1 Release 3.17.0 (18 Mar 2021)
2 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4 3.17.0 is a primarily a bug-fix release. There are also some architecture and
5 tool updates. Support for recent toolchains -- GCC 11 and Clang 11 -- has
8 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
9 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
10 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
11 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
12 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
14 * ==================== CORE CHANGES ===================
16 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
19 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
20 debugging information. When a debuginfo file cannot be found locally,
21 Valgrind is able to query debuginfod servers for the file using its
22 build-id. See the user manual for more information about debuginfod support.
24 * ================== PLATFORM CHANGES =================
28 - Inaccuracies resulting from double-rounding in the simulation of
29 floating-point multiply-add/subtract instructions have been fixed. These
30 should now behave exactly as the hardware does.
32 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
33 ongoing. Support for the half-word variants of at least the following
34 instructions has been added:
36 FABS <Vd>.<T>, <Vn>.<T>
38 FNEG <Vd>.<T>, <Vn>.<T>
40 FSQRT <Vd>.<T>, <Vn>.<T>
45 - Implement the new instructions/features that were added to z/Architecture
46 with the vector-enhancements facility 1. Also cover the instructions from
47 the vector-packed-decimal facility that are defined outside the chapter
48 "Vector Decimal Instructions", but not the ones from that chapter itself.
50 For a detailed list of newly supported instructions see the updates to
51 `docs/internals/s390-opcodes.csv'.
53 Since the miscellaneous instruction extensions facility 2 was already
54 added in Valgrind 3.16.0, this completes the support necessary to run
55 general programs built with `--march=z14' under Valgrind. The
56 vector-packed-decimal facility is currently not exploited by the standard
57 toolchain and libraries.
61 - Various bug fixes. Fix for the sync field to limit setting just two of
62 the two bits in the L-field. Fix the write size for the stxsibx and
63 stxsihx instructions. Fix the modsw and modsd instructions.
65 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
66 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
67 conversion instructions are still missing.
69 * ==================== TOOL CHANGES ====================
71 * General tool changes
73 - All the tools and their vgpreload libraries are now installed under
74 libexec because they cannot be executed directly and should be run through
75 the valgrind executable. This should be an internal, not user visible,
76 change, but might impact valgrind packagers.
78 - The --track-fds option now respects -q, --quiet and won't output anything
79 if no file descriptors are leaked. It also won't report the standard stdin
80 (0), stdout (1) or stderr (2) descriptors as being leaked with
81 --trace-fds=yes anymore. To track whether the standard file descriptors
82 are still open at the end of the program run use --trace-fds=all.
86 - DHAT has been extended, with two new modes of operation. The new
87 --mode=copy flag triggers copy profiling, which records calls to memcpy,
88 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
89 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
90 the new dhat/dhat.h file. This is useful for learning more about hot code
91 paths. See the user manual for more information about the new modes.
93 - Because of these changes, DHAT's file format has changed. DHAT output
94 files produced with earlier versions of DHAT will not work with this
95 version of DHAT's viewer, and DHAT output files produced with this version
96 of DHAT will not work with earlier versions of DHAT's viewer.
106 * ==================== OTHER CHANGES ====================
108 * ==================== FIXED BUGS ====================
110 The following bugs have been fixed or resolved. Note that "n-i-bz"
111 stands for "not in bugzilla" -- that is, a bug that was reported to us
112 but never got a bugzilla entry. We encourage you to file bugs in
113 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
114 than mailing the developers (or mailing lists) directly -- bugs that
115 are not entered into bugzilla tend to get forgotten about or ignored.
117 To see details of a given bug, visit
118 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
119 where XXXXXX is the bug number as listed below.
121 140178 open("/proc/self/exe", ...); doesn't quite work
122 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
123 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
124 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
125 345077 linux syscall execveat support (linux 3.19)
126 361770 Missing F_ADD_SEALS
127 369029 handle linux syscalls sched_getattr and sched_setattr
128 384729 __libc_freeres inhibits cross-platform valgrind
129 388787 Support for C++17 new/delete
130 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
131 396656 Warnings while reading debug info
132 397605 ioctl FICLONE mishandled
133 401416 Compile failure with openmpi 4.0
134 408663 Suppression file for musl libc
135 404076 s390x: z14 vector instructions not implemented
136 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
137 (actually affected all x86 and nanomips regardless of host bitness)
138 413547 regression test does not check for Arm 64 features.
139 414268 Enable AArch64 feature detection and decoding for v8.x instructions
140 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
141 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
142 422261 platform selection fails for unqualified client name
143 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
144 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
145 423195 PPC ISA 3.1 support is missing, part 1
146 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
147 424012 crash with readv/writev having invalid but not NULL arg2 iovec
148 424298 amd64: Implement RDSEED
149 425232 PPC ISA 3.1 support is missing, part 2
150 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
151 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
152 426123 PPC ISA 3.1 support is missing, part 3
153 426144 Fix "condition variable has not been initialized" on Fedora 33.
154 427400 PPC ISA 3.1 support is missing, part 4
155 427401 PPC ISA 3.1 support is missing, part 5
156 427404 PPC ISA 3.1 support is missing, part 6
157 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
158 427787 Support new faccessat2 linux syscall (439)
159 427969 debuginfo section duplicates a section in the main ELF file
160 428035 drd: Unbreak the musl build
161 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
162 428716 cppcheck detects potential leak in VEX/useful/smchash.c
163 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
164 429352 PPC ISA 3.1 support is missing, part 7
165 429354 PPC ISA 3.1 support is missing, part 8
166 429692 unhandled ppc64le-linux syscall: 147 (getsid)
167 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
169 429952 Errors when building regtest with clang
170 430354 ppc stxsibx and stxsihx instructions write too much data
171 430429 valgrind.h doesn't compile on s390x with clang
172 430485 expr_is_guardable doesn't handle Iex_Qop
173 431556 Complete arm64 FADDP v8.2 instruction support
174 432102 Add support for DWARF5 as produced by GCC11
175 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
176 432381 drd: Process STACK_REGISTER client requests
177 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
178 432672 vg_regtest: test-specific environment variables not reset between tests
179 432809 VEX should support REX.W + POPF
180 432861 PPC modsw and modsd give incorrect results for 1 mod 12
181 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
182 432215 Add debuginfod functionality
183 433323 Use pkglibexecdir as vglibdir
184 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
185 433629 valgrind/README has type "abd" instead of "and"
186 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
187 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
188 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
189 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
190 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
192 (3.17.0.RC1: 13 Mar 2021)
193 (3.17.0.RC2: 17 Mar 2021)
197 Release 3.16.1 (22 June 2020)
198 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
200 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
201 fixes character encoding problems in the documentation HTML.
203 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
204 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
206 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
210 Release 3.16.0 (27 May 2020)
211 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
213 3.16.0 is a feature release with many improvements and the usual collection of
216 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
217 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
218 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
219 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
220 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
222 * ==================== CORE CHANGES ===================
224 * It is now possible to dynamically change the value of many command line
225 options while your program (or its children) are running under Valgrind.
227 To see the list of dynamically changeable options, run
228 "valgrind --help-dyn-options".
230 You can change the options from the shell by using vgdb to launch
231 the monitor command "v.clo <clo option>...".
232 The same monitor command can be used from a gdb connected
233 to the valgrind gdbserver.
234 Your program can also change the dynamically changeable options using
235 the client request VALGRIND_CLO_CHANGE(option).
237 * ================== PLATFORM CHANGES =================
239 * MIPS: preliminary support for nanoMIPS instruction set has been added.
241 * ==================== TOOL CHANGES ====================
245 - The implicit memcpy done by each call to realloc now counts towards the
246 read and write counts of resized heap blocks, making those counts higher
251 - cg_annotate's --auto and --show-percs options now default to 'yes', because
252 they are usually wanted.
256 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
257 because they are usually wanted.
259 - The command option --collect-systime has been enhanced to specify
260 the unit used to record the elapsed time spent during system calls.
261 The command option now accepts the values no|yes|msec|usec|nsec,
262 where yes is a synonym of msec. When giving the value nsec, the
263 system cpu time of system calls is also recorded.
267 - Several memcheck options are now dynamically changeable.
268 Use valgrind --help-dyn-options to list them.
270 - The release 3.15 introduced a backward incompatible change for
271 some suppression entries related to preadv and pwritev syscalls.
272 When reading a suppression entry using the unsupported 3.14 format,
273 valgrind will now produce a warning to say the suppression entry will not
274 work, and suggest the needed change.
276 - Significantly fewer false positive errors on optimised code generated by
277 Clang and GCC. In particular, Memcheck now deals better with the
278 situation where the compiler will transform C-level "A && B" into "B && A"
279 under certain circumstances (in which the transformation is valid).
280 Handling of integer equality/non-equality checks on partially defined
281 values is also improved on some architectures.
285 - The exprimental Stack and Global Array Checking tool has been removed.
286 It only ever worked on x86 and amd64, and even on those it had a
287 high false positive rate and was slow. An alternative for detecting
288 stack and global array overruns is using the AddressSanitizer (ASAN)
289 facility of the GCC and Clang compilers, which require you to rebuild
290 your code with -fsanitize=address.
292 * ==================== OTHER CHANGES ====================
294 * New and modified GDB server monitor features:
296 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
298 - The gdbserver monitor commands that require an address and an optional
299 length argument now accepts the alternate 'C like' syntax "address[length]".
