
1178 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1179 380397 s390x: __GI_strcspn() replacement needed
1180 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1182 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1183 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1184 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1188 Release 3.12.0 (20 October 2016)
1189 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1191 3.12.0 is a feature release with many improvements and the usual
1192 collection of bug fixes.
1194 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1195 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1196 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1197 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1198 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1199 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1201 * ================== PLATFORM CHANGES =================
1203 * POWER: Support for ISA 3.0 has been added
1205 * mips: support for O32 FPXX ABI has been added.
1206 * mips: improved recognition of different processors
1207 * mips: determination of page size now done at run time
1209 * amd64: Partial support for AMD FMA4 instructions.
1211 * arm, arm64: Support for v8 crypto and CRC instructions.
1213 * Improvements and robustification of the Solaris port.
1215 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1217 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1218 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1219 where possible. Valgrind's support for 32-bit x86 has stagnated in
1220 recent years and has fallen far behind that for 64-bit x86
1221 instructions. By contrast 64-bit x86 is well supported, up to and
1224 * ==================== TOOL CHANGES ====================
1228 - Added meta mempool support for describing a custom allocator which:
1229 - Auto-frees all chunks assuming that destroying a pool destroys all
1231 - Uses itself to allocate other memory blocks
1233 - New flag --ignore-range-below-sp to ignore memory accesses below
1234 the stack pointer, if you really have to. The related flag
1235 --workaround-gcc296-bugs=yes is now deprecated. Use
1236 --ignore-range-below-sp=1024-1 as a replacement.
1240 - Improved thread startup time significantly on non-Linux platforms.
1244 - Added collection of the metric "tot-blocks-allocd"
1246 * ==================== OTHER CHANGES ====================
1248 * Replacement/wrapping of malloc/new related functions is now done not just
1249 for system libraries by default, but for any globally defined malloc/new
1250 related function (both in shared libraries and statically linked alternative
1251 malloc implementations). The dynamic (runtime) linker is excluded, though.
1252 To only intercept malloc/new related functions in
1253 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1254 "nouserintercepts" can be any non-existing library name).
1255 This new functionality is not implemented for MacOS X.
1257 * The maximum number of callers in a suppression entry is now equal to
1258 the maximum size for --num-callers (500).
1259 Note that --gen-suppressions=yes|all similarly generates suppressions
1260 containing up to --num-callers frames.
1262 * New and modified GDB server monitor features:
1264 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1265 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1268 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1269 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1272 * Valgrind is able to read compressed debuginfo sections in two formats:
1273 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1274 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1276 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1277 for the most common use case (x86_64-linux, Memcheck) has been
1280 * Improved performance for programs that do a lot of discarding of
1281 instruction address ranges of 8KB or less.
1283 * The C++ symbol demangler has been updated.
1285 * More robustness against invalid syscall parameters on Linux.
1287 * ==================== FIXED BUGS ====================
1289 The following bugs have been fixed or resolved. Note that "n-i-bz"
1290 stands for "not in bugzilla" -- that is, a bug that was reported to us
1291 but never got a bugzilla entry. We encourage you to file bugs in
1292 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1293 than mailing the developers (or mailing lists) directly -- bugs that
1294 are not entered into bugzilla tend to get forgotten about or ignored.
1296 To see details of a given bug, visit
1297 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1298 where XXXXXX is the bug number as listed below.
1300 191069 Exiting due to signal not reported in XML output
1301 199468 Suppressions: stack size limited to 25
1302 while --num-callers allows more frames
1303 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1304 278744 cvtps2pd with redundant RexW
1305 303877 valgrind doesn't support compressed debuginfo sections.
1306 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1307 348345 Assertion fails for negative lineno
1308 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1309 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1310 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1311 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1312 352197 mips: mmap2() not wrapped correctly for page size > 4096
1313 353083 arm64 doesn't implement various xattr system calls
1314 353084 arm64 doesn't support sigpending system call
1315 353137 www: update info for Supported Platforms
1316 353138 www: update "The Valgrind Developers" page
1317 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1320 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1321 353398 WARNING: unhandled amd64-solaris syscall: 207
1322 353660 XML in auxwhat tag not escaping reserved symbols properly
1323 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1324 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1325 353802 ELF debug info reader confused with multiple .rodata sections
1326 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1327 353917 unhandled amd64-solaris syscall fchdir(120)
1328 353920 unhandled amd64-solaris syscall: 170
1329 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1330 354392 unhandled amd64-solaris syscall: 171
1331 354797 Vbit test does not include Iops for Power 8 instruction support
1332 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1336 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1337 355188 valgrind should intercept all malloc related global functions
1338 355454 do not intercept malloc related symbols from the runtime linker
1339 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1340 356044 Dwarf line info reader misinterprets is_stmt register
1341 356112 mips: replace addi with addiu
1342 356393 valgrind (vex) crashes because isZeroU happened
1345 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1346 356678 arm64-linux: unhandled syscall 232 (mincore)
1347 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1348 356823 Unsupported ARM instruction: stlex
1349 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1350 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1351 357673 crash if I try to run valgrind with a binary link with libcurl
1352 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1353 357871 pthread_spin_destroy not properly wrapped
1354 357887 Calls to VG_(fclose) do not close the file descriptor
1355 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1356 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1357 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1358 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1359 359181 Buffer Overflow during Demangling
1360 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1361 359289 s390x: popcnt (B9E1) not implemented
1362 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1363 359503 Add missing syscalls for aarch64 (arm64)
1364 359645 "You need libc6-dbg" help message could be more helpful
1365 359703 s390: wire up separate socketcalls system calls
1366 359724 getsockname might crash - deref_UInt should call safe_to_deref
1367 359733 amd64 implement ld.so strchr/index override like x86
1368 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1369 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1371 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1372 359871 Incorrect mask handling in ppoll
1373 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1374 360008 Contents of Power vr registers contents is not printed correctly when
1375 the --vgdb-shadow-registers=yes option is used
1376 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1377 non-zero shadow bits
1378 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1379 360425 arm64 unsupported instruction ldpsw
1381 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1382 360571 Error about the Android Runtime reading below the stack pointer on ARM
1383 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1384 360749 kludge for multiple .rodata sections on Solaris no longer needed
1385 360752 raise the number of reserved fds in m_main.c from 10 to 12
1386 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1387 361226 s390x: risbgn (EC59) not implemented
1388 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1389 361354 ppc64[le]: wire up separate socketcalls system calls
1390 361615 Inconsistent termination for multithreaded process terminated by signal
1391 361926 Unhandled Solaris syscall: sysfs(84)
1392 362009 V dumps core on unimplemented functionality before threads are created
1393 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1394 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1395 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1396 362953 Request for an update to the Valgrind Developers page
1397 363680 add renameat2() support
1398 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1399 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1400 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1401 364058 clarify in manual limitations of array overruns detections
1402 364413 pselect sycallwrapper mishandles NULL sigmask
1403 364728 Power PC, missing support for several HW registers in
1404 get_otrack_shadow_offset_wrk()
1405 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1406 365273 Invalid write to stack location reported after signal handler runs
1407 365912 ppc64BE segfault during jm-insns test (RELRO)
1408 366079 FPXX Support for MIPS32 Valgrind
1409 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1410 366344 Multiple unhandled instruction for Aarch64
1411 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1412 367995 Integration of memcheck with custom memory allocator
1413 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1414 368412 False positive result for altivec capability check
1415 368416 Add tc06_two_races_xml.exp output for ppc64
1416 368419 Perf Events ioctls not implemented
1417 368461 mmapunmap test fails on ppc64
1418 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1419 369000 AMD64 fma4 instructions unsupported.
1420 369169 ppc64 fails jm_int_isa_2_07 test
1421 369175 jm_vec_isa_2_07 test crashes on ppc64
1422 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1423 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1424 369359 msghdr_foreachfield can crash when handling bad iovec
1425 369360 Bad sigprocmask old or new sets can crash valgrind
1426 369361 vmsplice syscall wrapper crashes on bad iovec
1427 369362 Bad sigaction arguments crash valgrind
1428 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1429 369402 Bad set/get_thread_area pointer crashes valgrind
1430 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1431 369446 valgrind crashes on unknown fcntl command
1432 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1433 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1434 370265 ISA 3.0 HW cap stuff needs updating
1435 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1436 372195 Power PC, xxsel instruction is not always recognized
1438 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1439 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1440 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1441 n-i-bz Fix ppoll_alarm exclusion on OS X
1442 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1443 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1444 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1445 n-i-bz amd64: memcheck false positive with shr %edx
1446 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1447 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1448 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1449 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1451 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1452 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1453 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1457 Release 3.11.0 (22 September 2015)
1458 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1460 3.11.0 is a feature release with many improvements and the usual
1461 collection of bug fixes.
1463 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1464 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1465 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1466 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1467 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1468 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1470 * ================== PLATFORM CHANGES =================
1472 * Support for Solaris/x86 and Solaris/amd64 has been added.
1474 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1476 * Preliminary support for the Tilera TileGX architecture has been added.
1478 * s390x: It is now required for the host to have the "long displacement"
1479 facility. The oldest supported machine model is z990.
1481 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1482 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1484 * The JIT's register allocator is significantly faster, making the JIT
1485 as a whole somewhat faster, so JIT-intensive activities, for example
1486 program startup, are modestly faster, around 5%.
1488 * There have been changes to the default settings of several command
1489 line flags, as detailed below.
1491 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1492 capable hosts, the simulated CPUID will now indicate AVX2 support.
1494 * ==================== TOOL CHANGES ====================
1498 - The default value for --leak-check-heuristics has been changed from
1499 "none" to "all". This helps to reduce the number of possibly
1500 lost blocks, in particular for C++ applications.
1502 - The default value for --keep-stacktraces has been changed from
1503 "malloc-then-free" to "malloc-and-free". This has a small cost in
1504 memory (one word per malloc-ed block) but allows Memcheck to show the
1505 3 stacktraces of a dangling reference: where the block was allocated,
1506 where it was freed, and where it is acccessed after being freed.
1508 - The default value for --partial-loads-ok has been changed from "no" to
1509 "yes", so as to avoid false positive errors resulting from some kinds
1510 of vectorised loops.
1512 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1513 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1514 than get_vbits when you need to associate byte data value with
1515 their corresponding validity bits.
1517 - The 'block_list' monitor command has been enhanced:
1518 o it can print a range of loss records
1519 o it now accepts an optional argument 'limited <max_blocks>'
1520 to control the number of blocks printed.
1521 o if a block has been found using a heuristic, then
1522 'block_list' now shows the heuristic after the block size.
1523 o the loss records/blocks to print can be limited to the blocks
1524 found via specified heuristics.
1526 - The C helper functions used to instrument loads on
1527 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1528 replaced by handwritten assembly sequences. This gives speedups
1529 in the region of 0% to 7% for those targets only.
1531 - A new command line option, --expensive-definedness-checks=yes|no,
1532 has been added. This is useful for avoiding occasional invalid
1533 uninitialised-value errors in optimised code. Watch out for
1534 runtime degradation, as this can be up to 25%. As always, though,
1535 the slowdown is highly application specific. The default setting
1540 - A new monitor command 'all_snapshots <filename>' dumps all
1541 snapshots taken so far.
1545 - Significant memory reduction and moderate speedups for
1546 --history-level=full for applications accessing a lot of memory
1547 with many different stacktraces.
1549 - The default value for --conflict-cache-size=N has been doubled to
1550 2000000. Users that were not using the default value should
1551 preferably also double the value they give.
1553 The default was changed due to the changes in the "full history"
1554 implementation. Doubling the value gives on average a slightly more
1555 complete history and uses similar memory (or significantly less memory
1556 in the worst case) than the previous implementation.
1558 - The Helgrind monitor command 'info locks' now accepts an optional
1559 argument 'lock_addr', which shows information about the lock at the
1562 - When using --history-level=full, the new Helgrind monitor command
1563 'accesshistory <addr> [<len>]' will show the recorded accesses for
1564 <len> (or 1) bytes at <addr>.
1566 * ==================== OTHER CHANGES ====================
1568 * The default value for the --smc-check option has been changed from
1569 "stack" to "all-non-file" on targets that provide automatic D-I
1570 cache coherence (x86, amd64 and s390x). The result is to provide,
1571 by default, transparent support for JIT generated and self-modifying
1572 code on all targets.
1574 * Mac OS X only: the default value for the --dsymutil option has been
1575 changed from "no" to "yes", since any serious usage on Mac OS X
1576 always required it to be "yes".
1578 * The command line options --db-attach and --db-command have been removed.
1579 They were deprecated in 3.10.0.
1581 * When a process dies due to a signal, Valgrind now shows the signal
1582 and the stacktrace at default verbosity (i.e. verbosity 1).
1584 * The address description logic used by Memcheck and Helgrind now
1585 describes addresses in anonymous segments, file mmap-ed segments,
1586 shared memory segments and the brk data segment.
1588 * The new option --error-markers=<begin>,<end> can be used to mark the
1589 begin/end of errors in textual output mode, to facilitate
1590 searching/extracting errors in output files that mix valgrind errors
1591 with program output.
1593 * The new option --max-threads=<number> can be used to change the number
1594 of threads valgrind can handle. The default is 500 threads which
1595 should be more than enough for most applications.
1597 * The new option --valgrind-stacksize=<number> can be used to change the
1598 size of the private thread stacks used by Valgrind. This is useful
1599 for reducing memory use or increasing the stack size if Valgrind
1600 segfaults due to stack overflow.
1602 * The new option --avg-transtab-entry-size=<number> can be used to specify
1603 the expected instrumented block size, either to reduce memory use or
1604 to avoid excessive retranslation.
1606 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1608 * New and modified GDB server monitor features:
1610 - When a signal is reported in GDB, you can now use the GDB convenience
1611 variable $_siginfo to examine detailed signal information.
1613 - Valgrind's gdbserver now allows the user to change the signal
1614 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1615 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1616 continue without passing the signal to the process.
1618 - With GDB >= 7.10, the command 'target remote'
1619 will automatically load the executable file of the process running
1620 under Valgrind. This means you do not need to specify the executable
1621 file yourself, GDB will discover it itself. See GDB documentation about
1622 'qXfer:exec-file:read' packet for more info.
1624 * ==================== FIXED BUGS ====================
1626 The following bugs have been fixed or resolved. Note that "n-i-bz"
1627 stands for "not in bugzilla" -- that is, a bug that was reported to us
1628 but never got a bugzilla entry. We encourage you to file bugs in
1629 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1630 than mailing the developers (or mailing lists) directly -- bugs that
1631 are not entered into bugzilla tend to get forgotten about or ignored.