300 For example, the memcheck command "monitor who_points_at 0x12345678 120"
301 can now also be given as "monitor who_points_at 0x12345678[120]".
303 * ==================== FIXED BUGS ====================
305 The following bugs have been fixed or resolved. Note that "n-i-bz"
306 stands for "not in bugzilla" -- that is, a bug that was reported to us
307 but never got a bugzilla entry. We encourage you to file bugs in
308 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
309 than mailing the developers (or mailing lists) directly -- bugs that
310 are not entered into bugzilla tend to get forgotten about or ignored.
312 To see details of a given bug, visit
313 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
314 where XXXXXX is the bug number as listed below.
316 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
317 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
318 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
319 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
320 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
321 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
322 400872 Add nanoMIPS support to Valgrind
323 403212 drd/tests/trylock hangs on FreeBSD
324 404406 s390x: z14 miscellaneous instructions not implemented
325 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
326 406561 mcinfcallWSRU gdbserver_test fails on ppc64
327 406824 Unsupported baseline
328 407218 Add support for the copy_file_range syscall
329 407307 Intercept stpcpy also in ld.so for arm64
330 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
332 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
333 408009 Expose rdrand and f16c even on avx if host cpu supports them
334 408091 Missing pkey syscalls
335 408414 Add support for missing for preadv2 and pwritev2 syscalls
336 409141 Valgrind hangs when SIGKILLed
337 409206 Support for Linux PPS and PTP ioctls
338 409367 exit_group() after signal to thread waiting in futex() causes hangs
339 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
340 409780 References to non-existent configure.in
341 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
342 410599 Non-deterministic behaviour of pth_self_kill_15_other test
343 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
344 411134 Allow the user to change a set of command line options during execution
345 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
346 412344 Problem setting mips flags with specific paths
347 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
348 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
349 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
350 413603 callgrind_annotate/cg_annotate truncate function names at '#'
351 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
352 415136 ARMv8.1 Compare-and-Swap instructions are not supported
353 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
354 416239 valgrind crashes when handling clock_adjtime
355 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
356 416286 DRD reports "conflicting load" error on std::mutex::lock()
357 416301 s390x: "compare and signal" not supported
358 416387 finit_module and bpf syscalls are unhandled on arm64
359 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
360 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
361 416753 new 32bit time syscalls for 2038+
362 417075 pwritev(vector[...]) suppression ignored
363 417075 is not fixed, but incompatible supp entries are detected
364 and a warning is produced for these.
365 417187 [MIPS] Conditional branch problem since 'grail' changes
366 417238 Test memcheck/tests/vbit-test fails on mips64 BE
367 417266 Make memcheck/tests/linux/sigqueue usable with musl
368 417281 s390x: /bin/true segfaults with "grail" enabled
369 417427 commit to fix vki_siginfo_t definition created numerous regression
371 417452 s390_insn_store_emit: dst->tag for HRcVec128
372 417578 Add suppressions for glibc DTV leaks
373 417906 clone with CLONE_VFORK and no CLONE_VM fails
374 418004 Grail code additions break ppc64.
375 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
376 418997 s390x: Support Iex_ITE for float and vector types
377 419503 s390x: Avoid modifying registers returned from isel functions
378 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
379 421570 std_mutex fails on Arm v8.1 h/w
380 434035 vgdb might crash if valgrind is killed
381 n-i-bz Fix minor one time leaks in dhat.
382 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
383 n-i-bz Add support for the Linux io_uring system calls
384 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
385 n-i-bz Fix non-glibc build of test suite with s390x_features
386 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
387 423195 PPC ISA 3.1 support is missing, part 1
389 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
390 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
391 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
395 Release 3.15.0 (12 April 2019)
396 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
398 3.15.0 is a feature release with many improvements and the usual collection of
401 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
402 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
403 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
404 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
405 support for X86/macOS 10.13 and AMD64/macOS 10.13.
407 * ==================== CORE CHANGES ===================
409 * The XTree Massif output format now makes use of the information obtained
410 when specifying --read-inline-info=yes.
412 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
414 * ==================== TOOL CHANGES ====================
418 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
419 it has been promoted from an experimental tool to a regular tool. Run it
420 with --tool=dhat instead of --tool=exp-dhat.
422 - DHAT now prints only minimal data when the program ends, instead writing
423 the bulk of the profiling data to a file. As a result, the --show-top-n
424 and --sort-by options have been removed.
426 - Profile results can be viewed with the new viewer, dh_view.html. When
427 a run ends, a short message is printed, explaining how to view the result.
429 - See the documentation for more details.
433 - cg_annotate has a new option, --show-percs, which prints percentages next
438 - callgrind_annotate has a new option, --show-percs, which prints percentages
439 next to all event counts.
441 - callgrind_annotate now inserts commas in call counts, and
442 sort the caller/callee lists in the call tree.
446 - The default value for --read-inline-info is now "yes" on
447 Linux/Android/Solaris. It is still "no" on other OS.
451 - The option --xtree-leak=yes (to output leak result in xtree format)
452 automatically activates the option --show-leak-kinds=all, as xtree
453 visualisation tools such as kcachegrind can in any case select what kind
454 of leak to visualise.
456 - There has been further work to avoid false positives. In particular,
457 integer equality on partially defined inputs (C == and !=) is now handled
460 * ==================== OTHER CHANGES ====================
462 * The new option --show-error-list=no|yes displays, at the end of the run, the
463 list of detected errors and the used suppressions. Prior to this change,
464 showing this information could only be done by specifying "-v -v", but that
465 also produced a lot of other possibly-non-useful messages. The option -s is
466 equivalent to --show-error-list=yes.
468 * ==================== FIXED BUGS ====================
470 The following bugs have been fixed or resolved. Note that "n-i-bz"
471 stands for "not in bugzilla" -- that is, a bug that was reported to us
472 but never got a bugzilla entry. We encourage you to file bugs in
473 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
474 than mailing the developers (or mailing lists) directly -- bugs that
475 are not entered into bugzilla tend to get forgotten about or ignored.
477 To see details of a given bug, visit
478 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
479 where XXXXXX is the bug number as listed below.
481 385411 s390x: z13 vector floating-point instructions not implemented
482 397187 z13 vector register support for vgdb gdbserver
483 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
484 398870 Please add support for instruction vcvtps2ph
485 399287 amd64 front end: Illegal Instruction vcmptrueps
486 399301 Use inlined frames in Massif XTree output.
487 399322 Improve callgrind_annotate output
488 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
489 400164 helgrind test encounters mips x-compiler warnings and assembler error
490 400490 s390x: VRs allocated as if separate from FPRs
491 400491 s390x: Operand of LOCH treated as unsigned integer
492 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
493 options, which specify a mips64 processor
494 401112 LLVM 5.0 generates comparison against partially initialized data
495 401277 More bugs in z13 support
496 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
497 401578 drd: crashes sometimes on fork()
498 401627 memcheck errors with glibc avx2 optimized wcsncmp
499 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
500 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
501 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
503 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
504 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
505 402123 invalid assembler opcodes for mips32r2
506 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
507 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
508 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
509 402351 mips64 libvexmultiarch_test fails on s390x
511 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
512 402480 Do not use %rsp in clobber list
513 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
514 402515 Implement new option --show-error-list=no|yes / -s
515 402519 POWER 3.0 addex instruction incorrectly implemented
516 402781 Redo the cache used to process indirect branch targets
517 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
518 403552 s390x: wrong facility bit checked for vector facility
519 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
520 404638 Add VG_(replaceIndexXA)
521 404843 s390x: backtrace sometimes ends prematurely
522 404888 autotools cleanup series
523 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
524 405182 Valgrind fails to build with Clang
525 405205 filter_libc: remove the line holding the futex syscall error entirely
526 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
527 the upper and lower 32-bits of the 64-bit result
528 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
529 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
530 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
532 405403 s390x disassembler cannot be used on x86
533 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
534 405458 MIPS mkFormVEC arguments swapped?
535 405716 drd: Fix an integer overflow in the stack margin calculation
536 405722 Support arm64 core dump
537 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
538 of the 64-bit destination field.
539 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
540 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
541 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
543 406256 PPC64, vector floating point instructions don't handle subnormal
544 according to VSCR[NJ] bit setting.
545 406352 cachegrind/callgrind fails ann tests because of missing a.c
546 406354 dhat is broken on x86 (32bit)
547 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
548 406357 gdbserver_tests fails because of gdb output change
549 406360 memcheck/tests/libstdc++.supp needs more supression variants
550 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
551 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
552 407340 PPC64, does not support the vlogefp, vexptefp instructions.
553 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
554 n-i-bz Fix callgrind_annotate non deterministic order for equal total
555 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
556 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
557 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
559 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
560 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
561 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
565 Release 3.14.0 (9 October 2018)
566 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
568 3.14.0 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, AMD64/macOS 10.13.
577 * ==================== CORE CHANGES ===================
579 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
580 debug info for unloaded code. This allows saved stack traces (e.g. for
581 memory leaks) to include file/line info for code that has been dlclose'd (or
582 similar). See the user manual for more information and known limitations.
584 * Ability to specify suppressions based on source file name and line number.
586 * Majorly overhauled register allocator. No end-user changes, but the JIT
587 generates code a bit more quickly now.