1633 To see details of a given bug, visit
1634 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1635 where XXXXXX is the bug number as listed below.
1637 116002 VG_(printf): Problems with justification of strings and integers
1638 155125 avoid cutting away file:lineno after long function name
1639 197259 Unsupported arch_prtctl PR_SET_GS option
1640 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
1641 201216 Fix Valgrind does not support pthread_sigmask() on OS X
1642 201435 Fix Darwin: -v does not show kernel version
1643 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
1644 211256 Fixed an outdated comment regarding the default platform.
1645 211529 Incomplete call stacks for code compiled by newer versions of MSVC
1646 211926 Avoid compilation warnings in valgrind.h with -pedantic
1647 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
1649 226609 Crediting upstream authors in man page
1650 231257 Valgrind omits path when executing script from shebang line
1651 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
1652 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
1653 269360 s390x: Fix addressing mode selection for compare-and-swap
1654 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
1656 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
1657 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
1658 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
1659 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
1660 330147 libmpiwrap PMPI_Get_count returns undefined value
1661 333051 mmap of huge pages fails due to incorrect alignment
1663 334802 valgrind does not always explain why a given option is bad
1664 335618 mov.w rN, pc/sp (ARM32)
1665 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
1670 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
1671 338602 AVX2 bit in CPUID missing
1672 338606 Strange message for scripts with invalid interpreter
1673 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
1674 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
1675 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
1677 339156 gdbsrv not called for fatal signal
1678 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
1679 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
1680 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
1681 339442 Fix testsuite build failure on OS X 10.9
1682 339542 Enable compilation with Intel's ICC compiler
1683 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
1684 339688 Mac-specific ASM does not support .version directive (cpuid,
1685 tronical and pushfpopf tests)
1686 339745 Valgrind crash when check Marmalade app (partial fix)
1687 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
1688 339778 Linux/TileGx platform support for Valgrind
1689 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
1690 339789 Fix none/tests/execve test on Mac OS X 10.9
1691 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
1692 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
1693 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
1694 340392 Allow user to select more accurate definedness checking in memcheck
1695 to avoid invalid complaints on optimised code
1696 340430 Fix some grammatical weirdness in the manual.
1697 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
1698 341419 Signal handler ucontext_t not filled out correctly on OS X
1699 341539 VG_(describe_addr) should not describe address as belonging to client
1700 segment if it is past the heap end
1701 341613 Enable building of manythreads and thread-exits tests on Mac OS X
1702 341615 Fix none/tests/darwin/access_extended test on Mac OS X
1703 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
1704 341789 aarch64: shmat fails with valgrind on ARMv8
1705 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
1706 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
1707 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
1708 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
1709 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
1710 342221 socket connect false positive uninit memory for unknown af family
1711 342353 Allow dumping full massif output while valgrind is still running
1712 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
1716 342603 Add I2C_SMBUS ioctl support
1717 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
1718 342683 Mark memory past the initial brk limit as unaddressable
1719 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
1720 342795 Internal glibc __GI_mempcpy call should be intercepted
1721 342841 s390x: Support instructions fiebr(a) and fidbr(a)
1722 343012 Unhandled syscall 319 (memfd_create)
1723 343069 Patch updating v4l2 API support
1724 343173 helgrind crash during stack unwind
1725 343219 fix GET_STARTREGS for arm
1726 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
1727 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
1728 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
1729 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
1730 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
1731 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
1732 343597 ppc64le: incorrect use of offseof macro
1733 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
1734 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
1735 343732 Unhandled syscall 144 (setgid) on aarch64
1736 343733 Unhandled syscall 187 (msgctl and related) on aarch64
1737 343802 s390x: False positive "conditional jump or move depends on [..]
1738 343902 --vgdb=yes doesn't break when --xml=yes is used
1739 343967 Don't warn about setuid/setgid/setcap executable for directories
1740 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
1741 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
1742 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
1743 344054 www - update info for Solaris/illumos
1744 344416 'make regtest' does not work cleanly on OS X
1745 344235 Remove duplicate include of pub_core_aspacemgr.h
1746 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
1747 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
1748 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
1749 344314 callgrind_annotate ... warnings about commands containing newlines
1750 344318 socketcall should wrap recvmmsg and sendmmsg
1751 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
1752 344416 Fix 'make regtest' does not work cleanly on OS X
1753 344499 Fix compilation for Linux kernel >= 4.0.0
1754 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
1755 unix:349 (__pthread_fchdir)
1756 344559 Garbage collection of unused segment names in address space manager
1757 344560 Fix stack traces missing penultimate frame on OS X
1758 344621 Fix memcheck/tests/err_disable4 test on OS X
1759 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
1760 344702 Fix missing libobjc suppressions on OS X 10.10
1762 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
1763 344939 Fix memcheck/tests/xml1 on OS X 10.10
1764 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
1765 345079 Fix build problems in VEX/useful/test_main.c
1766 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
1767 345177 arm64: prfm (reg) not implemented
1768 345215 Performance improvements for the register allocator
1769 345248 add support for Solaris OS in valgrind
1770 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
1771 345394 Fix memcheck/tests/strchr on OS X
1772 345637 Fix memcheck/tests/sendmsg on OS X
1773 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
1774 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
1775 345887 Fix an assertion in the address space manager
1776 345928 amd64: callstack only contains current function for small stacks
1777 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1778 345987 MIPS64: Implement cavium LHX instruction
1779 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1780 346185 Fix typo saving altivec register v24
1781 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1782 and LibVEX_GuestPPC64_get_CR()
1783 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1784 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1785 346307 fuse filesystem syscall deadlocks
1786 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1787 346411 MIPS: SysRes::_valEx handling is incorrect
1788 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1789 346474 PPC64 Power 8, spr TEXASRU register not supported
1790 346487 Compiler generates "note" about a future ABI change for PPC64
1791 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1792 and causing spurious "invalid read of size 8" warnings
1793 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1794 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1795 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1796 347322 Power PC regression test cleanup
1797 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1799 347389 unhandled syscall: 373 (Linux ARM syncfs)
1800 347686 Patch set to cleanup PPC64 regtests
1801 347978 Remove bash dependencies where not needed
1802 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1803 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1805 348102 Patch updating v4l2 API support
1806 348247 amd64 front end: jno jumps wrongly when overflow is not set
1807 348269 Improve mmap MAP_HUGETLB support.
1808 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1809 348345 Assertion fails for negative lineno
1810 348377 Unsupported ARM instruction: yield
1811 348565 Fix detection of command line option availability for clang
1812 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1813 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1814 348748 Fix redundant condition
1815 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1816 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1817 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1818 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1819 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1820 349626 Implemented additional Xen hypercalls
1821 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1822 349790 Clean up of the hardware capability checking utilities.
1823 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1824 349874 Fix typos in source code
1825 349879 memcheck: add handwritten assembly for helperc_LOADV*
1826 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1827 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1828 350202 Add limited param to 'monitor block_list'
1829 350290 s390x: Support instructions fixbr(a)
1830 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1831 350809 Fix none/tests/async-sigs for Solaris
1832 350811 Remove reference to --db-attach which has been removed.
1833 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1834 350854 hard-to-understand code in VG_(load_ELF)()
1835 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1836 351386 Solaris: Cannot run ld.so.1 under Valgrind
1837 351474 Fix VG_(iseqsigset) as obvious
1838 351531 Typo in /include/vki/vki-xen-physdev.h header guard
1839 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
1840 351858 ldsoexec support on Solaris
1841 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
1842 352130 helgrind reports false races for printfs using mempcpy on FILE* state
1843 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
1844 352320 arm64 crash on none/tests/nestedfs
1845 352765 Vbit test fails on Power 6
1846 352768 The mbar instruction is missing from the Power PC support
1847 352769 Power PC program priority register (PPR) is not supported
1848 n-i-bz Provide implementations of certain compiler builtins to support
1849 compilers that may not provide those
1850 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
1851 n-i-bz Fix compilation on distros with glibc < 2.5
1852 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
1853 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
1854 n-i-bz mremap did not work properly on shared memory
1855 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
1856 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
1858 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
1859 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
1860 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
1864 Release 3.10.1 (25 November 2014)
1865 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1866 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
1867 and backports fixes for all reported missing AArch64 ARMv8 instructions
1868 and syscalls from the trunk. If you package or deliver 3.10.0 for others
1869 to use, you might want to consider upgrading to 3.10.1 instead.
1871 The following bugs have been fixed or resolved. Note that "n-i-bz"
1872 stands for "not in bugzilla" -- that is, a bug that was reported to us
1873 but never got a bugzilla entry. We encourage you to file bugs in
1874 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1875 than mailing the developers (or mailing lists) directly -- bugs that
1876 are not entered into bugzilla tend to get forgotten about or ignored.
1878 To see details of a given bug, visit
1879 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1880 where XXXXXX is the bug number as listed below.
1882 335440 arm64: ld1 (single structure) is not implemented
1883 335713 arm64: unhanded instruction: prfm (immediate)
1884 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
1885 339182 ppc64: AvSplat ought to load destination vector register with [..]
1886 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
1887 339433 ppc64 lxvw4x instruction uses four 32-byte loads
1888 339645 Use correct tag names in sys_getdents/64 wrappers
1889 339706 Fix false positive for ioctl(TIOCSIG) on linux
1890 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
1891 339853 arm64 times syscall unknown
1892 339855 arm64 unhandled getsid/setsid syscalls
1893 339858 arm64 dmb sy not implemented
1894 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
1895 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
1896 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
1898 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
1899 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
1900 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
1901 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
1902 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
1903 340509 arm64: unhandled instruction fcvtas
1904 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
1905 340632 arm64: unhandled instruction fcvtas
1906 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
1907 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
1908 340788 warning: unhandled syscall: 318 (getrandom)
1909 340807 disInstr(arm): unhandled instruction: 0xEE989B20
1910 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
1911 340922 arm64: unhandled getgroups/setgroups syscalls
1912 350251 Fix typo in VEX utility program (test_main.c).
1913 350407 arm64: unhandled instruction ucvtf (vector, integer)
1914 350809 none/tests/async-sigs breaks when run under cron on Solaris
1915 350811 update README.solaris after r15445
1916 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
1917 350854 strange code in VG_(load_ELF)()
1918 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1919 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
1920 n-i-bz Add missing ]] to terminate CDATA.
1921 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
1922 n-i-bz Enable sys_fadvise64_64 on arm32.
1923 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
1924 n-i-bz Add test cases for all known arm64 load/store instructions.
1925 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
1926 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
1927 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
1928 n-i-bz arm64: Implement "fcvtpu w, s".
1929 n-i-bz arm64: implement ADDP and various others
1930 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
1931 n-i-bz arm64: enable FCVT{A,N}S X,S.
1933 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
1937 Release 3.10.0 (10 September 2014)
1938 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1940 3.10.0 is a feature release with many improvements and the usual
1941 collection of bug fixes.
1943 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1944 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1945 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
1946 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
1947 significantly improved relative to the 3.9.0 release.
1949 * ================== PLATFORM CHANGES =================
1951 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
1952 is mostly complete, and is usable, but some SIMD instructions are as
1955 * Support for little-endian variant of the 64-bit POWER architecture.
1957 * Support for Android on MIPS32.
1959 * Support for 64bit FPU on MIPS32 platforms.
1961 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
1963 * Configuration for and running on Android targets has changed.
1964 See README.android in the source tree for details.
1966 * ================== DEPRECATED FEATURES =================
1968 * --db-attach is now deprecated and will be removed in the next
1969 valgrind feature release. The built-in GDB server capabilities are
1970 superior and should be used instead. Learn more here:
1971 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
1973 * ==================== TOOL CHANGES ====================
1977 - Client code can now selectively disable and re-enable reporting of
1978 invalid address errors in specific ranges using the new client
1979 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
1980 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
1982 - Leak checker: there is a new leak check heuristic called
1983 "length64". This is used to detect interior pointers pointing 8
1984 bytes inside a block, on the assumption that the first 8 bytes
1985 holds the value "block size - 8". This is used by
1986 sqlite3MemMalloc, for example.
1988 - Checking of system call parameters: if a syscall parameter
1989 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
1990 several fields not initialised, an error is now reported for each
1991 field. Previously, an error was reported only for the first
1992 uninitialised field.
1994 - Mismatched alloc/free checking: a new flag
1995 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
1996 such checks if necessary.
2000 - Improvements to error messages:
2002 o Race condition error message involving heap allocated blocks also
2003 show the thread number that allocated the raced-on block.
2005 o All locks referenced by an error message are now announced.
2006 Previously, some error messages only showed the lock addresses.
2008 o The message indicating where a lock was first observed now also
2009 describes the address/location of the lock.
2011 - Helgrind now understands the Ada task termination rules and
2012 creates a happens-before relationship between a terminated task
2013 and its master. This avoids some false positives and avoids a big
2014 memory leak when a lot of Ada tasks are created and terminated.
2015 The interceptions are only activated with forthcoming releases of
2016 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
2018 - A new GDB server monitor command "info locks" giving the list of
2019 locks, their location, and their status.
2023 - callgrind_control now supports the --vgdb-prefix argument,
2024 which is needed if valgrind was started with this same argument.
2026 * ==================== OTHER CHANGES ====================
2028 * Unwinding through inlined function calls. Stack unwinding can now
2029 make use of Dwarf3 inlined-unwind information if it is available.
2030 The practical effect is that inlined calls become visible in stack
2031 traces. The suppression matching machinery has been adjusted
2032 accordingly. This is controlled by the new option
2033 --read-inline-info=yes|no. Currently this is enabled by default
2034 only on Linux and Android targets and only for the tools Memcheck,
2037 * Valgrind can now read EXIDX unwind information on 32-bit ARM
2038 targets. If an object contains both CFI and EXIDX unwind
2039 information, Valgrind will prefer the CFI over the EXIDX. This
2040 facilitates unwinding through system libraries on arm-android
2043 * Address description logic has been improved and is now common
2044 between Memcheck and Helgrind, resulting in better address
2045 descriptions for some kinds of error messages.
2047 * Error messages about dubious arguments (eg, to malloc or calloc) are
2048 output like other errors. This means that they can be suppressed
2049 and they have a stack trace.
2051 * The C++ demangler has been updated for better C++11 support.
2053 * New and modified GDB server monitor features:
2055 - Thread local variables/storage (__thread) can now be displayed.
2057 - The GDB server monitor command "v.info location <address>"
2058 displays information about an address. The information produced
2059 depends on the tool and on the options given to valgrind.
2060 Possibly, the following are described: global variables, local
2061 (stack) variables, allocated or freed blocks, ...
2063 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
2064 ask the GDB server to stop at the start of program execution, at
2065 the end of the program execution and on Valgrind internal errors.
2067 - A new monitor command "v.info stats" shows various Valgrind core
2068 and tool statistics.
2070 - A new monitor command "v.set hostvisibility" allows the GDB server
2071 to provide access to Valgrind internal host status/memory.
2073 * A new option "--aspace-minaddr=<address>" can in some situations
2074 allow the use of more memory by decreasing the address above which
2075 Valgrind maps memory. It can also be used to solve address
2076 conflicts with system libraries by increasing the default value.
2077 See user manual for details.
2079 * The amount of memory used by Valgrind to store debug info (unwind
2080 info, line number information and symbol data) has been
2081 significantly reduced, even though Valgrind now reads more
2082 information in order to support unwinding of inlined function calls.
2084 * Dwarf3 handling with --read-var-info=yes has been improved:
2086 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
2088 - Code compiled with
2089 -ffunction-sections -fdata-sections -Wl,--gc-sections
2090 no longer causes assertion failures.
2092 * Improved checking for the --sim-hints= and --kernel-variant=
2093 options. Unknown strings are now detected and reported to the user
2096 * The semantics of stack start/end boundaries in the valgrind.h
2097 VALGRIND_STACK_REGISTER client request has been clarified and
2098 documented. The convention is that start and end are respectively
2099 the lowest and highest addressable bytes of the stack.
2101 * ==================== FIXED BUGS ====================
2103 The following bugs have been fixed or resolved. Note that "n-i-bz"
2104 stands for "not in bugzilla" -- that is, a bug that was reported to us
2105 but never got a bugzilla entry. We encourage you to file bugs in
2106 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2107 than mailing the developers (or mailing lists) directly -- bugs that
2108 are not entered into bugzilla tend to get forgotten about or ignored.
2110 To see details of a given bug, visit
2111 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2112 where XXXXXX is the bug number as listed below.
2114 175819 Support for ipv6 socket reporting with --track-fds
2115 232510 make distcheck fails
2116 249435 Analyzing wine programs with callgrind triggers a crash
2117 278972 support for inlined function calls in stacktraces and suppression
2119 291310 FXSAVE instruction marks memory as undefined on amd64
2120 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2121 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2122 315199 vgcore file for threaded app does not show which thread crashed
2123 315952 tun/tap ioctls are not supported
2124 323178 Unhandled instruction: PLDW register (ARM)
2125 323179 Unhandled instruction: PLDW immediate (ARM)
2126 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2127 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2128 325124 [MIPSEL] Compilation error
2129 325477 Phase 4 support for IBM Power ISA 2.07
2130 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2131 325628 Phase 5 support for IBM Power ISA 2.07
2132 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2133 325751 Missing the two privileged Power PC Transactional Memory Instructions
2134 325816 Phase 6 support for IBM Power ISA 2.07
2135 325856 Make SGCheck fail gracefully on unsupported platforms
2136 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2137 326436 DRD: False positive in libstdc++ std::list::push_back
2138 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2139 326462 Refactor vgdb to isolate invoker stuff into separate module
2140 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2141 326623 DRD: false positive conflict report in a field assignment
2142 326724 Valgrind does not compile on OSX 1.9 Mavericks
2143 326816 Intercept for __strncpy_sse2_unaligned missing?
2144 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2145 326983 Clear direction flag after tests on amd64.
2146 327212 Do not prepend the current directory to absolute path names.
2147 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2148 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2149 327284 s390x: Fix translation of the risbg instruction
2150 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2151 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2152 327916 DW_TAG_typedef may have no name
2153 327943 s390x: add a redirection for the 'index' function
2154 328100 XABORT not implemented
2155 328205 Implement additional Xen hypercalls
2156 328454 add support Backtraces with ARM unwind tables (EXIDX)
2157 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2158 328711 valgrind.1 manpage "memcheck options" section is badly generated
2159 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2160 329612 Incorrect handling of AT_BASE for image execution
2161 329694 clang warns about using uninitialized variable
2162 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2163 330228 mmap must align to VKI_SHMLBA on mips32
2164 330257 LLVM does not support `-mno-dynamic-no-pic` option
2165 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2166 330459 --track-fds=yes doesn't track eventfds
2167 330469 Add clock_adjtime syscall support
2168 330594 Missing sysalls on PowerPC / uClibc
2169 330622 Add test to regression suite for POWER instruction: dcbzl
2170 330939 Support for AMD's syscall instruction on x86
2172 330941 Typo in PRE(poll) syscall wrapper
2173 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2174 331254 Fix expected output for memcheck/tests/dw4
2175 331255 Fix race condition in test none/tests/coolo_sigaction
2176 331257 Fix type of jump buffer in test none/tests/faultstatus
2177 331305 configure uses bash specific syntax
2178 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2179 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2180 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2181 331829 Unexpected ioctl opcode sign extension
2182 331830 ppc64: WARNING: unhandled syscall: 96/97
2183 331839 drd/tests/sem_open specifies invalid semaphore name
2184 331847 outcome of drd/tests/thread_name is nondeterministic
2185 332037 Valgrind cannot handle Thumb "add pc, reg"
2186 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2187 consistency checks enabled
2188 332263 intercepts for pthread_rwlock_timedrdlock and
2189 pthread_rwlock_timedwrlock are incorrect
2190 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2192 332276 Implement additional Xen hypercalls
2193 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2194 332765 Fix ms_print to create temporary files in a proper directory
2195 333072 drd: Add semaphore annotations
2196 333145 Tests for missaligned PC+#imm access for arm
2197 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2198 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2199 333248 WARNING: unhandled syscall: unix:443
2200 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2201 333501 cachegrind: assertion: Cache set count is not a power of two.
2204 333666 Recognize MPX instructions and bnd prefix.
2205 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2206 333817 Valgrind reports the memory areas written to by the SG_IO
2208 334049 lzcnt fails silently (x86_32)
2209 334384 Valgrind does not have support Little Endian support for
2211 334585 recvmmsg unhandled (+patch) (arm)
2212 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2213 334727 Build fails with -Werror=format-security
2214 334788 clarify doc about --log-file initial program directory
2215 334834 PPC64 Little Endian support, patch 2
2216 334836 PPC64 Little Endian support, patch 3 testcase fixes
2217 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2218 335034 Unhandled ioctl: HCIGETDEVLIST
2219 335155 vgdb, fix error print statement.
2220 335262 arm64: movi 8bit version is not supported
2221 335263 arm64: dmb instruction is not implemented
2222 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2223 335496 arm64: sbc/abc instructions are not implemented
2224 335554 arm64: unhandled instruction: abs
2225 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2226 335735 arm64: unhandled instruction: cnt
2227 335736 arm64: unhandled instruction: uaddlv
2228 335848 arm64: unhandled instruction: {s,u}cvtf
2229 335902 arm64: unhandled instruction: sli
2230 335903 arm64: unhandled instruction: umull (vector)
2231 336055 arm64: unhandled instruction: mov (element)
2232 336062 arm64: unhandled instruction: shrn{,2}
2233 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2234 336189 arm64: unhandled Instruction: mvn
2235 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2236 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2237 336772 Make moans about unknown ioctls more informative
2238 336957 Add a section about the Solaris/illumos port on the webpage
2239 337094 ifunc wrapper is broken on ppc64
2240 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2241 337528 leak check heuristic for block prefixed by length as 64bit number
2242 337740 Implement additional Xen hypercalls
2243 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2244 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2245 337871 deprecate --db-attach
2246 338023 Add support for all V4L2/media ioctls
2247 338024 inlined functions are not shown if DW_AT_ranges is used
2248 338106 Add support for 'kcmp' syscall
2249 338115 DRD: computed conflict set differs from actual after fork
2250 338160 implement display of thread local storage in gdbsrv
2251 338205 configure.ac and check for -Wno-tautological-compare
2252 338300 coredumps are missing one byte of every segment
2253 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2254 338499 --sim-hints parsing broken due to wrong order in tokens
2255 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2256 338681 Unable to unwind through clone thread created on i386-linux
2257 338698 race condition between gdbsrv and vgdb on startup
2258 338703 helgrind on arm-linux gets false positives in dynamic loader
2259 338791 alt dwz files can be relative of debug/main file
2260 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2261 338932 build V-trunk with gcc-trunk
2262 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2263 345079 Fix build problems in VEX/useful/test_main.c
2264 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2265 n-i-bz s390x: Fix memory corruption for multithreaded applications
2266 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2267 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2268 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2269 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2270 n-i-bz Add command-line option --default-suppressions
2271 n-i-bz Add support for BLKDISCARDZEROES ioctl
2272 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2273 n-i-bz Add support for sys_pivot_root and sys_unshare
2275 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2276 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2277 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2281 Release 3.9.0 (31 October 2013)
2282 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2283 3.9.0 is a feature release with many improvements and the usual
2284 collection of bug fixes.
2286 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2287 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2288 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2289 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2291 * ================== PLATFORM CHANGES =================
2293 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2294 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2296 * Support for MIPS DSP ASE on MIPS32 platforms.
2298 * Support for s390x Decimal Floating Point instructions on hosts that
2299 have the DFP facility installed.
2301 * Support for POWER8 (Power ISA 2.07) instructions
2303 * Support for Intel AVX2 instructions. This is available only on 64
2306 * Initial support for Intel Transactional Synchronization Extensions,
2309 * Initial support for Hardware Transactional Memory on POWER.
2311 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2312 run large GUI apps tolerably well.
2314 * ==================== TOOL CHANGES ====================
2318 - Improvements in handling of vectorised code, leading to
2319 significantly fewer false error reports. You need to use the flag
2320 --partial-loads-ok=yes to get the benefits of these changes.
2322 - Better control over the leak checker. It is now possible to
2323 specify which leak kinds (definite/indirect/possible/reachable)
2324 should be displayed, which should be regarded as errors, and which
2325 should be suppressed by a given leak suppression. This is done
2326 using the options --show-leak-kinds=kind1,kind2,..,
2327 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2328 "match-leak-kinds:" line in suppression entries, respectively.
2330 Note that generated leak suppressions contain this new line and
2331 are therefore more specific than in previous releases. To get the
2332 same behaviour as previous releases, remove the "match-leak-kinds:"
2333 line from generated suppressions before using them.
2335 - Reduced "possible leak" reports from the leak checker by the use
2336 of better heuristics. The available heuristics provide detection
2337 of valid interior pointers to std::stdstring, to new[] allocated
2338 arrays with elements having destructors and to interior pointers
2339 pointing to an inner part of a C++ object using multiple
2340 inheritance. They can be selected individually using the
2341 option --leak-check-heuristics=heur1,heur2,...
2343 - Better control of stacktrace acquisition for heap-allocated
2344 blocks. Using the --keep-stacktraces option, it is possible to
2345 control independently whether a stack trace is acquired for each
2346 allocation and deallocation. This can be used to create better
2347 "use after free" errors or to decrease Valgrind's resource
2348 consumption by recording less information.
2350 - Better reporting of leak suppression usage. The list of used
2351 suppressions (shown when the -v option is given) now shows, for
2352 each leak suppressions, how many blocks and bytes it suppressed
2353 during the last leak search.
2357 - False errors resulting from the use of statically initialised
2358 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2361 - False errors resulting from the use of pthread_cond_waits that
2362 timeout, have been removed.
2364 * ==================== OTHER CHANGES ====================
2366 * Some attempt to tune Valgrind's space requirements to the expected
2367 capabilities of the target:
2369 - The default size of the translation cache has been reduced from 8
2370 sectors to 6 on Android platforms, since each sector occupies
2371 about 40MB when using Memcheck.
2373 - The default size of the translation cache has been increased to 16
2374 sectors on all other platforms, reflecting the fact that large
2375 applications require instrumentation and storage of huge amounts
2376 of code. For similar reasons, the number of memory mapped
2377 segments that can be tracked has been increased by a factor of 6.
2379 - In all cases, the maximum number of sectors in the translation
2380 cache can be controlled by the new flag --num-transtab-sectors.
2382 * Changes in how debug info (line numbers, etc) is read:
2384 - Valgrind no longer temporarily mmaps the entire object to read
2385 from it. Instead, reading is done through a small fixed sized
2386 buffer. This avoids virtual memory usage spikes when Valgrind
2387 reads debuginfo from large shared objects.
2389 - A new experimental remote debug info server. Valgrind can read
2390 debug info from a different machine (typically, a build host)
2391 where debuginfo objects are stored. This can save a lot of time
2392 and hassle when running Valgrind on resource-constrained targets
2393 (phones, tablets) when the full debuginfo objects are stored
2394 somewhere else. This is enabled by the --debuginfo-server=
2397 - Consistency checking between main and debug objects can be
2398 disabled using the --allow-mismatched-debuginfo option.
2400 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2401 scanning can recover stack traces in some cases when the normal
2402 unwind mechanisms fail. Stack scanning is best described as "a
2403 nasty, dangerous and misleading hack" and so is disabled by default.
2404 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2407 * Detection and merging of recursive stack frame cycles. When your
2408 program has recursive algorithms, this limits the memory used by
2409 Valgrind for recorded stack traces and avoids recording
2410 uninteresting repeated calls. This is controlled by the command
2411 line option --merge-recursive-frame and by the monitor command
2412 "v.set merge-recursive-frames".
2414 * File name and line numbers for used suppressions. The list of used
2415 suppressions (shown when the -v option is given) now shows, for each
2416 used suppression, the file name and line number where the suppression
2419 * New and modified GDB server monitor features:
2421 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2422 that can be used to execute gdbserver monitor commands from the
2425 - A new monitor command, "v.info open_fds", that gives the list of
2426 open file descriptors and additional details.
2428 - An optional message in the "v.info n_errs_found" monitor command,
2429 for example "v.info n_errs_found test 1234 finished", allowing a
2430 comment string to be added to the process output, perhaps for the
2431 purpose of separating errors of different tests or test phases.
2433 - A new monitor command "v.info execontext" that shows information
2434 about the stack traces recorded by Valgrind.
2436 - A new monitor command "v.do expensive_sanity_check_general" to run
2437 some internal consistency checks.
2439 * New flag --sigill-diagnostics to control whether a diagnostic
2440 message is printed when the JIT encounters an instruction it can't
2441 translate. The actual behavior -- delivery of SIGILL to the
2442 application -- is unchanged.
2444 * The maximum amount of memory that Valgrind can use on 64 bit targets
2445 has been increased from 32GB to 64GB. This should make it possible
2446 to run applications on Memcheck that natively require up to about 35GB.