589 * ================== PLATFORM CHANGES =================
591 * Preliminary support for macOS 10.13 has been added.
593 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
595 * mips: support for MIPS SIMD architecture (MSA) has been added.
597 * mips: support for MIPS N32 ABI has been added.
599 * s390: partial support for vector instructions (integer and string) has been
602 * ==================== TOOL CHANGES ====================
604 * Helgrind: Addition of a flag
605 --delta-stacktrace=no|yes [yes on linux amd64/x86]
606 which specifies how full history stack traces should be computed.
607 Setting this to =yes can speed up Helgrind by 25% when using
608 --history-level=full.
610 * Memcheck: reduced false positive rate for optimised code created by Clang 6
611 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
612 blocks more carefully to determine where it can avoid expensive definedness
613 checks without loss of precision. This is controlled by the flag
614 --expensive-definedness-checks=no|auto|yes [auto].
616 * ==================== OTHER CHANGES ====================
618 * Valgrind is now buildable with link-time optimisation (LTO). A new
619 configure option --enable-lto=yes allows building Valgrind with LTO. If the
620 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
621 Note that if you are doing Valgrind development, --enable-lto=yes massively
622 slows down the build process.
624 * ==================== FIXED BUGS ====================
626 The following bugs have been fixed or resolved. Note that "n-i-bz"
627 stands for "not in bugzilla" -- that is, a bug that was reported to us
628 but never got a bugzilla entry. We encourage you to file bugs in
629 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
630 than mailing the developers (or mailing lists) directly -- bugs that
631 are not entered into bugzilla tend to get forgotten about or ignored.
633 To see details of a given bug, visit
634 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
635 where XXXXXX is the bug number as listed below.
637 79362 Debug info is lost for .so files when they are dlclose'd
638 208052 strlcpy error when n = 0
639 255603 exp-sgcheck Assertion '!already_present' failed
640 338252 building valgrind with -flto (link time optimisation) fails
641 345763 MIPS N32 ABI support
642 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
643 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
644 372347 Replacement problem of the additional c++14/c++17 new/delete operators
645 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
646 376257 helgrind history full speed up using a cached stack
647 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
649 379748 Fix missing pselect syscall (OS X 10.11)
650 379754 Fix missing syscall ulock_wait (OS X 10.12)
651 380397 s390x: __GI_strcspn() replacemenet needed
652 381162 possible array overrun in VEX register allocator
653 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
654 381274 powerpc too chatty even with --sigill-diagnostics=no
655 381289 epoll_pwait can have a NULL sigmask
656 381553 VEX register allocator v3
657 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
658 381769 Use ucontext_t instead of struct ucontext
659 381805 arm32 needs ld.so index hardwire for new glibc security fixes
660 382256 gz compiler flag test doesn't work for gold
661 382407 vg_perf needs "--terse" command line option
662 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
663 382563 MIPS MSA ASE support
664 382998 xml-socket doesn't work
665 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
666 383723 Fix missing kevent_qos syscall (macOS 10.11)
667 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
668 384096 Mention AddrCheck at Memcheck's command line option [..]
669 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
670 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
671 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
672 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
673 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
674 384337 performance improvements to VEX register allocator v2 and v3
675 384526 reduce number of spill insns generated by VEX register allocator v3
676 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
677 384631 Sanitise client args as printed with -v
678 384633 Add a simple progress-reporting facility
679 384987 VEX regalloc: allocate caller-save registers for short lived vregs
680 385055 PPC VEX temporary storage exhausted
681 385182 PPC64 is missing support for the DSCR
682 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
683 385207 PPC64, generate_store_FPRF() generates too many Iops
684 385208 PPC64, xxperm instruction exhausts temporary memory
685 385210 PPC64, vpermr instruction could exhaust temporary memory
686 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
687 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
688 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
689 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
690 385408 s390x: z13 vector "support" instructions not implemented
691 385409 s390x: z13 vector integer instructions not implemented
692 385410 s390x: z13 vector string instructions not implemented
693 385412 s390x: new non-vector z13 instructions not implemented
694 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
695 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
696 385939 Optionally exit on the first error
697 386318 valgrind.org/info/tools.html is missing SGCheck
698 386425 running valgrind + wine on armv7l gives illegal opcode
699 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
700 387410 MIPSr6 support
701 387664 Memcheck: make expensive-definedness-checks be the default
702 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
703 387766 asm shifts cause false positive "Conditional jump or move depends
704 on uninitialised value"
705 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
706 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
707 388786 Support bpf syscall in amd64 Linux
708 388862 Add replacements for wmemchr and wcsnlen on Linux
709 389065 valgrind meets gcc flag -Wlogical-op
710 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
711 390471 suppression by specification of source-file line number
712 390723 make xtree dump files world wide readable, similar to log files
713 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
714 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
715 392118 unhandled amd64-linux syscall: 332 (statx)
716 392449 callgrind not clearing the number of calls properly
717 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
718 lxssp, stxssp and stxvl instructions.
719 393023 callgrind_control risks using the wrong vgdb
720 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
721 393099 posix_memalign() invalid write if alignment == 0
722 393146 failing assert "is_DebugInfo_active(di)"
723 395709 PPC64 is missing support for the xvnegsp instruction
724 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
726 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
727 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
728 396839 s390x: Trap instructions not implemented
729 396887 arch_prctl should return EINVAL on unknown option
730 == 397286 crash before launching binary (Unsupported arch_prctl option)
731 == 397393 valgrind: the 'impossible' happened: (Archlinux)
732 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
733 396906 compile tests failure on mips32-linux: broken inline asm in tests on
735 397012 glibc ld.so uses arch_prctl on i386
736 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
737 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
738 397424 glibc 2.27 and gdb_server tests
739 398028 Assertion `cfsi_fits` failing in simple C program
740 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
742 n-i-bz Fix missing workq_ops operations (macOS)
743 n-i-bz fix bug in strspn replacement
744 n-i-bz Add support for the Linux BLKFLSBUF ioctl
745 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
746 n-i-bz Fix possible stack trashing by semctl syscall wrapping
747 n-i-bz Add support for the Linux membarrier() system call
748 n-i-bz x86 front end: recognise and handle UD2 correctly
749 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
750 correctly aligned before entering the handler.
752 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
753 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
754 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
758 Release 3.13.0 (15 June 2017)
759 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
761 3.13.0 is a feature release with many improvements and the usual collection of
764 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
765 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
766 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
767 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
769 * ==================== CORE CHANGES ===================
771 * The translation cache size has been increased to keep up with the demands of
772 large applications. The maximum number of sectors has increased from 24 to
773 48. The default number of sectors has increased from 16 to 32 on all
774 targets except Android, where the increase is from 6 to 12.
776 * The amount of memory that Valgrind can use has been increased from 64GB to
777 128GB. In particular this means your application can allocate up to about
778 60GB when running on Memcheck.
780 * Valgrind's default load address has been changed from 0x3800'0000 to
781 0x5800'0000, so as to make it possible to load larger executables. This
782 should make it possible to load executables of size at least 1200MB.
784 * A massive spaceleak caused by reading compressed debuginfo files has been
785 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
788 * The C++ demangler has been updated.
790 * Support for demangling Rust symbols has been added.
792 * A new representation of stack traces, the "XTree", has been added. An XTree
793 is a tree of stacktraces with data associated with the stacktraces. This is
794 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
795 consumption of your program. Reporting is controlled by the new options
796 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
798 A report can also be produced on demand using the gdbserver monitor command
799 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
800 format' and 'massif format. The existing visualisers for these formats (e.g.
801 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
802 analyse these reports.
804 Memcheck can also produce XTree leak reports using the Callgrind file
805 format. For more details, see the user manual.
807 * ================== PLATFORM CHANGES =================
809 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
811 * amd64: fixes for JIT failure problems on long AVX2 code blocks
813 * amd64 and x86: support for CET prefixes has been added
815 * arm32: a few missing ARMv8 instructions have been implemented
817 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
818 Store-Conditional instructions has been added. This is to deal with
819 processor implementations that implement the LL/SC specifications strictly
820 and as a result cause Valgrind to hang in certain situations. The
821 alternative implementation is automatically enabled at startup, as required.
822 You can use the option --sim-hints=fallback-llsc to force-enable it if you
825 * Support for OSX 10.12 has been improved.
827 * On Linux, clone handling has been improved to honour CLONE_VFORK that
828 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
829 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
830 CLONE_VM exact semantics will (still) not work.
832 * The TileGX/Linux port has been removed because it appears to be both unused
835 * ==================== TOOL CHANGES ====================
839 - Memcheck should give fewer false positives when running optimised
840 Clang/LLVM generated code.
842 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
844 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
845 to produce the end of execution leak report in a xtree callgrind format
848 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
849 the leak report in an xtree file.
853 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
855 - For some workloads (typically, for big applications), Massif memory
856 consumption and CPU consumption has decreased significantly.
860 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
862 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
863 for Ada gnat compiled applications.
865 * ==================== OTHER CHANGES ====================
867 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
868 append the inner guest stacktrace to the inner host stacktrace. This helps
869 to investigate the errors reported by the outer, when they are caused by the
870 inner guest program (such as an inner regtest). See README_DEVELOPERS for
873 * To allow fast detection of callgrind files by desktop environments and file
874 managers, the format was extended to have an optional first line that
875 uniquely identifies the format ("# callgrind format"). Callgrind creates
876 this line now, as does the new xtree functionality.