2448 * ==================== FIXED BUGS ====================
2450 The following bugs have been fixed or resolved. Note that "n-i-bz"
2451 stands for "not in bugzilla" -- that is, a bug that was reported to us
2452 but never got a bugzilla entry. We encourage you to file bugs in
2453 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2454 than mailing the developers (or mailing lists) directly -- bugs that
2455 are not entered into bugzilla tend to get forgotten about or ignored.
2457 To see details of a given bug, visit
2458 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2459 where XXXXXX is the bug number as listed below.
2461 123837 system call: 4th argument is optional, depending on cmd
2462 135425 memcheck should tell you where Freed blocks were Mallocd
2463 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2464 207815 Adds some of the drm ioctls to syswrap-linux.c
2465 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2466 252955 Impossible to compile with ccache
2467 253519 Memcheck reports auxv pointer accesses as invalid reads.
2468 263034 Crash when loading some PPC64 binaries
2469 269599 Increase deepest backtrace
2470 274695 s390x: Support "compare to/from logical" instructions (z196)
2471 275800 s390x: Autodetect cache info (part 2)
2472 280271 Valgrind reports possible memory leaks on still-reachable std::string
2473 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2474 289578 Backtraces with ARM unwind tables (stack scan flags)
2475 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2476 304832 ppc32: build failure
2477 305431 Use find_buildid shdr fallback for separate .debug files
2478 305728 Add support for AVX2 instructions
2479 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2480 306035 s390x: Fix IR generation for LAAG and friends
2481 306054 s390x: Condition code computation for convert-to-int/logical
2482 306098 s390x: alternate opcode form for convert to/from fixed
2483 306587 Fix cache line detection from auxiliary vector for PPC.
2484 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2485 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2486 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2487 307101 sys_capget second argument can be NULL
2488 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2489 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2490 307113 s390x: DFP support
2491 307141 valgrind does't work in mips-linux system
2492 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2493 307285 x86_amd64 feature test for avx in test suite is wrong
2494 307290 memcheck overlap testcase needs memcpy version filter
2495 307463 Please add "&limit=0" to the "all open bugs" link
2496 307465 --show-possibly-lost=no should reduce the error count / exit code
2497 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2498 307729 pkgconfig support broken valgrind.pc
2499 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2500 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2501 308089 Unhandled syscall on ppc64: prctl
2502 308135 PPC32 MPC8xx has 16 bytes cache size
2503 308321 testsuite memcheck filter interferes with gdb_filter
2505 308341 vgdb should report process exit (or fatal signal)
2506 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2507 308495 Remove build dependency on installed Xen headers
2508 308573 Internal error on 64-bit instruction executed in 32-bit mode
2510 308627 pmovmskb validity bit propagation is imprecise
2511 308644 vgdb command for having the info for the track-fds option
2512 308711 give more info about aspacemgr and arenas in out_of_memory
2513 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2514 308718 ARM implement SMLALBB family of instructions
2515 308886 Missing support for PTRACE_SET/GETREGSET
2516 308930 syscall name_to_handle_at (303 on amd64) not handled
2517 309229 V-bit tester does not report number of tests generated
2518 309323 print unrecognized instuction on MIPS
2519 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2520 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2521 309430 Self hosting ppc64 encounters a vassert error on operand type
2522 309600 valgrind is a bit confused about 0-sized sections
2523 309823 Generate errors for still reachable blocks
2524 309921 PCMPISTRI validity bit propagation is imprecise
2525 309922 none/tests/ppc64/test_dfp5 sometimes fails
2526 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2527 310424 --read-var-info does not properly describe static variables
2528 310792 search additional path for debug symbols
2529 310931 s390x: Message-security assist (MSA) instruction extension [..]
2530 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2531 311318 ARM: "128-bit constant is not implemented" error message
2532 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2533 311690 V crashes because it redirects branches inside of a redirected function
2534 311880 x86_64: make regtest hangs at shell_valid1
2535 311922 WARNING: unhandled syscall: 170
2537 312171 ppc: insn selection for DFP
2538 312571 Rounding mode call wrong for the DFP Iops [..]
2539 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2540 312913 Dangling pointers error should also report the alloc stack trace
2541 312980 Building on Mountain Lion generates some compiler warnings
2542 313267 Adding MIPS64/Linux port to Valgrind
2545 313811 Buffer overflow in assert_fail
2546 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2547 314269 ppc: dead code in insn selection
2548 314718 ARM: implement integer divide instruction (sdiv and udiv)
2549 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2550 315441 sendmsg syscall should ignore unset msghdr msg_flags
2551 315534 msgrcv inside a thread causes valgrind to hang (block)
2552 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2553 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2554 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2555 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2556 316144 valgrind.1 manpage contains unknown ??? strings [..]
2557 316145 callgrind command line options in manpage reference (unknown) [..]
2558 316145 callgrind command line options in manpage reference [..]
2559 316181 drd: Fixed a 4x slowdown for certain applications
2560 316503 Valgrind does not support SSE4 "movntdqa" instruction
2561 316535 Use of |signed int| instead of |size_t| in valgrind messages
2562 316696 fluidanimate program of parsec 2.1 stuck
2563 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2564 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2565 317186 "Impossible happens" when occurs VCVT instruction on ARM
2566 317318 Support for Threading Building Blocks "scalable_malloc"
2567 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2568 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2569 317463 bmi testcase IR SANITY CHECK FAILURE
2570 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2571 318050 libmpiwrap fails to compile with out-of-source build
2572 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2573 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2574 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2575 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2576 318932 Add missing PPC64 and PPC32 system call support
2577 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2578 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2579 319494 VEX Makefile-gcc standalone build update after r2702
2580 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2581 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2582 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2583 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2584 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2585 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2586 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2587 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2588 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2589 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2590 320895 add fanotify support (patch included)
2591 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2592 321065 Valgrind updates for Xen 4.3
2593 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2594 321363 Unhandled instruction: SSAX (ARM + Thumb)
2595 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2596 321466 Unhandled instruction: SHASX (ARM + Thumb)
2597 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2598 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2599 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2600 321620 Unhandled instruction: UASX (ARM + Thumb)
2601 321621 Unhandled instruction: USAX (ARM + Thumb)
2602 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2603 321693 Unhandled instruction: LDRSBT (Thumb)
2604 321694 Unhandled instruction: UQASX (ARM + Thumb)
2605 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2606 321697 Unhandled instruction: UHASX (ARM + Thumb)
2607 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2608 321704 Unhandled instruction: REVSH (ARM + Thumb)
2609 321730 Add cg_diff and cg_merge man pages
2610 321738 Add vgdb and valgrind-listener man pages
2612 321891 Unhandled instruction: LDRHT (Thumb)
2613 321960 pthread_create() then alloca() causing invalid stack write errors
2614 321969 ppc32 and ppc64 don't support [lf]setxattr
2615 322254 Show threadname together with tid if set by application
2616 322294 Add initial support for IBM Power ISA 2.07
2617 322368 Assertion failure in wqthread_hijack under OS X 10.8
2618 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2619 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2620 322851 0bXXX binary literal syntax is not standard
2621 323035 Unhandled instruction: LDRSHT(Thumb)
2622 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2623 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2624 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2625 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2626 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2627 323437 Phase 2 support for IBM Power ISA 2.07
2628 323713 Support mmxext (integer sse) subset on i386 (athlon)
2629 323803 Transactional memory instructions are not supported for Power
2630 323893 SSE3 not available on amd cpus in valgrind
2631 323905 Probable false positive from Valgrind/drd on close()
2632 323912 valgrind.h header isn't compatible for mingw64
2633 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
2634 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
2635 324181 mmap does not handle MAP_32BIT
2636 324227 memcheck false positive leak when a thread calls exit+block [..]
2637 324421 Support for fanotify API on ARM architecture
2638 324514 gdbserver monitor cmd output behaviour consistency [..]
2639 324518 ppc64: Emulation of dcbt instructions does not handle [..]
2640 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
2641 324582 When access is made to freed memory, report both allocation [..]
2642 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
2643 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
2644 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
2645 324834 Unhandled instructions in Microsoft C run-time for x86_64
2646 324894 Phase 3 support for IBM Power ISA 2.07
2647 326091 drd: Avoid false race reports from optimized strlen() impls
2648 326113 valgrind libvex hwcaps error on AMD64
2649 n-i-bz Some wrong command line options could be ignored
2650 n-i-bz patch to allow fair-sched on android
2651 n-i-bz report error for vgdb snapshot requested before execution
2652 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
2654 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
2658 Release 3.8.1 (19 September 2012)
2659 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2660 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
2661 that occur moderately frequently in real use cases, adds support for
2662 some missing instructions on ARM, and fixes a deadlock condition on
2663 MacOSX. If you package or deliver 3.8.0 for others to use, you might
2664 want to consider upgrading to 3.8.1 instead.
2666 The following bugs have been fixed or resolved. Note that "n-i-bz"
2667 stands for "not in bugzilla" -- that is, a bug that was reported to us
2668 but never got a bugzilla entry. We encourage you to file bugs in
2669 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2670 than mailing the developers (or mailing lists) directly -- bugs that
2671 are not entered into bugzilla tend to get forgotten about or ignored.
2673 To see details of a given bug, visit
2674 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2675 where XXXXXX is the bug number as listed below.
2678 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
2679 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
2680 298281 wcslen causes false(?) uninitialised value warnings
2681 301281 valgrind hangs on OS X when the process calls system()
2682 304035 disInstr(arm): unhandled instruction 0xE1023053
2683 304867 implement MOVBE instruction in x86 mode
2684 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
2685 305042 amd64: implement 0F 7F encoding of movq between two registers
2686 305199 ARM: implement QDADD and QDSUB
2687 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
2688 305513 killed by fatal signal: SIGSEGV
2689 305690 DRD reporting invalid semaphore when sem_trywait fails
2690 305926 Invalid alignment checks for some AVX instructions
2691 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
2692 306310 3.8.0 release tarball missing some files
2693 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
2694 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
2695 n-i-bz shmat of a segment > 4Gb does not work
2696 n-i-bz simulate_control_c script wrong USR1 signal number on mips
2697 n-i-bz vgdb ptrace calls wrong on mips [...]
2698 n-i-bz Fixes for more MPI false positives
2699 n-i-bz exp-sgcheck's memcpy causes programs to segfault
2700 n-i-bz OSX build w/ clang: asserts at startup
2701 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
2702 n-i-bz fix a couple of union tag-vs-field mixups
2703 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
2705 The following bugs were fixed in 3.8.0 but not listed in this NEWS
2708 254088 Valgrind should know about UD2 instruction
2711 304754 NEWS blows TeX's little mind
2713 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
2717 Release 3.8.0 (10 August 2012)
2718 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2719 3.8.0 is a feature release with many improvements and the usual
2720 collection of bug fixes.
2722 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2723 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
2724 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
2725 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
2726 There is initial support for MacOSX 10.8, but it is not usable for
2727 serious work at present.
2729 * ================== PLATFORM CHANGES =================
2731 * Support for MIPS32 platforms running Linux. Valgrind has been
2732 tested on MIPS32 and MIPS32r2 platforms running different Debian
2733 Squeeze and MeeGo distributions. Both little-endian and big-endian
2734 cores are supported. The tools Memcheck, Massif and Lackey have
2735 been tested and are known to work. See README.mips for more details.
2737 * Preliminary support for Android running on x86.
2739 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
2741 * Support for Intel AVX instructions and for AES instructions. This
2742 support is available only for 64 bit code.
2744 * Support for POWER Decimal Floating Point instructions.
2746 * ==================== TOOL CHANGES ====================
2748 * Non-libc malloc implementations are now supported. This is useful
2749 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
2750 Using the new option --soname-synonyms, such tools can be informed
2751 that the malloc implementation is either linked statically into the
2752 executable, or is present in some other shared library different
2753 from libc.so. This makes it possible to process statically linked
2754 programs, and programs using other malloc libraries, for example
2755 TCMalloc or JEMalloc.
2757 * For tools that provide their own replacement for malloc et al, the
2758 option --redzone-size=<number> allows users to specify the size of
2759 the padding blocks (redzones) added before and after each client
2760 allocated block. Smaller redzones decrease the memory needed by
2761 Valgrind. Bigger redzones increase the chance to detect blocks
2762 overrun or underrun. Prior to this change, the redzone size was
2763 hardwired to 16 bytes in Memcheck.
2767 - The leak_check GDB server monitor command now can
2768 control the maximum nr of loss records to output.
2770 - Reduction of memory use for applications allocating
2771 many blocks and/or having many partially defined bytes.
2773 - Addition of GDB server monitor command 'block_list' that lists
2774 the addresses/sizes of the blocks of a leak search loss record.
2776 - Addition of GDB server monitor command 'who_points_at' that lists
2777 the locations pointing at a block.
2779 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2780 detect an invalid access of these redzones, by marking them
2781 noaccess. Similarly, if a redzone size is given for a memory
2782 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2783 This still allows to find some bugs if the user has forgotten to
2784 mark the pool superblock noaccess.
2786 - Performance of memory leak check has been improved, especially in
2787 cases where there are many leaked blocks and/or many suppression
2788 rules used to suppress leak reports.
2790 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2791 more precise analysis, which is important for LLVM/Clang
2792 generated code. This is at the cost of somewhat reduced
2793 performance. Note there is no change to analysis precision or
2794 costs on Linux targets.
2798 - Added even more facilities that can help finding the cause of a data
2799 race, namely the command-line option --ptrace-addr and the macro
2800 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2802 - Fixed a subtle bug that could cause false positive data race reports.
2804 * ==================== OTHER CHANGES ====================
2806 * The C++ demangler has been updated so as to work well with C++
2807 compiled by up to at least g++ 4.6.
2809 * Tool developers can make replacement/wrapping more flexible thanks
2810 to the new option --soname-synonyms. This was reported above, but
2811 in fact is very general and applies to all function
2812 replacement/wrapping, not just to malloc-family functions.
2814 * Round-robin scheduling of threads can be selected, using the new
2815 option --fair-sched= yes. Prior to this change, the pipe-based
2816 thread serialisation mechanism (which is still the default) could
2817 give very unfair scheduling. --fair-sched=yes improves
2818 responsiveness of interactive multithreaded applications, and
2819 improves repeatability of results from the thread checkers Helgrind
2822 * For tool developers: support to run Valgrind on Valgrind has been
2823 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2825 * gdbserver now shows the float shadow registers as integer
2826 rather than float values, as the shadow values are mostly
2827 used as bit patterns.
2829 * Increased limit for the --num-callers command line flag to 500.
2831 * Performance improvements for error matching when there are many
2832 suppression records in use.
2834 * Improved support for DWARF4 debugging information (bug 284184).
2836 * Initial support for DWZ compressed Dwarf debug info.
2838 * Improved control over the IR optimiser's handling of the tradeoff
2839 between performance and precision of exceptions. Specifically,
2840 --vex-iropt-precise-memory-exns has been removed and replaced by
2841 --vex-iropt-register-updates, with extended functionality. This
2842 allows the Valgrind gdbserver to always show up to date register
2845 * Modest performance gains through the use of translation chaining for
2848 * ==================== FIXED BUGS ====================
2850 The following bugs have been fixed or resolved. Note that "n-i-bz"
2851 stands for "not in bugzilla" -- that is, a bug that was reported to us
2852 but never got a bugzilla entry. We encourage you to file bugs in
2853 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2854 than mailing the developers (or mailing lists) directly -- bugs that
2855 are not entered into bugzilla tend to get forgotten about or ignored.
2857 To see details of a given bug, visit
2858 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2859 where XXXXXX is the bug number as listed below.
2861 197914 Building valgrind from svn now requires automake-1.10
2862 203877 increase to 16Mb maximum allowed alignment for memalign et al
2863 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
2864 247386 make perf does not run all performance tests
2865 270006 Valgrind scheduler unfair
2866 270777 Adding MIPS/Linux port to Valgrind
2867 270796 s390x: Removed broken support for the TS insn
2868 271438 Fix configure for proper SSE4.2 detection
2869 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
2870 273475 Add support for AVX instructions
2871 274078 improved configure logic for mpicc
2872 276993 fix mremap 'no thrash checks'
2873 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
2874 281482 memcheck incorrect byte allocation count in realloc() for silly argument
2875 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
2876 283413 Fix wrong sanity check
2877 283671 Robustize alignment computation in LibVEX_Alloc
2878 283961 Adding support for some HCI IOCTLs
2879 284124 parse_type_DIE: confused by: DWARF 4
2880 284864 == 273475 (Add support for AVX instructions)
2881 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
2882 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
2883 285725 == 273475 (Add support for AVX instructions)
2884 286261 add wrapper for linux I2C_RDWR ioctl
2885 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
2886 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
2887 286384 configure fails "checking for a supported version of gcc"
2888 286497 == 273475 (Add support for AVX instructions)
2889 286596 == 273475 (Add support for AVX instructions)
2890 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
2891 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
2892 287260 Incorrect conditional jump or move depends on uninitialised value(s)
2893 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
2894 287307 == 273475 (Add support for AVX instructions)
2895 287858 VG_(strerror): unknown error
2896 288298 (MacOSX) unhandled syscall shm_unlink
2897 288995 == 273475 (Add support for AVX instructions)
2898 289470 Loading of large Mach-O thin binaries fails.
2899 289656 == 273475 (Add support for AVX instructions)
2900 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
2901 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
2902 289839 s390x: Provide support for unicode conversion instructions
2903 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
2904 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
2905 290655 Add support for AESKEYGENASSIST instruction
2906 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
2907 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
2908 291253 ES register not initialised in valgrind simulation
2909 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
2910 291865 s390x: Support the "Compare Double and Swap" family of instructions
2911 292300 == 273475 (Add support for AVX instructions)
2912 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
2913 292493 == 273475 (Add support for AVX instructions)
2914 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
2915 292627 Missing support for some SCSI ioctls
2916 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
2917 292841 == 273475 (Add support for AVX instructions)
2918 292993 implement the getcpu syscall on amd64-linux
2919 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
2920 293088 Add some VEX sanity checks for ppc64 unhandled instructions
2921 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
2922 293754 PCMPxSTRx not implemented for 16-bit characters
2923 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
2924 293808 CLFLUSH not supported by latest VEX for amd64
2925 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
2926 294048 MPSADBW instruction not implemented
2927 294055 regtest none/tests/shell fails when locale is not set to C
2928 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
2929 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
2930 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
2931 294260 disInstr_AMD64: disInstr miscalculated next %rip
2932 294523 --partial-loads-ok=yes causes false negatives
2933 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
2934 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
2935 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
2936 295089 can not annotate source for both helgrind and drd
2937 295221 POWER Processor decimal floating point instruction support missing
2938 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
2939 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
2940 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
2941 295617 ARM - Add some missing syscalls
2942 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
2943 296229 Linux user input device ioctls missing wrappers
2944 296318 ELF Debug info improvements (more than one rx/rw mapping)
2945 296422 Add translation chaining support
2946 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
2947 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
2948 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
2949 297078 gdbserver signal handling problems [..]
2950 297147 drd false positives on newly allocated memory
2951 297329 disallow decoding of IBM Power DFP insns on some machines
2952 297497 POWER Processor decimal floating point instruction support missing
2953 297701 Another alias for strncasecmp_l in libc-2.13.so
2954 297911 'invalid write' not reported when using APIs for custom mem allocators.
2955 297976 s390x: revisit EX implementation
2956 297991 Valgrind interferes with mmap()+ftell()
2957 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
2958 297993 Fix compilation of valgrind with gcc -g3.
2959 298080 POWER Processor DFP support missing, part 3
2960 298227 == 273475 (Add support for AVX instructions)
2961 298335 == 273475 (Add support for AVX instructions)
2962 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
2963 298394 s390x: Don't bail out on an unknown machine model. [..]
2964 298421 accept4() syscall (366) support is missing for ARM
2965 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
2966 298732 valgrind installation problem in ubuntu with kernel version 3.x
2967 298862 POWER Processor DFP instruction support missing, part 4
2968 298864 DWARF reader mis-parses DW_FORM_ref_addr
2969 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
2970 299053 Support DWARF4 DW_AT_high_pc constant form
2971 299104 == 273475 (Add support for AVX instructions)
2972 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
2973 299629 dup3() syscall (358) support is missing for ARM
2974 299694 POWER Processor DFP instruction support missing, part 5
2975 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
2976 299803 == 273475 (Add support for AVX instructions)
2977 299804 == 273475 (Add support for AVX instructions)
2978 299805 == 273475 (Add support for AVX instructions)
2979 300140 ARM - Missing (T1) SMMUL
2980 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
2981 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
2982 300414 FCOM and FCOMP unimplemented for amd64 guest
2983 301204 infinite loop in canonicaliseSymtab with ifunc symbol
2984 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
2985 301265 add x86 support to Android build
2986 301984 configure script doesn't detect certain versions of clang
2987 302205 Fix compiler warnings for POWER VEX code and POWER test cases
2988 302287 Unhandled movbe instruction on Atom processors
2989 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
2990 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
2991 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
2992 302656 == 273475 (Add support for AVX instructions)
2993 302709 valgrind for ARM needs extra tls support for android emulator [..]
2994 302827 add wrapper for CDROM_GET_CAPABILITY
2995 302901 Valgrind crashes with dwz optimized debuginfo
2996 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
2997 303116 Add support for the POWER instruction popcntb
2998 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
2999 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
3000 303466 == 273475 (Add support for AVX instructions)
3001 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
3002 303963 strstr() function produces wrong results under valgrind callgrind
3003 304054 CALL_FN_xx macros need to enforce stack alignment
3004 304561 tee system call not supported
3005 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
3006 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
3007 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
3008 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
3009 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
3010 n-i-bz s390x: Shadow registers can now be examined using vgdb
3012 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
3013 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
3017 Release 3.7.0 (5 November 2011)
3018 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3019 3.7.0 is a feature release with many significant improvements and the
3020 usual collection of bug fixes.
3022 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3023 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
3024 Support for recent distros and toolchain components (glibc 2.14, gcc
3025 4.6, MacOSX 10.7) has been added.
3027 * ================== PLATFORM CHANGES =================
3029 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
3030 analyse 64-bit programs running on z/Architecture. Most user space
3031 instructions up to and including z10 are supported. Valgrind has
3032 been tested extensively on z9, z10, and z196 machines running SLES
3033 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
3034 known to work well. Callgrind, Helgrind, and DRD work reasonably
3035 well on z9 and later models. See README.s390 for more details.
3037 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
3038 64-bit processes are supported. Some complex threaded applications
3039 (Firefox) are observed to hang when run as 32 bit applications,
3040 whereas 64-bit versions run OK. The cause is unknown. Memcheck
3041 will likely report some false errors. In general, expect some rough
3042 spots. This release also supports MacOSX 10.6, but drops support
3045 * Preliminary support for Android (on ARM). Valgrind can now run
3046 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
3047 README.android for more details, plus instructions on how to get
3050 * Support for the IBM Power ISA 2.06 (Power7 instructions)
3052 * General correctness and performance improvements for ARM/Linux, and,
3053 by extension, ARM/Android.
3055 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
3056 instruction set support is under development but is not available in
3059 * Support for AIX5 has been removed.
3061 * ==================== TOOL CHANGES ====================
3063 * Memcheck: some incremental changes:
3065 - reduction of memory use in some circumstances
3067 - improved handling of freed memory, which in some circumstances
3068 can cause detection of use-after-free that would previously have
3071 - fix of a longstanding bug that could cause false negatives (missed
3072 errors) in programs doing vector saturated narrowing instructions.
3074 * Helgrind: performance improvements and major memory use reductions,
3075 particularly for large, long running applications which perform many
3076 synchronisation (lock, unlock, etc) events. Plus many smaller
3079 - display of locksets for both threads involved in a race
3081 - general improvements in formatting/clarity of error messages
3083 - addition of facilities and documentation regarding annotation
3084 of thread safe reference counted C++ classes
3086 - new flag --check-stack-refs=no|yes [yes], to disable race checking
3087 on thread stacks (a performance hack)
3089 - new flag --free-is-write=no|yes [no], to enable detection of races
3090 where one thread accesses heap memory but another one frees it,
3091 without any coordinating synchronisation event
3093 * DRD: enabled XML output; added support for delayed thread deletion
3094 in order to detect races that occur close to the end of a thread
3095 (--join-list-vol); fixed a memory leak triggered by repeated client
3096 memory allocatation and deallocation; improved Darwin support.
3098 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
3100 * exp-sgcheck: this tool has been reduced in scope so as to improve
3101 performance and remove checking that Memcheck does better.
3102 Specifically, the ability to check for overruns for stack and global
3103 arrays is unchanged, but the ability to check for overruns of heap
3104 blocks has been removed. The tool has accordingly been renamed to
3105 exp-sgcheck ("Stack and Global Array Checking").
3107 * ==================== OTHER CHANGES ====================
3109 * GDB server: Valgrind now has an embedded GDB server. That means it
3110 is possible to control a Valgrind run from GDB, doing all the usual
3111 things that GDB can do (single stepping, breakpoints, examining
3112 data, etc). Tool-specific functionality is also available. For
3113 example, it is possible to query the definedness state of variables
3114 or memory from within GDB when running Memcheck; arbitrarily large
3115 memory watchpoints are supported, etc. To use the GDB server, start
3116 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3119 * Improved support for unfriendly self-modifying code: a new option
3120 --smc-check=all-non-file is available. This adds the relevant
3121 consistency checks only to code that originates in non-file-backed
3122 mappings. In effect this confines the consistency checking only to
3123 code that is or might be JIT generated, and avoids checks on code
3124 that must have been compiled ahead of time. This significantly
3125 improves performance on applications that generate code at run time.
3127 * It is now possible to build a working Valgrind using Clang-2.9 on
3130 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3131 These enable and disable error reporting on a per-thread, and
3132 nestable, basis. This is useful for hiding errors in particularly
3133 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3134 now uses this facility.
3136 * Added the --mod-funcname option to cg_diff.
3138 * ==================== FIXED BUGS ====================
3140 The following bugs have been fixed or resolved. Note that "n-i-bz"
3141 stands for "not in bugzilla" -- that is, a bug that was reported to us
3142 but never got a bugzilla entry. We encourage you to file bugs in
3143 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3144 mailing the developers (or mailing lists) directly -- bugs that are
3145 not entered into bugzilla tend to get forgotten about or ignored.
3147 To see details of a given bug, visit
3148 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3149 where XXXXXX is the bug number as listed below.
3151 79311 malloc silly arg warning does not give stack trace
3152 210935 port valgrind.h (not valgrind) to win32 to support client requests
3153 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3154 243404 Port to zSeries
3155 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3156 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3157 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3158 253206 Some fixes for the faultstatus testcase
3159 255223 capget testcase fails when running as root
3160 256703 xlc_dbl_u32.c testcase broken
3161 256726 Helgrind tests have broken inline asm
3162 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3163 264800 testcase compile failure on zseries
3164 265762 make public VEX headers compilable by G++ 3.x
3165 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3166 266753 configure script does not give the user the option to not use QtCore
3167 266931 gen_insn_test.pl is broken
3168 266961 ld-linux.so.2 i?86-linux strlen issues
3169 266990 setns instruction causes false positive
3170 267020 Make directory for temporary files configurable at run-time.