878 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
879 have a new %n format letter that is replaced by a sequence number.
881 * "--version -v" now shows the SVN revision numbers from which Valgrind was
884 * ==================== FIXED BUGS ====================
886 The following bugs have been fixed or resolved. Note that "n-i-bz"
887 stands for "not in bugzilla" -- that is, a bug that was reported to us
888 but never got a bugzilla entry. We encourage you to file bugs in
889 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
890 than mailing the developers (or mailing lists) directly -- bugs that
891 are not entered into bugzilla tend to get forgotten about or ignored.
893 To see details of a given bug, visit
894 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
895 where XXXXXX is the bug number as listed below.
897 162848 --log-file output isn't split when a program forks
898 340777 Illegal instruction on mips (ar71xx)
899 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
900 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
901 to a different stack.
902 344139 x86 stack-seg overrides, needed by the Wine people
903 344524 store conditional of guest applications always fail - observed on
905 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
906 352395 Please provide SVN revision info in --version -v
907 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
908 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
909 INVALID_POSIX_THREADID' failed
910 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
911 barrier implementation
912 358697 valgrind.h: Some code remains even when defining NVALGRIND
913 359202 Add musl libc configure/compile
914 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
915 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
916 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
917 362223 assertion failed when .valgrindrc is a directory instead of a file
918 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
919 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
920 368507 can't malloc chunks larger than about 34GB
921 368529 Android arm target link error, missing atexit and pthread_atfork
922 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
923 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
924 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
925 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
926 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
927 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
928 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
929 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
930 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
931 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
932 370028 Reduce the number of compiler warnings on MIPS platforms
933 370635 arm64 missing syscall getcpu
934 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
935 371227 Clean AArch64 syscall table
936 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
937 371471 Valgrind complains about non legit memory leaks on placement new (C++)
938 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
939 371503 disInstr(arm64): unhandled instruction 0xF89F0000
940 371869 support '%' in symbol Z-encoding
941 371916 execution tree xtree concept
942 372120 c++ demangler demangles symbols which are not c++
943 372185 Support of valgrind on ARMv8 with 32 bit executable
944 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
945 372195 Power PC, xxsel instruction is not always recognized.
946 372504 Hanging on exit_group
947 372600 process loops forever when fatal signals are arriving quickly
948 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
949 373046 Stacks registered by core are never deregistered
950 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
951 373086 Implement additional Xen hypercalls
952 373192 Calling posix_spawn in glibc 2.24 completely broken
953 373488 Support for fanotify API on ARM64 architecture
954 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
955 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
956 373938 const IRExpr arguments for matchIRExpr()
957 374719 some spelling fixes
958 374963 increase valgrind's load address to prevent mmap failure
959 375514 valgrind_get_tls_addr() does not work in case of static TLS
960 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
961 for ML_(find_rx_mapping)()
962 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
963 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
964 == 377159 "vex: the `impossible' happened" still present
965 == 375150 Assertion 'tres.status == VexTransOK' failed
966 == 378068 valgrind crashes on AVX2 function in FFmpeg
967 376142 Segfaults on MIPS Cavium Octeon boards
968 376279 disInstr(arm64): unhandled instruction 0xD50320FF
969 376455 Solaris: unhandled syscall lgrpsys(180)
970 376518 Solaris: unhandled fast trap getlgrp(6)
971 376611 ppc64 and arm64 don't know about prlimit64 syscall
972 376729 PPC64, remove R2 from the clobber list
974 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
975 to be wrongly marked as addressable
976 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
977 PIE enabled by default
978 377376 memcheck/tests/linux/getregset fails with glibc2.24
979 377427 PPC64, lxv instruction failing on odd destination register
980 377478 PPC64: ISA 3.0 setup fixes
981 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
982 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
983 and 2 args for FUTEX_TRYLOCK_PI
984 377717 Fix massive space leak when reading compressed debuginfo sections
985 377891 Update Xen 4.6 domctl wrappers
986 377930 fcntl syscall wrapper is missing flock structure check
987 378524 libvexmultiarch_test regression on s390x and ppc64
988 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
989 378673 Update libiberty demangler
990 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
991 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
992 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
993 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
994 (task_register_dyld_image_infos)
995 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
996 (task_register_dyld_shared_cache_image_info)
997 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
998 379473 MIPS: add support for rdhwr cycle counter register
999 379504 remove TileGX/Linux port
1000 379525 Support more x86 nop opcodes
1001 379838 disAMode(x86): not an addr!
1002 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1003 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1004 379895 clock_gettime does not execute POST syscall wrapper
1005 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1006 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1007 380200 xtree generated callgrind files refer to files without directory name
1008 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1009 380397 s390x: __GI_strcspn() replacement needed
1010 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1012 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1013 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1014 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1018 Release 3.12.0 (20 October 2016)
1019 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1021 3.12.0 is a feature release with many improvements and the usual
1022 collection of bug fixes.
1024 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1025 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1026 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1027 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1028 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1029 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1031 * ================== PLATFORM CHANGES =================
1033 * POWER: Support for ISA 3.0 has been added
1035 * mips: support for O32 FPXX ABI has been added.
1036 * mips: improved recognition of different processors
1037 * mips: determination of page size now done at run time
1039 * amd64: Partial support for AMD FMA4 instructions.
1041 * arm, arm64: Support for v8 crypto and CRC instructions.
1043 * Improvements and robustification of the Solaris port.
1045 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1047 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1048 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1049 where possible. Valgrind's support for 32-bit x86 has stagnated in
1050 recent years and has fallen far behind that for 64-bit x86
1051 instructions. By contrast 64-bit x86 is well supported, up to and
1054 * ==================== TOOL CHANGES ====================
1058 - Added meta mempool support for describing a custom allocator which:
1059 - Auto-frees all chunks assuming that destroying a pool destroys all
1061 - Uses itself to allocate other memory blocks
1063 - New flag --ignore-range-below-sp to ignore memory accesses below
1064 the stack pointer, if you really have to. The related flag
1065 --workaround-gcc296-bugs=yes is now deprecated. Use
1066 --ignore-range-below-sp=1024-1 as a replacement.
1070 - Improved thread startup time significantly on non-Linux platforms.
1074 - Added collection of the metric "tot-blocks-allocd"
1076 * ==================== OTHER CHANGES ====================
1078 * Replacement/wrapping of malloc/new related functions is now done not just
1079 for system libraries by default, but for any globally defined malloc/new
1080 related function (both in shared libraries and statically linked alternative
1081 malloc implementations). The dynamic (runtime) linker is excluded, though.
1082 To only intercept malloc/new related functions in
1083 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1084 "nouserintercepts" can be any non-existing library name).
1085 This new functionality is not implemented for MacOS X.
1087 * The maximum number of callers in a suppression entry is now equal to
1088 the maximum size for --num-callers (500).
1089 Note that --gen-suppressions=yes|all similarly generates suppressions
1090 containing up to --num-callers frames.
1092 * New and modified GDB server monitor features:
1094 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1095 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1098 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1099 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1102 * Valgrind is able to read compressed debuginfo sections in two formats:
1103 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1104 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1106 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1107 for the most common use case (x86_64-linux, Memcheck) has been
1110 * Improved performance for programs that do a lot of discarding of
1111 instruction address ranges of 8KB or less.
1113 * The C++ symbol demangler has been updated.
1115 * More robustness against invalid syscall parameters on Linux.
1117 * ==================== FIXED BUGS ====================
1119 The following bugs have been fixed or resolved. Note that "n-i-bz"
1120 stands for "not in bugzilla" -- that is, a bug that was reported to us
1121 but never got a bugzilla entry. We encourage you to file bugs in
1122 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1123 than mailing the developers (or mailing lists) directly -- bugs that
1124 are not entered into bugzilla tend to get forgotten about or ignored.
1126 To see details of a given bug, visit
1127 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1128 where XXXXXX is the bug number as listed below.
1130 191069 Exiting due to signal not reported in XML output
1131 199468 Suppressions: stack size limited to 25
1132 while --num-callers allows more frames
1133 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1134 278744 cvtps2pd with redundant RexW
1135 303877 valgrind doesn't support compressed debuginfo sections.
1136 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1137 348345 Assertion fails for negative lineno
1138 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1139 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1140 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1141 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1142 352197 mips: mmap2() not wrapped correctly for page size > 4096
1143 353083 arm64 doesn't implement various xattr system calls
1144 353084 arm64 doesn't support sigpending system call
1145 353137 www: update info for Supported Platforms
1146 353138 www: update "The Valgrind Developers" page
1147 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1150 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1151 353398 WARNING: unhandled amd64-solaris syscall: 207
1152 353660 XML in auxwhat tag not escaping reserved symbols properly
1153 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1154 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1155 353802 ELF debug info reader confused with multiple .rodata sections
1156 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1157 353917 unhandled amd64-solaris syscall fchdir(120)
1158 353920 unhandled amd64-solaris syscall: 170
1159 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1160 354392 unhandled amd64-solaris syscall: 171
1161 354797 Vbit test does not include Iops for Power 8 instruction support
1162 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1166 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1167 355188 valgrind should intercept all malloc related global functions
1168 355454 do not intercept malloc related symbols from the runtime linker
1169 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1170 356044 Dwarf line info reader misinterprets is_stmt register
1171 356112 mips: replace addi with addiu
1172 356393 valgrind (vex) crashes because isZeroU happened
1175 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1176 356678 arm64-linux: unhandled syscall 232 (mincore)
1177 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1178 356823 Unsupported ARM instruction: stlex
1179 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1180 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1181 357673 crash if I try to run valgrind with a binary link with libcurl
1182 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1183 357871 pthread_spin_destroy not properly wrapped
1184 357887 Calls to VG_(fclose) do not close the file descriptor
1185 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1186 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1187 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1188 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1189 359181 Buffer Overflow during Demangling
1190 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1191 359289 s390x: popcnt (B9E1) not implemented
1192 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1193 359503 Add missing syscalls for aarch64 (arm64)
1194 359645 "You need libc6-dbg" help message could be more helpful
1195 359703 s390: wire up separate socketcalls system calls
1196 359724 getsockname might crash - deref_UInt should call safe_to_deref
1197 359733 amd64 implement ld.so strchr/index override like x86
1198 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1199 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1201 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1202 359871 Incorrect mask handling in ppoll
1203 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1204 360008 Contents of Power vr registers contents is not printed correctly when
1205 the --vgdb-shadow-registers=yes option is used
1206 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1207 non-zero shadow bits
1208 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1209 360425 arm64 unsupported instruction ldpsw
1211 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1212 360571 Error about the Android Runtime reading below the stack pointer on ARM
1213 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1214 360749 kludge for multiple .rodata sections on Solaris no longer needed
1215 360752 raise the number of reserved fds in m_main.c from 10 to 12
1216 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1217 361226 s390x: risbgn (EC59) not implemented
1218 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1219 361354 ppc64[le]: wire up separate socketcalls system calls
1220 361615 Inconsistent termination for multithreaded process terminated by signal
1221 361926 Unhandled Solaris syscall: sysfs(84)
1222 362009 V dumps core on unimplemented functionality before threads are created
1223 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1224 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1225 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1226 362953 Request for an update to the Valgrind Developers page
1227 363680 add renameat2() support
1228 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1229 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1230 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1231 364058 clarify in manual limitations of array overruns detections
1232 364413 pselect sycallwrapper mishandles NULL sigmask
1233 364728 Power PC, missing support for several HW registers in
1234 get_otrack_shadow_offset_wrk()
1235 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1236 365273 Invalid write to stack location reported after signal handler runs
1237 365912 ppc64BE segfault during jm-insns test (RELRO)
1238 366079 FPXX Support for MIPS32 Valgrind
1239 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1240 366344 Multiple unhandled instruction for Aarch64
1241 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1242 367995 Integration of memcheck with custom memory allocator
1243 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1244 368412 False positive result for altivec capability check
1245 368416 Add tc06_two_races_xml.exp output for ppc64
1246 368419 Perf Events ioctls not implemented
1247 368461 mmapunmap test fails on ppc64
1248 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1249 369000 AMD64 fma4 instructions unsupported.
1250 369169 ppc64 fails jm_int_isa_2_07 test
1251 369175 jm_vec_isa_2_07 test crashes on ppc64
1252 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1253 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1254 369359 msghdr_foreachfield can crash when handling bad iovec
1255 369360 Bad sigprocmask old or new sets can crash valgrind
1256 369361 vmsplice syscall wrapper crashes on bad iovec
1257 369362 Bad sigaction arguments crash valgrind
1258 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1259 369402 Bad set/get_thread_area pointer crashes valgrind
1260 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1261 369446 valgrind crashes on unknown fcntl command
1262 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1263 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1264 370265 ISA 3.0 HW cap stuff needs updating
1265 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1266 372195 Power PC, xxsel instruction is not always recognized
1268 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1269 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1270 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1271 n-i-bz Fix ppoll_alarm exclusion on OS X
1272 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1273 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1274 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1275 n-i-bz amd64: memcheck false positive with shr %edx
1276 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1277 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1278 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1279 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1281 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1282 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1283 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1287 Release 3.11.0 (22 September 2015)
1288 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1290 3.11.0 is a feature release with many improvements and the usual
1291 collection of bug fixes.
1293 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1294 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1295 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1296 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1297 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1298 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1300 * ================== PLATFORM CHANGES =================
1302 * Support for Solaris/x86 and Solaris/amd64 has been added.
1304 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1306 * Preliminary support for the Tilera TileGX architecture has been added.
1308 * s390x: It is now required for the host to have the "long displacement"
1309 facility. The oldest supported machine model is z990.
1311 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1312 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1314 * The JIT's register allocator is significantly faster, making the JIT
1315 as a whole somewhat faster, so JIT-intensive activities, for example
1316 program startup, are modestly faster, around 5%.
1318 * There have been changes to the default settings of several command
1319 line flags, as detailed below.
1321 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1322 capable hosts, the simulated CPUID will now indicate AVX2 support.
1324 * ==================== TOOL CHANGES ====================
1328 - The default value for --leak-check-heuristics has been changed from
1329 "none" to "all". This helps to reduce the number of possibly
1330 lost blocks, in particular for C++ applications.
1332 - The default value for --keep-stacktraces has been changed from
1333 "malloc-then-free" to "malloc-and-free". This has a small cost in
1334 memory (one word per malloc-ed block) but allows Memcheck to show the
1335 3 stacktraces of a dangling reference: where the block was allocated,
1336 where it was freed, and where it is acccessed after being freed.
1338 - The default value for --partial-loads-ok has been changed from "no" to
1339 "yes", so as to avoid false positive errors resulting from some kinds
1340 of vectorised loops.
1342 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1343 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1344 than get_vbits when you need to associate byte data value with
1345 their corresponding validity bits.
1347 - The 'block_list' monitor command has been enhanced:
1348 o it can print a range of loss records
1349 o it now accepts an optional argument 'limited <max_blocks>'
1350 to control the number of blocks printed.
1351 o if a block has been found using a heuristic, then
1352 'block_list' now shows the heuristic after the block size.
1353 o the loss records/blocks to print can be limited to the blocks
1354 found via specified heuristics.
1356 - The C helper functions used to instrument loads on
1357 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1358 replaced by handwritten assembly sequences. This gives speedups
1359 in the region of 0% to 7% for those targets only.
1361 - A new command line option, --expensive-definedness-checks=yes|no,
1362 has been added. This is useful for avoiding occasional invalid
1363 uninitialised-value errors in optimised code. Watch out for
1364 runtime degradation, as this can be up to 25%. As always, though,
1365 the slowdown is highly application specific. The default setting
1370 - A new monitor command 'all_snapshots <filename>' dumps all
1371 snapshots taken so far.
1375 - Significant memory reduction and moderate speedups for
1376 --history-level=full for applications accessing a lot of memory
1377 with many different stacktraces.
1379 - The default value for --conflict-cache-size=N has been doubled to
1380 2000000. Users that were not using the default value should
1381 preferably also double the value they give.
1383 The default was changed due to the changes in the "full history"
1384 implementation. Doubling the value gives on average a slightly more
1385 complete history and uses similar memory (or significantly less memory
1386 in the worst case) than the previous implementation.
1388 - The Helgrind monitor command 'info locks' now accepts an optional
1389 argument 'lock_addr', which shows information about the lock at the
1392 - When using --history-level=full, the new Helgrind monitor command
1393 'accesshistory <addr> [<len>]' will show the recorded accesses for
1394 <len> (or 1) bytes at <addr>.
1396 * ==================== OTHER CHANGES ====================
1398 * The default value for the --smc-check option has been changed from
1399 "stack" to "all-non-file" on targets that provide automatic D-I
1400 cache coherence (x86, amd64 and s390x). The result is to provide,
1401 by default, transparent support for JIT generated and self-modifying
1402 code on all targets.
1404 * Mac OS X only: the default value for the --dsymutil option has been
1405 changed from "no" to "yes", since any serious usage on Mac OS X
1406 always required it to be "yes".
1408 * The command line options --db-attach and --db-command have been removed.
1409 They were deprecated in 3.10.0.
1411 * When a process dies due to a signal, Valgrind now shows the signal
1412 and the stacktrace at default verbosity (i.e. verbosity 1).
1414 * The address description logic used by Memcheck and Helgrind now
1415 describes addresses in anonymous segments, file mmap-ed segments,
1416 shared memory segments and the brk data segment.
1418 * The new option --error-markers=<begin>,<end> can be used to mark the
1419 begin/end of errors in textual output mode, to facilitate
1420 searching/extracting errors in output files that mix valgrind errors
1421 with program output.
1423 * The new option --max-threads=<number> can be used to change the number
1424 of threads valgrind can handle. The default is 500 threads which
1425 should be more than enough for most applications.
1427 * The new option --valgrind-stacksize=<number> can be used to change the
1428 size of the private thread stacks used by Valgrind. This is useful
1429 for reducing memory use or increasing the stack size if Valgrind
1430 segfaults due to stack overflow.
1432 * The new option --avg-transtab-entry-size=<number> can be used to specify
1433 the expected instrumented block size, either to reduce memory use or
1434 to avoid excessive retranslation.
1436 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1438 * New and modified GDB server monitor features:
1440 - When a signal is reported in GDB, you can now use the GDB convenience
1441 variable $_siginfo to examine detailed signal information.