3171 267342 == 267997 (segmentation fault on Mac OS 10.6)
3172 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3173 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3174 267488 regtest: darwin support for 64-bit build
3175 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3176 267630 Add support for IBM Power ISA 2.06 -- stage 1
3177 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3178 267819 Add client request for informing the core about reallocation
3179 267925 laog data structure quadratic for a single sequence of lock
3180 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3181 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3182 268513 missed optimizations in fold_Expr
3183 268619 s390x: fpr - gpr transfer facility
3184 268620 s390x: reconsider "long displacement" requirement
3185 268621 s390x: improve IR generation for XC
3186 268715 s390x: FLOGR is not universally available
3187 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3188 268930 s390x: MHY is not universally available
3189 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3190 269079 Support ptrace system call on ARM
3191 269144 missing "Bad option" error message
3192 269209 conditional load and store facility (z196)
3193 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3194 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3195 269736 s390x: minor code generation tweaks
3196 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3197 269863 s390x: remove unused function parameters
3198 269864 s390x: tweak s390_emit_load_cc
3199 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3200 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3201 270115 s390x: rewrite some testcases
3202 270309 == 267997 (valgrind crash on startup)
3203 270320 add support for Linux FIOQSIZE ioctl() call
3204 270326 segfault while trying to sanitize the environment passed to execle
3205 270794 IBM POWER7 support patch causes regression in none/tests
3206 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3207 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3208 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3209 270959 s390x: invalid use of R0 as base register
3210 271042 VSX configure check fails when it should not
3211 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3212 271259 s390x: fix code confusion
3213 271337 == 267997 (Valgrind segfaults on MacOS X)
3214 271385 s390x: Implement Ist_MBE
3215 271501 s390x: misc cleanups
3216 271504 s390x: promote likely and unlikely
3217 271579 ppc: using wrong enum type
3218 271615 unhandled instruction "popcnt" (arch=amd10h)
3219 271730 Fix bug when checking ioctls: duplicate check
3220 271776 s390x: provide STFLE instruction support
3221 271779 s390x: provide clock instructions like STCK
3222 271799 Darwin: ioctls without an arg report a memory error
3223 271820 arm: fix type confusion
3224 271917 pthread_cond_timedwait failure leads to not-locked false positive
3225 272067 s390x: fix DISP20 macro
3226 272615 A typo in debug output in mc_leakcheck.c
3227 272661 callgrind_annotate chokes when run from paths containing regex chars
3228 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3229 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3230 272967 make documentation build-system more robust
3231 272986 Fix gcc-4.6 warnings with valgrind.h
3232 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3233 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3234 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3235 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3236 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3237 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3238 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3239 273778 exp-ptrcheck: unhandled sysno == 259
3240 274089 exp-ptrcheck: unhandled sysno == 208
3241 274378 s390x: Various dispatcher tweaks
3242 274447 WARNING: unhandled syscall: 340
3243 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3244 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3245 274926 valgrind does not build against linux-3
3246 275148 configure FAIL with glibc-2.14
3247 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3248 275168 Make Valgrind work for MacOSX 10.7 Lion
3249 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3250 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3251 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3252 275308 Fix implementation for ppc64 fres instruc
3253 275339 s390x: fix testcase compile warnings
3254 275517 s390x: Provide support for CKSM instruction
3255 275710 s390x: get rid of redundant address mode calculation
3256 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3257 275852 == 250101 (valgrind uses all swap space and is killed)
3258 276784 Add support for IBM Power ISA 2.06 -- stage 3
3259 276987 gdbsrv: fix tests following recent commits
3260 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3261 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3262 277471 Unhandled syscall: 340
3263 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3264 277653 ARM: support Thumb2 PLD instruction
3265 277663 ARM: NEON float VMUL by scalar incorrect
3266 277689 ARM: tests for VSTn with register post-index are broken
3267 277694 ARM: BLX LR instruction broken in ARM mode
3268 277780 ARM: VMOV.F32 (immediate) instruction is broken
3269 278057 fuse filesystem syscall deadlocks
3270 278078 Unimplemented syscall 280 on ppc32
3271 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3272 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3273 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3274 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3275 279027 Support for MVCL and CLCL instruction
3276 279027 s390x: Provide support for CLCL and MVCL instructions
3277 279062 Remove a redundant check in the insn selector for ppc.
3278 279071 JDK creates PTEST with redundant REX.W prefix
3279 279212 gdbsrv: add monitor cmd v.info scheduler.
3280 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3281 279698 memcheck discards valid-bits for packuswb
3282 279795 memcheck reports uninitialised values for mincore on amd64
3283 279994 Add support for IBM Power ISA 2.06 -- stage 3
3284 280083 mempolicy syscall check errors
3285 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3286 280710 s390x: config files for nightly builds
3287 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3288 280965 Valgrind breaks fcntl locks when program does mmap
3289 281138 WARNING: unhandled syscall: 340
3290 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3291 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3292 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3293 281468 s390x: handle do_clone and gcc clones in call traces
3294 281488 ARM: VFP register corruption
3295 281828 == 275284 (false memmove warning: "Source and destination overlap")
3296 281883 s390x: Fix system call wrapper for "clone".
3297 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3298 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3299 282238 SLES10: make check fails
3300 282979 strcasestr needs replacement with recent(>=2.12) glibc
3301 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3302 283243 Regression in ppc64 memcheck tests
3303 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3304 283427 re-connect epoll_pwait syscall on ARM linux
3305 283600 gdbsrv: android: port vgdb.c
3306 283709 none/tests/faultstatus needs to account for page size
3307 284305 filter_gdb needs enhancement to work on ppc64
3308 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3309 284472 Thumb2 ROR.W encoding T2 not implemented
3310 284621 XML-escape process command line in XML output
3311 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3312 that have non-power-of-2 sizes (also AMDs)
3313 n-i-bz don't be spooked by libraries mashed by elfhack
3314 n-i-bz don't be spooked by libxul.so linked with gold
3315 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3317 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3318 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3319 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3323 Release 3.6.1 (16 February 2011)
3324 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3325 3.6.1 is a bug fix release. It adds support for some SSE4
3326 instructions that were omitted in 3.6.0 due to lack of time. Initial
3327 support for glibc-2.13 has been added. A number of bugs causing
3328 crashing or assertion failures have been fixed.
3330 The following bugs have been fixed or resolved. Note that "n-i-bz"
3331 stands for "not in bugzilla" -- that is, a bug that was reported to us
3332 but never got a bugzilla entry. We encourage you to file bugs in
3333 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3334 mailing the developers (or mailing lists) directly -- bugs that are
3335 not entered into bugzilla tend to get forgotten about or ignored.
3337 To see details of a given bug, visit
3338 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3339 where XXXXXX is the bug number as listed below.
3341 188572 Valgrind on Mac should suppress setenv() mem leak
3342 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3343 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3344 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3345 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3346 254420 memory pool tracking broken
3347 254957 Test code failing to compile due to changes in memcheck.h
3348 255009 helgrind/drd: crash on chmod with invalid parameter
3349 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3350 255355 helgrind/drd: crash on threaded programs doing fork
3352 255418 (SSE4.x) rint call compiled with ICC
3353 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3354 255888 closing valgrindoutput tag outputted to log-stream on error
3355 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3356 255966 Slowness when using mempool annotations
3357 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3358 256600 super-optimized strcasecmp() false positive
3359 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3360 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3361 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3362 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3363 257276 Missing case in memcheck --track-origins=yes
3364 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3365 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3366 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3367 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3368 263099 callgrind_annotate counts Ir improperly [...]
3369 263877 undefined coprocessor instruction on ARMv7
3370 265964 configure FAIL with glibc-2.13
3371 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3372 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3373 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3374 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3375 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3377 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3381 Release 3.6.0 (21 October 2010)
3382 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3383 3.6.0 is a feature release with many significant improvements and the
3384 usual collection of bug fixes.
3386 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3387 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3388 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3390 -------------------------
3392 Here are some highlights. Details are shown further down:
3394 * Support for ARM/Linux.
3396 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3398 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3400 * Support for the SSE4.2 instruction set.
3402 * Enhancements to the Callgrind profiler, including the ability to
3403 handle CPUs with three levels of cache.
3405 * A new experimental heap profiler, DHAT.
3407 * A huge number of bug fixes and small enhancements.
3409 -------------------------
3411 Here are details of the above changes, together with descriptions of
3412 many other changes, and a list of fixed bugs.
3414 * ================== PLATFORM CHANGES =================
3416 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3417 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3418 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3420 This requires a CPU capable of running the ARMv7-A instruction set
3421 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3422 of the user space instruction set, including ARM and Thumb integer
3423 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3424 Cachegrind and Massif tools work properly; other tools work to
3427 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3428 with support for recent releases of the underlying toolchain
3429 components, notably gcc-4.5 and glibc-2.12.
3431 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3432 support also works much better on OS X 10.5, and is as solid as
3435 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3436 64-bit mode. In 32-bit mode, support is only available up to and
3437 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3438 supported in 64-bit mode, and 32-bit mode does in fact support the
3439 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3440 10.6 on 32-bit targets.
3442 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3443 and including version 2.05 is supported.
3445 * ==================== TOOL CHANGES ====================
3447 * Cachegrind has a new processing script, cg_diff, which finds the
3448 difference between two profiles. It's very useful for evaluating
3449 the performance effects of a change in a program.
3451 Related to this change, the meaning of cg_annotate's (rarely-used)
3452 --threshold option has changed; this is unlikely to affect many
3453 people, if you do use it please see the user manual for details.
3455 * Callgrind now can do branch prediction simulation, similar to
3456 Cachegrind. In addition, it optionally can count the number of
3457 executed global bus events. Both can be used for a better
3458 approximation of a "Cycle Estimation" as derived event (you need to
3459 update the event formula in KCachegrind yourself).
3461 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3462 rather than the L2 cache. This is to accommodate machines with
3463 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3464 cache configuration of such a machine it will run the simulation as
3465 if the L2 cache isn't present. This means the results are less
3466 likely to match the true result for the machine, but
3467 Cachegrind/Callgrind's results are already only approximate, and
3468 should not be considered authoritative. The results are still
3469 useful for giving a general idea about a program's locality.
3471 * Massif has a new option, --pages-as-heap, which is disabled by
3472 default. When enabled, instead of tracking allocations at the level
3473 of heap blocks (as allocated with malloc/new/new[]), it instead
3474 tracks memory allocations at the level of memory pages (as mapped by
3475 mmap, brk, etc). Each mapped page is treated as its own block.
3476 Interpreting the page-level output is harder than the heap-level
3477 output, but this option is useful if you want to account for every
3478 byte of memory used by a program.
3480 * DRD has two new command-line options: --free-is-write and
3481 --trace-alloc. The former allows to detect reading from already freed
3482 memory, and the latter allows tracing of all memory allocations and
3485 * DRD has several new annotations. Custom barrier implementations can
3486 now be annotated, as well as benign races on static variables.
3488 * DRD's happens before / happens after annotations have been made more
3489 powerful, so that they can now also be used to annotate e.g. a smart
3490 pointer implementation.
3492 * Helgrind's annotation set has also been drastically improved, so as
3493 to provide to users a general set of annotations to describe locks,
3494 semaphores, barriers and condition variables. Annotations to
3495 describe thread-safe reference counted heap objects have also been
3498 * Memcheck has a new command-line option, --show-possibly-lost, which
3499 is enabled by default. When disabled, the leak detector will not
3500 show possibly-lost blocks.
3502 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3503 has been added. DHAT keeps track of allocated heap blocks, and also
3504 inspects every memory reference to see which block (if any) is being
3505 accessed. This gives a lot of insight into block lifetimes,
3506 utilisation, turnover, liveness, and the location of hot and cold
3507 fields. You can use DHAT to do hot-field profiling.
3509 * ==================== OTHER CHANGES ====================
3511 * Improved support for unfriendly self-modifying code: the extra
3512 overhead incurred by --smc-check=all has been reduced by
3513 approximately a factor of 5 as compared with 3.5.0.
3515 * Ability to show directory names for source files in error messages.
3516 This is combined with a flexible mechanism for specifying which
3517 parts of the paths should be shown. This is enabled by the new flag
3520 * A new flag, --require-text-symbol, which will stop the run if a
3521 specified symbol is not found it a given shared object when it is
3522 loaded into the process. This makes advanced working with function
3523 intercepting and wrapping safer and more reliable.
3525 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3526 and control of Valgrind is now available for the tools Memcheck and
3527 Helgrind. XML output from Valgrind is available for Memcheck,
3528 Helgrind and exp-Ptrcheck.
3530 * More reliable stack unwinding on amd64-linux, particularly in the
3531 presence of function wrappers, and with gcc-4.5 compiled code.
3533 * Modest scalability (performance improvements) for massive
3534 long-running applications, particularly for those with huge amounts
3537 * Support for analyzing programs running under Wine with has been
3538 improved. The header files <valgrind/valgrind.h>,
3539 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3540 Windows-programs compiled with MinGW or one of the Microsoft Visual
3543 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3544 The 32-bit simulator was not affected. This did not occur often,
3545 but when it did would usually crash the program under test.
3548 * A large number of bugs were fixed. These are shown below.
3550 * A number of bugs were investigated, and were candidates for fixing,
3551 but are not fixed in 3.6.0, due to lack of developer time. They may
3552 get fixed in later releases. They are:
3554 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3555 212419 false positive "lock order violated" (A+B vs A)
3556 213685 Undefined value propagates past dependency breaking instruction
3557 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3558 237920 valgrind segfault on fork failure
3559 242137 support for code compiled by LLVM-2.8
3560 242423 Another unknown Intel cache config value
3561 243232 Inconsistent Lock Orderings report with trylock
3562 243483 ppc: callgrind triggers VEX assertion failure
3563 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3564 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3566 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3567 249435 Analyzing wine programs with callgrind triggers a crash
3568 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3569 250065 Handling large allocations
3570 250101 huge "free" memory usage due to m_mallocfree.c
3571 "superblocks fragmentation"
3572 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3573 252091 Callgrind on ARM does not detect function returns correctly
3574 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3575 254420 memory pool tracking broken
3576 n-i-bz support for adding symbols for JIT generated code
3579 The following bugs have been fixed or resolved. Note that "n-i-bz"
3580 stands for "not in bugzilla" -- that is, a bug that was reported to us
3581 but never got a bugzilla entry. We encourage you to file bugs in
3582 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3583 mailing the developers (or mailing lists) directly -- bugs that are
3584 not entered into bugzilla tend to get forgotten about or ignored.
3586 To see details of a given bug, visit
3587 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3588 where XXXXXX is the bug number as listed below.