1443 - Valgrind's gdbserver now allows the user to change the signal
1444 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1445 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1446 continue without passing the signal to the process.
1448 - With GDB >= 7.10, the command 'target remote'
1449 will automatically load the executable file of the process running
1450 under Valgrind. This means you do not need to specify the executable
1451 file yourself, GDB will discover it itself. See GDB documentation about
1452 'qXfer:exec-file:read' packet for more info.
1454 * ==================== FIXED BUGS ====================
1456 The following bugs have been fixed or resolved. Note that "n-i-bz"
1457 stands for "not in bugzilla" -- that is, a bug that was reported to us
1458 but never got a bugzilla entry. We encourage you to file bugs in
1459 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1460 than mailing the developers (or mailing lists) directly -- bugs that
1461 are not entered into bugzilla tend to get forgotten about or ignored.
1463 To see details of a given bug, visit
1464 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1465 where XXXXXX is the bug number as listed below.
1467 116002 VG_(printf): Problems with justification of strings and integers
1468 155125 avoid cutting away file:lineno after long function name
1469 197259 Unsupported arch_prtctl PR_SET_GS option
1470 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
1471 201216 Fix Valgrind does not support pthread_sigmask() on OS X
1472 201435 Fix Darwin: -v does not show kernel version
1473 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
1474 211256 Fixed an outdated comment regarding the default platform.
1475 211529 Incomplete call stacks for code compiled by newer versions of MSVC
1476 211926 Avoid compilation warnings in valgrind.h with -pedantic
1477 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
1479 226609 Crediting upstream authors in man page
1480 231257 Valgrind omits path when executing script from shebang line
1481 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
1482 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
1483 269360 s390x: Fix addressing mode selection for compare-and-swap
1484 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
1486 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
1487 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
1488 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
1489 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
1490 330147 libmpiwrap PMPI_Get_count returns undefined value
1491 333051 mmap of huge pages fails due to incorrect alignment
1493 334802 valgrind does not always explain why a given option is bad
1494 335618 mov.w rN, pc/sp (ARM32)
1495 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
1500 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
1501 338602 AVX2 bit in CPUID missing
1502 338606 Strange message for scripts with invalid interpreter
1503 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
1504 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
1505 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
1507 339156 gdbsrv not called for fatal signal
1508 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
1509 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
1510 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
1511 339442 Fix testsuite build failure on OS X 10.9
1512 339542 Enable compilation with Intel's ICC compiler
1513 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
1514 339688 Mac-specific ASM does not support .version directive (cpuid,
1515 tronical and pushfpopf tests)
1516 339745 Valgrind crash when check Marmalade app (partial fix)
1517 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
1518 339778 Linux/TileGx platform support for Valgrind
1519 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
1520 339789 Fix none/tests/execve test on Mac OS X 10.9
1521 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
1522 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
1523 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
1524 340392 Allow user to select more accurate definedness checking in memcheck
1525 to avoid invalid complaints on optimised code
1526 340430 Fix some grammatical weirdness in the manual.
1527 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
1528 341419 Signal handler ucontext_t not filled out correctly on OS X
1529 341539 VG_(describe_addr) should not describe address as belonging to client
1530 segment if it is past the heap end
1531 341613 Enable building of manythreads and thread-exits tests on Mac OS X
1532 341615 Fix none/tests/darwin/access_extended test on Mac OS X
1533 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
1534 341789 aarch64: shmat fails with valgrind on ARMv8
1535 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
1536 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
1537 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
1538 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
1539 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
1540 342221 socket connect false positive uninit memory for unknown af family
1541 342353 Allow dumping full massif output while valgrind is still running
1542 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
1546 342603 Add I2C_SMBUS ioctl support
1547 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
1548 342683 Mark memory past the initial brk limit as unaddressable
1549 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
1550 342795 Internal glibc __GI_mempcpy call should be intercepted
1551 342841 s390x: Support instructions fiebr(a) and fidbr(a)
1552 343012 Unhandled syscall 319 (memfd_create)
1553 343069 Patch updating v4l2 API support
1554 343173 helgrind crash during stack unwind
1555 343219 fix GET_STARTREGS for arm
1556 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
1557 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
1558 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
1559 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
1560 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
1561 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
1562 343597 ppc64le: incorrect use of offseof macro
1563 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
1564 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
1565 343732 Unhandled syscall 144 (setgid) on aarch64
1566 343733 Unhandled syscall 187 (msgctl and related) on aarch64
1567 343802 s390x: False positive "conditional jump or move depends on [..]
1568 343902 --vgdb=yes doesn't break when --xml=yes is used
1569 343967 Don't warn about setuid/setgid/setcap executable for directories
1570 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
1571 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
1572 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
1573 344054 www - update info for Solaris/illumos
1574 344416 'make regtest' does not work cleanly on OS X
1575 344235 Remove duplicate include of pub_core_aspacemgr.h
1576 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
1577 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
1578 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
1579 344314 callgrind_annotate ... warnings about commands containing newlines
1580 344318 socketcall should wrap recvmmsg and sendmmsg
1581 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
1582 344416 Fix 'make regtest' does not work cleanly on OS X
1583 344499 Fix compilation for Linux kernel >= 4.0.0
1584 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
1585 unix:349 (__pthread_fchdir)
1586 344559 Garbage collection of unused segment names in address space manager
1587 344560 Fix stack traces missing penultimate frame on OS X
1588 344621 Fix memcheck/tests/err_disable4 test on OS X
1589 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
1590 344702 Fix missing libobjc suppressions on OS X 10.10
1592 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
1593 344939 Fix memcheck/tests/xml1 on OS X 10.10
1594 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
1595 345079 Fix build problems in VEX/useful/test_main.c
1596 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
1597 345177 arm64: prfm (reg) not implemented
1598 345215 Performance improvements for the register allocator
1599 345248 add support for Solaris OS in valgrind
1600 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
1601 345394 Fix memcheck/tests/strchr on OS X
1602 345637 Fix memcheck/tests/sendmsg on OS X
1603 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
1604 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
1605 345887 Fix an assertion in the address space manager
1606 345928 amd64: callstack only contains current function for small stacks
1607 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1608 345987 MIPS64: Implement cavium LHX instruction
1609 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1610 346185 Fix typo saving altivec register v24
1611 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1612 and LibVEX_GuestPPC64_get_CR()
1613 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1614 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1615 346307 fuse filesystem syscall deadlocks
1616 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1617 346411 MIPS: SysRes::_valEx handling is incorrect
1618 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1619 346474 PPC64 Power 8, spr TEXASRU register not supported
1620 346487 Compiler generates "note" about a future ABI change for PPC64
1621 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1622 and causing spurious "invalid read of size 8" warnings
1623 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1624 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1625 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1626 347322 Power PC regression test cleanup
1627 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1629 347389 unhandled syscall: 373 (Linux ARM syncfs)
1630 347686 Patch set to cleanup PPC64 regtests
1631 347978 Remove bash dependencies where not needed
1632 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1633 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1635 348102 Patch updating v4l2 API support
1636 348247 amd64 front end: jno jumps wrongly when overflow is not set
1637 348269 Improve mmap MAP_HUGETLB support.
1638 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1639 348345 Assertion fails for negative lineno
1640 348377 Unsupported ARM instruction: yield
1641 348565 Fix detection of command line option availability for clang
1642 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1643 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1644 348748 Fix redundant condition
1645 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1646 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1647 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1648 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1649 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1650 349626 Implemented additional Xen hypercalls
1651 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1652 349790 Clean up of the hardware capability checking utilities.
1653 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1654 349874 Fix typos in source code
1655 349879 memcheck: add handwritten assembly for helperc_LOADV*
1656 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1657 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1658 350202 Add limited param to 'monitor block_list'
1659 350290 s390x: Support instructions fixbr(a)
1660 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1661 350809 Fix none/tests/async-sigs for Solaris
1662 350811 Remove reference to --db-attach which has been removed.
1663 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1664 350854 hard-to-understand code in VG_(load_ELF)()
1665 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1666 351386 Solaris: Cannot run ld.so.1 under Valgrind
1667 351474 Fix VG_(iseqsigset) as obvious
1668 351531 Typo in /include/vki/vki-xen-physdev.h header guard
1669 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
1670 351858 ldsoexec support on Solaris
1671 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
1672 352130 helgrind reports false races for printfs using mempcpy on FILE* state
1673 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
1674 352320 arm64 crash on none/tests/nestedfs
1675 352765 Vbit test fails on Power 6
1676 352768 The mbar instruction is missing from the Power PC support
1677 352769 Power PC program priority register (PPR) is not supported
1678 n-i-bz Provide implementations of certain compiler builtins to support
1679 compilers that may not provide those
1680 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
1681 n-i-bz Fix compilation on distros with glibc < 2.5
1682 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
1683 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
1684 n-i-bz mremap did not work properly on shared memory
1685 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
1686 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
1688 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
1689 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
1690 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
1694 Release 3.10.1 (25 November 2014)
1695 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1696 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
1697 and backports fixes for all reported missing AArch64 ARMv8 instructions
1698 and syscalls from the trunk. If you package or deliver 3.10.0 for others
1699 to use, you might want to consider upgrading to 3.10.1 instead.