3590 135264 dcbzl instruction missing
3592 153699 Valgrind should report unaligned reads with movdqa
3594 190429 Valgrind reports lost of errors in ld.so
3595 with x86_64 2.9.90 glibc
3596 197266 valgrind appears to choke on the xmms instruction
3598 197988 Crash when demangling very large symbol names
3599 202315 unhandled syscall: 332 (inotify_init1)
3600 203256 Add page-level profiling to Massif
3601 205093 dsymutil=yes needs quotes, locking (partial fix)
3602 205241 Snow Leopard 10.6 support (partial fix)
3603 206600 Leak checker fails to upgrade indirect blocks when their
3604 parent becomes reachable
3605 210935 port valgrind.h (not valgrind) to win32 so apps run under
3606 wine can make client requests
3607 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3608 within Linux ip-stack checksum functions
3609 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3611 213685 Undefined value propagates past dependency breaking instruction
3613 215914 Valgrind inserts bogus empty environment variable
3615 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3616 222545 shmat fails under valgind on some arm targets
3617 222560 ARM NEON support
3620 232509 Docs build fails with formatting inside <title></title> elements
3622 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3623 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3624 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3625 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3626 237485 symlink (syscall 57) is not supported on Mac OS
3627 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3629 238208 is_just_below_ESP doesn't take into account red-zone
3630 238345 valgrind passes wrong $0 when executing a shell script
3631 238679 mq_timedreceive syscall doesn't flag the reception buffer
3633 238696 fcntl command F_DUPFD_CLOEXEC not supported
3634 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3635 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3636 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
3638 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
3644 242606 unhandled syscall: setegid (in Ptrcheck)
3645 242814 Helgrind "Impossible has happened" during
3646 QApplication::initInstance();
3647 243064 Valgrind attempting to read debug information from iso
3648 243270 Make stack unwinding in Valgrind wrappers more reliable
3649 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
3650 sysno = 277 (mq_open)
3651 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
3652 244493 ARM VFP d16-d31 registers support
3653 244670 add support for audit_session_self syscall on Mac OS 10.6
3654 244921 The xml report of helgrind tool is not well format
3655 244923 In the xml report file, the <preamble> not escape the
3656 xml char, eg '<','&','>'
3657 245535 print full path names in plain text reports
3658 245925 x86-64 red zone handling problem
3659 246258 Valgrind not catching integer underruns + new [] s
3660 246311 reg/reg cmpxchg doesn't work on amd64
3661 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
3662 246888 Improve Makefile.vex.am
3663 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
3664 to [f]chmod_extended
3665 247526 IBM POWER6 (ISA 2.05) support is incomplete
3666 247561 Some leak testcases fails due to reachable addresses in
3668 247875 sizeofIRType to handle Ity_I128
3669 247894 [PATCH] unhandled syscall sys_readahead
3670 247980 Doesn't honor CFLAGS passed to configure
3671 248373 darwin10.supp is empty in the trunk
3672 248822 Linux FIBMAP ioctl has int parameter instead of long
3673 248893 [PATCH] make readdwarf.c big endianess safe to enable
3674 unwinding on big endian systems
3675 249224 Syscall 336 not supported (SYS_proc_info)
3677 249775 Incorrect scheme for detecting NEON capabilities of host CPU
3678 249943 jni JVM init fails when using valgrind
3679 249991 Valgrind incorrectly declares AESKEYGENASSIST support
3681 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
3682 250799 frexp$fenv_access_off function generates SIGILL
3683 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
3684 251251 support pclmulqdq insn
3685 251362 valgrind: ARM: attach to debugger either fails or provokes
3687 251674 Unhandled syscall 294
3690 254257 Add support for debugfiles found by build-id
3691 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
3692 254646 Wrapped functions cause stack misalignment on OS X
3693 (and possibly Linux)
3694 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
3696 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
3700 Release 3.5.0 (19 August 2009)
3701 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3702 3.5.0 is a feature release with many significant improvements and the
3703 usual collection of bug fixes. The main improvement is that Valgrind
3704 now works on Mac OS X.
3706 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
3707 and X86/Darwin. Support for recent distros and toolchain components
3708 (glibc 2.10, gcc 4.5) has been added.
3710 -------------------------
3712 Here is a short summary of the changes. Details are shown further
3715 * Support for Mac OS X (10.5.x).
3717 * Improvements and simplifications to Memcheck's leak checker.
3719 * Clarification and simplifications in various aspects of Valgrind's
3722 * XML output for Helgrind and Ptrcheck.
3724 * Performance and stability improvements for Helgrind and DRD.
3726 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
3728 * A new experimental tool, BBV, useful for computer architecture
3731 * Improved Wine support, including ability to read Windows PDB
3734 -------------------------
3736 Here are details of the above changes, followed by descriptions of
3737 many other minor changes, and a list of fixed bugs.
3740 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
3741 called "Darwin" because that is the name of the OS core, which is the
3742 level that Valgrind works at.)
3746 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
3747 because it would require work and 10.4 is only becoming less common.
3749 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
3750 fairly well. For 10.5.x, 32-bit programs are the default even on
3751 64-bit machines, so it handles most current programs.
3753 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
3754 officially supported, but simple programs at least will probably work.
3755 However, start-up is slow.
3757 - PowerPC machines are not supported.
3759 Things that don't work:
3761 - The Ptrcheck tool.
3763 - Objective-C garbage collection.
3767 - If you have Rogue Amoeba's "Instant Hijack" program installed,
3768 Valgrind will fail with a SIGTRAP at start-up. See
3769 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
3774 - You will likely find --dsymutil=yes a useful option, as error
3775 messages may be imprecise without it.
3777 - Mac OS X support is new and therefore will be less robust than the
3778 Linux support. Please report any bugs you find.
3780 - Threaded programs may run more slowly than on Linux.
3782 Many thanks to Greg Parker for developing this port over several years.
3785 * Memcheck's leak checker has been improved.
3787 - The results for --leak-check=summary now match the summary results
3788 for --leak-check=full. Previously they could differ because
3789 --leak-check=summary counted "indirectly lost" blocks and
3790 "suppressed" blocks as "definitely lost".
3792 - Blocks that are only reachable via at least one interior-pointer,
3793 but are directly pointed to by a start-pointer, were previously
3794 marked as "still reachable". They are now correctly marked as
3797 - The default value for the --leak-resolution option has been
3798 changed from "low" to "high". In general, this means that more
3799 leak reports will be produced, but each leak report will describe
3800 fewer leaked blocks.
3802 - With --leak-check=full, "definitely lost" and "possibly lost"
3803 leaks are now considered as proper errors, ie. they are counted
3804 for the "ERROR SUMMARY" and affect the behaviour of
3805 --error-exitcode. These leaks are not counted as errors if
3806 --leak-check=summary is specified, however.
3808 - Documentation for the leak checker has been improved.
3811 * Various aspects of Valgrind's text output have changed.
3813 - Valgrind's start-up message has changed. It is shorter but also
3814 includes the command being run, which makes it easier to use
3815 --trace-children=yes. An example:
3817 - Valgrind's shut-down messages have also changed. This is most
3818 noticeable with Memcheck, where the leak summary now occurs before
3819 the error summary. This change was necessary to allow leaks to be
3820 counted as proper errors (see the description of the leak checker
3821 changes above for more details). This was also necessary to fix a
3822 longstanding bug in which uses of suppressions against leaks were
3823 not "counted", leading to difficulties in maintaining suppression
3824 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3826 - Behavior of -v has changed. In previous versions, -v printed out
3827 a mixture of marginally-user-useful information, and tool/core
3828 statistics. The statistics printing has now been moved to its own
3829 flag, --stats=yes. This means -v is less verbose and more likely
3830 to convey useful end-user information.
3832 - The format of some (non-XML) stack trace entries has changed a
3833 little. Previously there were six possible forms:
3835 0x80483BF: really (a.c:20)
3836 0x80483BF: really (in /foo/a.out)
3838 0x80483BF: (within /foo/a.out)
3839 0x80483BF: ??? (a.c:20)
3842 The third and fourth of these forms have been made more consistent
3843 with the others. The six possible forms are now:
3845 0x80483BF: really (a.c:20)
3846 0x80483BF: really (in /foo/a.out)
3847 0x80483BF: really (in ???)
3848 0x80483BF: ??? (in /foo/a.out)
3849 0x80483BF: ??? (a.c:20)
3852 Stack traces produced when --xml=yes is specified are different
3856 * Helgrind and Ptrcheck now support XML output, so they can be used
3857 from GUI tools. Also, the XML output mechanism has been
3860 - The XML format has been overhauled and generalised, so it is more
3861 suitable for error reporting tools in general. The Memcheck
3862 specific aspects of it have been removed. The new format, which
3863 is an evolution of the old format, is described in
3864 docs/internals/xml-output-protocol4.txt.
3866 - Memcheck has been updated to use the new format.
3868 - Helgrind and Ptrcheck are now able to emit output in this format.
3870 - The XML output mechanism has been overhauled. XML is now output
3871 to its own file descriptor, which means that:
3873 * Valgrind can output text and XML independently.
3875 * The longstanding problem of XML output being corrupted by
3876 unexpected un-tagged text messages is solved.
3878 As before, the destination for text output is specified using
3879 --log-file=, --log-fd= or --log-socket=.
3881 As before, XML output for a tool is enabled using --xml=yes.
3883 Because there's a new XML output channel, the XML output
3884 destination is now specified by --xml-file=, --xml-fd= or
3887 Initial feedback has shown this causes some confusion. To
3888 clarify, the two envisaged usage scenarios are:
3890 (1) Normal text output. In this case, do not specify --xml=yes
3891 nor any of --xml-file=, --xml-fd= or --xml-socket=.
3893 (2) XML output. In this case, specify --xml=yes, and one of
3894 --xml-file=, --xml-fd= or --xml-socket= to select the XML
3895 destination, one of --log-file=, --log-fd= or --log-socket=
3896 to select the destination for any remaining text messages,
3897 and, importantly, -q.
3899 -q makes Valgrind completely silent on the text channel,
3900 except in the case of critical failures, such as Valgrind
3901 itself segfaulting, or failing to read debugging information.
3902 Hence, in this scenario, it suffices to check whether or not
3903 any output appeared on the text channel. If yes, then it is
3904 likely to be a critical error which should be brought to the
3905 attention of the user. If no (the text channel produced no
3906 output) then it can be assumed that the run was successful.
3908 This allows GUIs to make the critical distinction they need to
3909 make (did the run fail or not?) without having to search or
3910 filter the text output channel in any way.
3912 It is also recommended to use --child-silent-after-fork=yes in
3916 * Improvements and changes in Helgrind:
3918 - XML output, as described above
3920 - Checks for consistent association between pthread condition
3921 variables and their associated mutexes are now performed.
3923 - pthread_spinlock functions are supported.
3925 - Modest performance improvements.
3927 - Initial (skeletal) support for describing the behaviour of
3928 non-POSIX synchronisation objects through ThreadSanitizer
3929 compatible ANNOTATE_* macros.
3931 - More controllable tradeoffs between performance and the level of
3932 detail of "previous" accesses in a race. There are now three
3935 * --history-level=full. This is the default, and was also the
3936 default in 3.4.x. It shows both stacks involved in a race, but
3937 requires a lot of memory and can be very slow in programs that
3938 do many inter-thread synchronisation events.
3940 * --history-level=none. This only shows the later stack involved
3941 in a race. This can be much faster than --history-level=full,
3942 but makes it much more difficult to find the other access
3943 involved in the race.
3945 The new intermediate setting is
3947 * --history-level=approx
3949 For the earlier (other) access, two stacks are presented. The
3950 earlier access is guaranteed to be somewhere in between the two
3951 program points denoted by those stacks. This is not as useful
3952 as showing the exact stack for the previous access (as per
3953 --history-level=full), but it is better than nothing, and it's
3954 almost as fast as --history-level=none.
3957 * New features and improvements in DRD:
3959 - The error messages printed by DRD are now easier to interpret.
3960 Instead of using two different numbers to identify each thread
3961 (Valgrind thread ID and DRD thread ID), DRD does now identify
3962 threads via a single number (the DRD thread ID). Furthermore
3963 "first observed at" information is now printed for all error
3964 messages related to synchronization objects.
3966 - Added support for named semaphores (sem_open() and sem_close()).
3968 - Race conditions between pthread_barrier_wait() and
3969 pthread_barrier_destroy() calls are now reported.
3971 - Added support for custom allocators through the macros
3972 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
3973 in <valgrind/valgrind.h>). An alternative for these two macros is
3974 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
3977 - Added support for annotating non-POSIX synchronization objects
3978 through several new ANNOTATE_*() macros.
3980 - OpenMP: added support for the OpenMP runtime (libgomp) included
3981 with gcc versions 4.3.0 and 4.4.0.
3985 - Added two new command-line options (--first-race-only and
3986 --segment-merging-interval).
3989 * Genuinely atomic support for x86/amd64/ppc atomic instructions
3991 Valgrind will now preserve (memory-access) atomicity of LOCK-
3992 prefixed x86/amd64 instructions, and any others implying a global
3993 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
3995 This means that Valgrinded processes will "play nicely" in
3996 situations where communication with other processes, or the kernel,
3997 is done through shared memory and coordinated with such atomic
3998 instructions. Prior to this change, such arrangements usually
3999 resulted in hangs, races or other synchronisation failures, because
4000 Valgrind did not honour atomicity of such instructions.
4003 * A new experimental tool, BBV, has been added. BBV generates basic
4004 block vectors for use with the SimPoint analysis tool, which allows
4005 a program's overall behaviour to be approximated by running only a
4006 fraction of it. This is useful for computer architecture
4007 researchers. You can run BBV by specifying --tool=exp-bbv (the
4008 "exp-" prefix is short for "experimental"). BBV was written by
4012 * Modestly improved support for running Windows applications under
4013 Wine. In particular, initial support for reading Windows .PDB debug
4014 information has been added.
4017 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
4018 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
4022 * The Valgrind client requests VALGRIND_PRINTF and
4023 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
4024 the string was always printed immediately on its own line. Now, the
4025 string will be added to a buffer but not printed until a newline is
4026 encountered, or other Valgrind output is printed (note that for
4027 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
4028 "other Valgrind output"). This allows you to use multiple
4029 VALGRIND_PRINTF calls to build up a single output line, and also to
4030 print multiple output lines with a single request (by embedding
4031 multiple newlines in the string).
4034 * The graphs drawn by Massif's ms_print program have changed slightly:
4036 - The half-height chars '.' and ',' are no longer drawn, because
4037 they are confusing. The --y option can be used if the default
4038 y-resolution is not high enough.
4040 - Horizontal lines are now drawn after the top of a snapshot if
4041 there is a gap until the next snapshot. This makes it clear that
4042 the memory usage has not dropped to zero between snapshots.
4045 * Something that happened in 3.4.0, but wasn't clearly announced: the
4046 option --read-var-info=yes can be used by some tools (Memcheck,
4047 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
4048 variable type and location information. This makes those tools
4049 start up more slowly and increases memory consumption, but
4050 descriptions of data addresses in error messages become more
4054 * exp-Omega, an experimental instantaneous leak-detecting tool, was
4055 disabled in 3.4.0 due to a lack of interest and maintenance,
4056 although the source code was still in the distribution. The source
4057 code has now been removed from the distribution. For anyone
4058 interested, the removal occurred in SVN revision r10247.
4061 * Some changes have been made to the build system.
4063 - VEX/ is now integrated properly into the build system. This means
4064 that dependency tracking within VEX/ now works properly, "make
4065 install" will work without requiring "make" before it, and
4066 parallel builds (ie. 'make -j') now work (previously a
4067 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
4068 was effectively ignored).
4070 - The --with-vex configure option has been removed. It was of
4071 little use and removing it simplified the build system.
4073 - The location of some install files has changed. This should not
4074 affect most users. Those who might be affected:
4076 * For people who use Valgrind with MPI programs, the installed
4077 libmpiwrap.so library has moved from
4078 $(INSTALL)/<platform>/libmpiwrap.so to
4079 $(INSTALL)/libmpiwrap-<platform>.so.
4081 * For people who distribute standalone Valgrind tools, the
4082 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
4083 have moved to $(INSTALL)/libcoregrind-<platform>.a.
4085 These changes simplify the build system.
4087 - Previously, all the distributed suppression (*.supp) files were
4088 installed. Now, only default.supp is installed. This should not
4089 affect users as the other installed suppression files were not
4090 read; the fact that they were installed was a mistake.