1701 The following bugs have been fixed or resolved. Note that "n-i-bz"
1702 stands for "not in bugzilla" -- that is, a bug that was reported to us
1703 but never got a bugzilla entry. We encourage you to file bugs in
1704 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1705 than mailing the developers (or mailing lists) directly -- bugs that
1706 are not entered into bugzilla tend to get forgotten about or ignored.
1708 To see details of a given bug, visit
1709 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1710 where XXXXXX is the bug number as listed below.
1712 335440 arm64: ld1 (single structure) is not implemented
1713 335713 arm64: unhanded instruction: prfm (immediate)
1714 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
1715 339182 ppc64: AvSplat ought to load destination vector register with [..]
1716 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
1717 339433 ppc64 lxvw4x instruction uses four 32-byte loads
1718 339645 Use correct tag names in sys_getdents/64 wrappers
1719 339706 Fix false positive for ioctl(TIOCSIG) on linux
1720 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
1721 339853 arm64 times syscall unknown
1722 339855 arm64 unhandled getsid/setsid syscalls
1723 339858 arm64 dmb sy not implemented
1724 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
1725 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
1726 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
1728 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
1729 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
1730 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
1731 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
1732 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
1733 340509 arm64: unhandled instruction fcvtas
1734 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
1735 340632 arm64: unhandled instruction fcvtas
1736 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
1737 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
1738 340788 warning: unhandled syscall: 318 (getrandom)
1739 340807 disInstr(arm): unhandled instruction: 0xEE989B20
1740 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
1741 340922 arm64: unhandled getgroups/setgroups syscalls
1742 350251 Fix typo in VEX utility program (test_main.c).
1743 350407 arm64: unhandled instruction ucvtf (vector, integer)
1744 350809 none/tests/async-sigs breaks when run under cron on Solaris
1745 350811 update README.solaris after r15445
1746 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
1747 350854 strange code in VG_(load_ELF)()
1748 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1749 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
1750 n-i-bz Add missing ]] to terminate CDATA.
1751 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
1752 n-i-bz Enable sys_fadvise64_64 on arm32.
1753 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
1754 n-i-bz Add test cases for all known arm64 load/store instructions.
1755 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
1756 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
1757 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
1758 n-i-bz arm64: Implement "fcvtpu w, s".
1759 n-i-bz arm64: implement ADDP and various others
1760 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
1761 n-i-bz arm64: enable FCVT{A,N}S X,S.
1763 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
1767 Release 3.10.0 (10 September 2014)
1768 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1770 3.10.0 is a feature release with many improvements and the usual
1771 collection of bug fixes.
1773 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1774 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1775 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
1776 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
1777 significantly improved relative to the 3.9.0 release.
1779 * ================== PLATFORM CHANGES =================
1781 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
1782 is mostly complete, and is usable, but some SIMD instructions are as
1785 * Support for little-endian variant of the 64-bit POWER architecture.
1787 * Support for Android on MIPS32.
1789 * Support for 64bit FPU on MIPS32 platforms.
1791 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
1793 * Configuration for and running on Android targets has changed.
1794 See README.android in the source tree for details.
1796 * ================== DEPRECATED FEATURES =================
1798 * --db-attach is now deprecated and will be removed in the next
1799 valgrind feature release. The built-in GDB server capabilities are
1800 superior and should be used instead. Learn more here:
1801 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
1803 * ==================== TOOL CHANGES ====================
1807 - Client code can now selectively disable and re-enable reporting of
1808 invalid address errors in specific ranges using the new client
1809 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
1810 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
1812 - Leak checker: there is a new leak check heuristic called
1813 "length64". This is used to detect interior pointers pointing 8
1814 bytes inside a block, on the assumption that the first 8 bytes
1815 holds the value "block size - 8". This is used by
1816 sqlite3MemMalloc, for example.
1818 - Checking of system call parameters: if a syscall parameter
1819 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
1820 several fields not initialised, an error is now reported for each
1821 field. Previously, an error was reported only for the first
1822 uninitialised field.
1824 - Mismatched alloc/free checking: a new flag
1825 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
1826 such checks if necessary.
1830 - Improvements to error messages:
1832 o Race condition error message involving heap allocated blocks also
1833 show the thread number that allocated the raced-on block.
1835 o All locks referenced by an error message are now announced.
1836 Previously, some error messages only showed the lock addresses.
1838 o The message indicating where a lock was first observed now also
1839 describes the address/location of the lock.
1841 - Helgrind now understands the Ada task termination rules and
1842 creates a happens-before relationship between a terminated task
1843 and its master. This avoids some false positives and avoids a big
1844 memory leak when a lot of Ada tasks are created and terminated.
1845 The interceptions are only activated with forthcoming releases of
1846 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
1848 - A new GDB server monitor command "info locks" giving the list of
1849 locks, their location, and their status.
1853 - callgrind_control now supports the --vgdb-prefix argument,
1854 which is needed if valgrind was started with this same argument.
1856 * ==================== OTHER CHANGES ====================
1858 * Unwinding through inlined function calls. Stack unwinding can now
1859 make use of Dwarf3 inlined-unwind information if it is available.
1860 The practical effect is that inlined calls become visible in stack
1861 traces. The suppression matching machinery has been adjusted
1862 accordingly. This is controlled by the new option
1863 --read-inline-info=yes|no. Currently this is enabled by default
1864 only on Linux and Android targets and only for the tools Memcheck,
1867 * Valgrind can now read EXIDX unwind information on 32-bit ARM
1868 targets. If an object contains both CFI and EXIDX unwind
1869 information, Valgrind will prefer the CFI over the EXIDX. This
1870 facilitates unwinding through system libraries on arm-android
1873 * Address description logic has been improved and is now common
1874 between Memcheck and Helgrind, resulting in better address
1875 descriptions for some kinds of error messages.
1877 * Error messages about dubious arguments (eg, to malloc or calloc) are
1878 output like other errors. This means that they can be suppressed
1879 and they have a stack trace.
1881 * The C++ demangler has been updated for better C++11 support.
1883 * New and modified GDB server monitor features:
1885 - Thread local variables/storage (__thread) can now be displayed.
1887 - The GDB server monitor command "v.info location <address>"
1888 displays information about an address. The information produced
1889 depends on the tool and on the options given to valgrind.
1890 Possibly, the following are described: global variables, local
1891 (stack) variables, allocated or freed blocks, ...
1893 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
1894 ask the GDB server to stop at the start of program execution, at
1895 the end of the program execution and on Valgrind internal errors.
1897 - A new monitor command "v.info stats" shows various Valgrind core
1898 and tool statistics.
1900 - A new monitor command "v.set hostvisibility" allows the GDB server
1901 to provide access to Valgrind internal host status/memory.
1903 * A new option "--aspace-minaddr=<address>" can in some situations
1904 allow the use of more memory by decreasing the address above which
1905 Valgrind maps memory. It can also be used to solve address
1906 conflicts with system libraries by increasing the default value.
1907 See user manual for details.
1909 * The amount of memory used by Valgrind to store debug info (unwind
1910 info, line number information and symbol data) has been
1911 significantly reduced, even though Valgrind now reads more
1912 information in order to support unwinding of inlined function calls.
1914 * Dwarf3 handling with --read-var-info=yes has been improved:
1916 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
1918 - Code compiled with
1919 -ffunction-sections -fdata-sections -Wl,--gc-sections
1920 no longer causes assertion failures.
1922 * Improved checking for the --sim-hints= and --kernel-variant=
1923 options. Unknown strings are now detected and reported to the user
1926 * The semantics of stack start/end boundaries in the valgrind.h
1927 VALGRIND_STACK_REGISTER client request has been clarified and
1928 documented. The convention is that start and end are respectively
1929 the lowest and highest addressable bytes of the stack.
1931 * ==================== FIXED BUGS ====================
1933 The following bugs have been fixed or resolved. Note that "n-i-bz"
1934 stands for "not in bugzilla" -- that is, a bug that was reported to us
1935 but never got a bugzilla entry. We encourage you to file bugs in
1936 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1937 than mailing the developers (or mailing lists) directly -- bugs that
1938 are not entered into bugzilla tend to get forgotten about or ignored.
1940 To see details of a given bug, visit
1941 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1942 where XXXXXX is the bug number as listed below.
1944 175819 Support for ipv6 socket reporting with --track-fds
1945 232510 make distcheck fails
1946 249435 Analyzing wine programs with callgrind triggers a crash
1947 278972 support for inlined function calls in stacktraces and suppression
1949 291310 FXSAVE instruction marks memory as undefined on amd64
1950 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
1951 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
1952 315199 vgcore file for threaded app does not show which thread crashed
1953 315952 tun/tap ioctls are not supported
1954 323178 Unhandled instruction: PLDW register (ARM)
1955 323179 Unhandled instruction: PLDW immediate (ARM)
1956 324050 Helgrind: SEGV because of unaligned stack when using movdqa
1957 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
1958 325124 [MIPSEL] Compilation error
1959 325477 Phase 4 support for IBM Power ISA 2.07
1960 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
1961 325628 Phase 5 support for IBM Power ISA 2.07
1962 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
1963 325751 Missing the two privileged Power PC Transactional Memory Instructions
1964 325816 Phase 6 support for IBM Power ISA 2.07
1965 325856 Make SGCheck fail gracefully on unsupported platforms
1966 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
1967 326436 DRD: False positive in libstdc++ std::list::push_back
1968 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
1969 326462 Refactor vgdb to isolate invoker stuff into separate module
1970 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
1971 326623 DRD: false positive conflict report in a field assignment
1972 326724 Valgrind does not compile on OSX 1.9 Mavericks
1973 326816 Intercept for __strncpy_sse2_unaligned missing?