4093 * KNOWN LIMITATIONS:
4095 - Memcheck is unusable with the Intel compiler suite version 11.1,
4096 when it generates code for SSE2-and-above capable targets. This
4097 is because of icc's use of highly optimised inlined strlen
4098 implementations. It causes Memcheck to report huge numbers of
4099 false errors even in simple programs. Helgrind and DRD may also
4102 Versions 11.0 and earlier may be OK, but this has not been
4106 The following bugs have been fixed or resolved. Note that "n-i-bz"
4107 stands for "not in bugzilla" -- that is, a bug that was reported to us
4108 but never got a bugzilla entry. We encourage you to file bugs in
4109 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4110 mailing the developers (or mailing lists) directly -- bugs that are
4111 not entered into bugzilla tend to get forgotten about or ignored.
4113 To see details of a given bug, visit
4114 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4115 where XXXXXX is the bug number as listed below.
4117 84303 How about a LockCheck tool?
4118 91633 dereference of null ptr in vgPlain_st_basetype
4119 97452 Valgrind doesn't report any pthreads problems
4120 100628 leak-check gets assertion failure when using
4121 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4122 108528 NPTL pthread cleanup handlers not called
4123 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4124 110128 mallinfo is not implemented...
4125 110770 VEX: Generated files not always updated when making valgrind
4126 111102 Memcheck: problems with large (memory footprint) applications
4127 115673 Vex's decoder should never assert
4128 117564 False positive: Syscall param clone(child_tidptr) contains
4129 uninitialised byte(s)
4130 119404 executing ssh from inside valgrind fails
4131 133679 Callgrind does not write path names to sources with dwarf debug
4133 135847 configure.in problem with non gnu compilers (and possible fix)
4134 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4135 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4136 136230 memcheck reports "possibly lost", should be "still reachable"
4137 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4138 137904 Valgrind reports a memory leak when using POSIX threads,
4140 139076 valgrind VT_GETSTATE error
4141 142228 complaint of elf_dynamic_do_rela in trivial usage
4142 145347 spurious warning with USBDEVFS_REAPURB
4143 148441 (wine) can't find memory leak in Wine, win32 binary
4145 148742 Leak-check fails assert on exit
4146 149878 add (proper) check for calloc integer overflow
4147 150606 Call graph is broken when using callgrind control
4148 152393 leak errors produce an exit code of 0. I need some way to
4149 cause leak errors to result in a nonzero exit code.
4150 157154 documentation (leak-resolution doc speaks about num-callers
4151 def=4) + what is a loss record
4152 159501 incorrect handling of ALSA ioctls
4153 162020 Valgrinding an empty/zero-byte file crashes valgrind
4154 162482 ppc: Valgrind crashes while reading stabs information
4155 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4156 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4157 163560 VEX/test_main.c is missing from valgrind-3.3.1
4158 164353 malloc_usable_size() doesn't return a usable size
4159 165468 Inconsistent formatting in memcheck manual -- please fix
4160 169505 main.c:286 (endOfInstr):
4161 Assertion 'ii->cost_offset == *cost_offset' failed
4162 177206 Generate default.supp during compile instead of configure
4163 177209 Configure valt_load_address based on arch+os
4164 177305 eventfd / syscall 323 patch lost
4165 179731 Tests fail to build because of inlining of non-local asm labels
4166 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4167 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4168 181594 Bogus warning for empty text segment
4169 181707 dwarf doesn't require enumerations to have name
4170 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4171 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4172 Assertion '!already_present' failed.
4173 185359 exp-ptrcheck: unhandled syscall getresuid()
4174 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4175 185816 Valgrind is unable to handle debug info for files with split
4176 debug info that are prelinked afterwards
4177 185980 [darwin] unhandled syscall: sem_open
4178 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4179 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4180 186790 Suppression pattern used for leaks are not reported
4181 186796 Symbols with length>200 in suppression files are ignored
4182 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4183 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4184 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4185 188046 bashisms in the configure script
4186 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4187 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4188 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4189 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4190 assertion fail "!lock->heldBy"
4191 188427 Add support for epoll_create1 (with patch)
4192 188530 Support for SIOCGSTAMPNS
4193 188560 Include valgrind.spec in the tarball
4194 188572 Valgrind on Mac should suppress setenv() mem leak
4195 189054 Valgrind fails to build because of duplicate non-local asm labels
4196 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4197 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4198 189763 drd assertion failure: s_threadinfo[tid].is_recording
4199 190219 unhandled syscall: 328 (x86-linux)
4200 190391 dup of 181394; see above
4201 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4202 190820 No debug information on powerpc-linux
4203 191095 PATCH: Improve usbdevfs ioctl handling
4204 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4206 191189 --xml=yes should obey --gen-suppressions=all
4207 191192 syslog() needs a suppression on macosx
4208 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4209 191761 getrlimit on MacOSX
4210 191992 multiple --fn-skip only works sometimes; dependent on order
4211 192634 V. reports "aspacem sync_check_mapping_callback:
4212 segment mismatch" on Darwin
4213 192954 __extension__ missing on 2 client requests
4214 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4215 194474 "INSTALL" file has different build instructions than "README"
4216 194671 Unhandled syscall (sem_wait?) from mac valgrind
4217 195069 memcheck: reports leak (memory still reachable) for
4219 195169 drd: (vgDrd_barrier_post_wait):
4220 Assertion 'r->sg[p->post_iteration]' failed.
4221 195268 valgrind --log-file doesn't accept ~/...
4222 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4223 195860 WARNING: unhandled syscall: unix:223
4224 196528 need a error suppression for pthread_rwlock_init under os x?
4225 197227 Support aio_* syscalls on Darwin
4226 197456 valgrind should reject --suppressions=(directory)
4227 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4228 197591 unhandled syscall 27 (mincore)
4229 197793 Merge DCAS branch to the trunk == 85756, 142103
4230 197794 Avoid duplicate filenames in Vex
4231 197898 make check fails on current SVN
4232 197901 make check fails also under exp-ptrcheck in current SVN
4233 197929 Make --leak-resolution=high the default
4234 197930 Reduce spacing between leak reports
4235 197933 Print command line of client at start-up, and shorten preamble
4236 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4237 198395 add BBV to the distribution as an experimental tool
4238 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4239 198649 callgrind_annotate doesn't cumulate counters
4240 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4241 199977 Valgrind complains about an unrecognized instruction in the
4242 atomic_incs test program
4243 200029 valgrind isn't able to read Fedora 12 debuginfo
4244 200760 darwin unhandled syscall: unix:284
4245 200827 DRD doesn't work on Mac OS X
4246 200990 VG_(read_millisecond_timer)() does not work correctly
4247 201016 Valgrind does not support pthread_kill() on Mac OS
4248 201169 Document --read-var-info
4249 201323 Pre-3.5.0 performance sanity checking
4250 201384 Review user manual for the 3.5.0 release
4251 201585 mfpvr not implemented on ppc
4252 201708 tests failing because x86 direction flag is left set
4253 201757 Valgrind doesn't handle any recent sys_futex additions
4254 204377 64-bit valgrind can not start a shell script
4255 (with #!/path/to/shell) if the shell is a 32-bit executable
4256 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4257 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4258 about memory allocation events with memory access tracing enabled
4259 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4261 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4265 Release 3.4.1 (28 February 2009)
4266 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4267 3.4.1 is a bug-fix release that fixes some regressions and assertion
4268 failures in debug info reading in 3.4.0, most notably incorrect stack
4269 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4270 other debug info problems are also fixed. A number of bugs in the
4271 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4273 In view of the fact that 3.4.0 contains user-visible regressions
4274 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4275 encouraged to ship 3.4.1 in preference to 3.4.0.
4277 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4278 bugzilla" -- that is, a bug that was reported to us but never got a
4279 bugzilla entry. We encourage you to file bugs in bugzilla
4280 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4281 developers (or mailing lists) directly -- bugs that are not entered
4282 into bugzilla tend to get forgotten about or ignored.
4284 n-i-bz Fix various bugs reading icc-11 generated debug info
4285 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4286 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4287 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4288 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4289 179618 exp-ptrcheck crashed / exit prematurely
4290 179624 helgrind: false positive races with pthread_create and
4291 recv/open/close/read
4292 134207 pkg-config output contains @VG_PLATFORM@
4293 176926 floating point exception at valgrind startup with PPC 440EPX
4294 181594 Bogus warning for empty text segment
4295 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4296 181707 Dwarf3 doesn't require enumerations to have name
4297 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4298 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4299 Assertion '!already_present' failed.
4300 185359 exp-ptrcheck unhandled syscall getresuid()
4302 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4303 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4307 Release 3.4.0 (2 January 2009)
4308 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4309 3.4.0 is a feature release with many significant improvements and the
4310 usual collection of bug fixes. This release supports X86/Linux,
4311 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4312 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4314 3.4.0 brings some significant tool improvements. Memcheck can now
4315 report the origin of uninitialised values, the thread checkers
4316 Helgrind and DRD are much improved, and we have a new experimental
4317 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4318 global arrays. In detail:
4320 * Memcheck is now able to track the origin of uninitialised values.
4321 When it reports an uninitialised value error, it will try to show
4322 the origin of the value, as either a heap or stack allocation.
4323 Origin tracking is expensive and so is not enabled by default. To
4324 use it, specify --track-origins=yes. Memcheck's speed will be
4325 essentially halved, and memory usage will be significantly
4326 increased. Nevertheless it can drastically reduce the effort
4327 required to identify the root cause of uninitialised value errors,
4328 and so is often a programmer productivity win, despite running more
4331 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4332 3.4.0, will be released shortly.
4334 * Helgrind's race detection algorithm has been completely redesigned
4335 and reimplemented, to address usability and scalability concerns:
4337 - The new algorithm has a lower false-error rate: it is much less
4338 likely to report races that do not really exist.
4340 - Helgrind will display full call stacks for both accesses involved
4341 in a race. This makes it easier to identify the root causes of
4344 - Limitations on the size of program that can run have been removed.
4346 - Performance has been modestly improved, although that is very
4349 - Direct support for Qt4 threading has been added.
4351 - pthread_barriers are now directly supported.
4353 - Helgrind works well on all supported Linux targets.
4355 * The DRD thread debugging tool has seen major improvements:
4357 - Greatly improved performance and significantly reduced memory
4360 - Support for several major threading libraries (Boost.Thread, Qt4,
4361 glib, OpenMP) has been added.
4363 - Support for atomic instructions, POSIX semaphores, barriers and
4364 reader-writer locks has been added.
4366 - Works now on PowerPC CPUs too.
4368 - Added support for printing thread stack usage at thread exit time.
4370 - Added support for debugging lock contention.
4372 - Added a manual for Drd.
4374 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4375 checks for misuses of pointers. In that sense it is a bit like
4376 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4377 detect overruns of stack and global arrays, it can detect
4378 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4379 detect accesses heap blocks that have been freed a very long time
4380 ago (millions of blocks in the past).
4382 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4383 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4384 of the main Valgrind documentation. As this is an experimental
4385 tool, we would be particularly interested in hearing about your
4386 experiences with it.
4388 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4389 longer built by default, although the code remains in the repository
4390 and the tarball. This is due to three factors: a perceived lack of
4391 users, a lack of maintenance, and concerns that it may not be
4392 possible to achieve reliable operation using the existing design.
4394 * As usual, support for the latest Linux distros and toolchain
4395 components has been added. It should work well on Fedora Core 10,
4396 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4397 state) is supported, as is glibc-2.9. The C++ demangler has been
4398 updated so as to work well with C++ compiled by even the most recent
4401 * You can now use frame-level wildcards in suppressions. This was a
4402 frequently-requested enhancement. A line "..." in a suppression now
4403 matches zero or more frames. This makes it easier to write
4404 suppressions which are precise yet insensitive to changes in
4407 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4409 * Very basic support for IBM Power6 has been added (64-bit processes only).
4411 * Valgrind is now cross-compilable. For example, it is possible to
4412 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4413 on a ppc32/64-linux target.
4415 * You can set the main thread's stack size at startup using the
4416 new --main-stacksize= flag (subject of course to ulimit settings).
4417 This is useful for running apps that need a lot of stack space.
4419 * The limitation that you can't use --trace-children=yes together
4420 with --db-attach=yes has been removed.
4422 * The following bugs have been fixed. Note that "n-i-bz" stands for
4423 "not in bugzilla" -- that is, a bug that was reported to us but
4424 never got a bugzilla entry. We encourage you to file bugs in
4425 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4426 mailing the developers (or mailing lists) directly.
4428 n-i-bz Make return types for some client requests 64-bit clean
4429 n-i-bz glibc 2.9 support
4430 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4431 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4432 n-i-bz Building in an env without gdb gives bogus gdb attach
4433 92456 Tracing the origin of uninitialised memory
4434 106497 Valgrind does not demangle some C++ template symbols
4436 151612 Suppression with "..." (frame-level wildcards in .supp files)
4437 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4438 159285 unhandled syscall:25 (stime, on x86-linux)
4439 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4440 160954 ppc build of valgrind crashes with illegal instruction (isel)
4441 160956 mallinfo implementation, w/ patch
4442 162092 Valgrind fails to start gnome-system-monitor
4443 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4444 163794 assertion failure with "--track-origins=yes"
4445 163933 sigcontext.err and .trapno must be set together
4446 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4447 164476 Missing kernel module loading system calls
4448 164669 SVN regression: mmap() drops posix file locks
4449 166581 Callgrind output corruption when program forks
4450 167288 Patch file for missing system calls on Cell BE
4451 168943 unsupported scas instruction pentium
4452 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4453 172417 x86->IR: 0x82 ...
4454 172563 amd64->IR: 0xD9 0xF5 - fprem1
4455 173099 .lds linker script generation error
4456 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4457 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4459 174908 --log-file value not expanded correctly for core file
4460 175044 Add lookup_dcookie for amd64
4461 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4463 Developer-visible changes:
4465 * Valgrind's debug-info reading machinery has been majorly overhauled.
4466 It can now correctly establish the addresses for ELF data symbols,
4467 which is something that has never worked properly before now.
4469 Also, Valgrind can now read DWARF3 type and location information for
4470 stack and global variables. This makes it possible to use the
4471 framework to build tools that rely on knowing the type and locations
4472 of stack and global variables, for example exp-Ptrcheck.
4474 Reading of such information is disabled by default, because most
4475 tools don't need it, and because it is expensive in space and time.
4476 However, you can force Valgrind to read it, using the
4477 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4478 make use of such information, if present, to provide source-level
4479 descriptions of data addresses in the error messages they create.
4481 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4482 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).