1974 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
1975 326983 Clear direction flag after tests on amd64.
1976 327212 Do not prepend the current directory to absolute path names.
1977 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
1978 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
1979 327284 s390x: Fix translation of the risbg instruction
1980 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
1981 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
1982 327916 DW_TAG_typedef may have no name
1983 327943 s390x: add a redirection for the 'index' function
1984 328100 XABORT not implemented
1985 328205 Implement additional Xen hypercalls
1986 328454 add support Backtraces with ARM unwind tables (EXIDX)
1987 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
1988 328711 valgrind.1 manpage "memcheck options" section is badly generated
1989 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
1990 329612 Incorrect handling of AT_BASE for image execution
1991 329694 clang warns about using uninitialized variable
1992 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
1993 330228 mmap must align to VKI_SHMLBA on mips32
1994 330257 LLVM does not support `-mno-dynamic-no-pic` option
1995 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
1996 330459 --track-fds=yes doesn't track eventfds
1997 330469 Add clock_adjtime syscall support
1998 330594 Missing sysalls on PowerPC / uClibc
1999 330622 Add test to regression suite for POWER instruction: dcbzl
2000 330939 Support for AMD's syscall instruction on x86
2002 330941 Typo in PRE(poll) syscall wrapper
2003 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2004 331254 Fix expected output for memcheck/tests/dw4
2005 331255 Fix race condition in test none/tests/coolo_sigaction
2006 331257 Fix type of jump buffer in test none/tests/faultstatus
2007 331305 configure uses bash specific syntax
2008 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2009 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2010 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2011 331829 Unexpected ioctl opcode sign extension
2012 331830 ppc64: WARNING: unhandled syscall: 96/97
2013 331839 drd/tests/sem_open specifies invalid semaphore name
2014 331847 outcome of drd/tests/thread_name is nondeterministic
2015 332037 Valgrind cannot handle Thumb "add pc, reg"
2016 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2017 consistency checks enabled
2018 332263 intercepts for pthread_rwlock_timedrdlock and
2019 pthread_rwlock_timedwrlock are incorrect
2020 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2022 332276 Implement additional Xen hypercalls
2023 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2024 332765 Fix ms_print to create temporary files in a proper directory
2025 333072 drd: Add semaphore annotations
2026 333145 Tests for missaligned PC+#imm access for arm
2027 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2028 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2029 333248 WARNING: unhandled syscall: unix:443
2030 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2031 333501 cachegrind: assertion: Cache set count is not a power of two.
2034 333666 Recognize MPX instructions and bnd prefix.
2035 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2036 333817 Valgrind reports the memory areas written to by the SG_IO
2038 334049 lzcnt fails silently (x86_32)
2039 334384 Valgrind does not have support Little Endian support for
2041 334585 recvmmsg unhandled (+patch) (arm)
2042 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2043 334727 Build fails with -Werror=format-security
2044 334788 clarify doc about --log-file initial program directory
2045 334834 PPC64 Little Endian support, patch 2
2046 334836 PPC64 Little Endian support, patch 3 testcase fixes
2047 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2048 335034 Unhandled ioctl: HCIGETDEVLIST
2049 335155 vgdb, fix error print statement.
2050 335262 arm64: movi 8bit version is not supported
2051 335263 arm64: dmb instruction is not implemented
2052 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2053 335496 arm64: sbc/abc instructions are not implemented
2054 335554 arm64: unhandled instruction: abs
2055 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2056 335735 arm64: unhandled instruction: cnt
2057 335736 arm64: unhandled instruction: uaddlv
2058 335848 arm64: unhandled instruction: {s,u}cvtf
2059 335902 arm64: unhandled instruction: sli
2060 335903 arm64: unhandled instruction: umull (vector)
2061 336055 arm64: unhandled instruction: mov (element)
2062 336062 arm64: unhandled instruction: shrn{,2}
2063 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2064 336189 arm64: unhandled Instruction: mvn
2065 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2066 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2067 336772 Make moans about unknown ioctls more informative
2068 336957 Add a section about the Solaris/illumos port on the webpage
2069 337094 ifunc wrapper is broken on ppc64
2070 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2071 337528 leak check heuristic for block prefixed by length as 64bit number
2072 337740 Implement additional Xen hypercalls
2073 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2074 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2075 337871 deprecate --db-attach
2076 338023 Add support for all V4L2/media ioctls
2077 338024 inlined functions are not shown if DW_AT_ranges is used
2078 338106 Add support for 'kcmp' syscall
2079 338115 DRD: computed conflict set differs from actual after fork
2080 338160 implement display of thread local storage in gdbsrv
2081 338205 configure.ac and check for -Wno-tautological-compare
2082 338300 coredumps are missing one byte of every segment
2083 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2084 338499 --sim-hints parsing broken due to wrong order in tokens
2085 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2086 338681 Unable to unwind through clone thread created on i386-linux
2087 338698 race condition between gdbsrv and vgdb on startup
2088 338703 helgrind on arm-linux gets false positives in dynamic loader
2089 338791 alt dwz files can be relative of debug/main file
2090 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2091 338932 build V-trunk with gcc-trunk
2092 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2093 345079 Fix build problems in VEX/useful/test_main.c
2094 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2095 n-i-bz s390x: Fix memory corruption for multithreaded applications
2096 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2097 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2098 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2099 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2100 n-i-bz Add command-line option --default-suppressions
2101 n-i-bz Add support for BLKDISCARDZEROES ioctl
2102 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2103 n-i-bz Add support for sys_pivot_root and sys_unshare
2105 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2106 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2107 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2111 Release 3.9.0 (31 October 2013)
2112 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2113 3.9.0 is a feature release with many improvements and the usual
2114 collection of bug fixes.
2116 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2117 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2118 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2119 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2121 * ================== PLATFORM CHANGES =================
2123 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2124 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2126 * Support for MIPS DSP ASE on MIPS32 platforms.
2128 * Support for s390x Decimal Floating Point instructions on hosts that
2129 have the DFP facility installed.
2131 * Support for POWER8 (Power ISA 2.07) instructions
2133 * Support for Intel AVX2 instructions. This is available only on 64
2136 * Initial support for Intel Transactional Synchronization Extensions,
2139 * Initial support for Hardware Transactional Memory on POWER.
2141 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2142 run large GUI apps tolerably well.
2144 * ==================== TOOL CHANGES ====================
2148 - Improvements in handling of vectorised code, leading to
2149 significantly fewer false error reports. You need to use the flag
2150 --partial-loads-ok=yes to get the benefits of these changes.
2152 - Better control over the leak checker. It is now possible to
2153 specify which leak kinds (definite/indirect/possible/reachable)
2154 should be displayed, which should be regarded as errors, and which
2155 should be suppressed by a given leak suppression. This is done
2156 using the options --show-leak-kinds=kind1,kind2,..,
2157 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2158 "match-leak-kinds:" line in suppression entries, respectively.
2160 Note that generated leak suppressions contain this new line and
2161 are therefore more specific than in previous releases. To get the
2162 same behaviour as previous releases, remove the "match-leak-kinds:"
2163 line from generated suppressions before using them.
2165 - Reduced "possible leak" reports from the leak checker by the use
2166 of better heuristics. The available heuristics provide detection
2167 of valid interior pointers to std::stdstring, to new[] allocated
2168 arrays with elements having destructors and to interior pointers
2169 pointing to an inner part of a C++ object using multiple
2170 inheritance. They can be selected individually using the
2171 option --leak-check-heuristics=heur1,heur2,...
2173 - Better control of stacktrace acquisition for heap-allocated
2174 blocks. Using the --keep-stacktraces option, it is possible to
2175 control independently whether a stack trace is acquired for each
2176 allocation and deallocation. This can be used to create better
2177 "use after free" errors or to decrease Valgrind's resource
2178 consumption by recording less information.
2180 - Better reporting of leak suppression usage. The list of used
2181 suppressions (shown when the -v option is given) now shows, for
2182 each leak suppressions, how many blocks and bytes it suppressed
2183 during the last leak search.
2187 - False errors resulting from the use of statically initialised
2188 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2191 - False errors resulting from the use of pthread_cond_waits that
2192 timeout, have been removed.
2194 * ==================== OTHER CHANGES ====================
2196 * Some attempt to tune Valgrind's space requirements to the expected
2197 capabilities of the target:
2199 - The default size of the translation cache has been reduced from 8
2200 sectors to 6 on Android platforms, since each sector occupies
2201 about 40MB when using Memcheck.
2203 - The default size of the translation cache has been increased to 16
2204 sectors on all other platforms, reflecting the fact that large
2205 applications require instrumentation and storage of huge amounts
2206 of code. For similar reasons, the number of memory mapped
2207 segments that can be tracked has been increased by a factor of 6.
2209 - In all cases, the maximum number of sectors in the translation
2210 cache can be controlled by the new flag --num-transtab-sectors.
2212 * Changes in how debug info (line numbers, etc) is read:
2214 - Valgrind no longer temporarily mmaps the entire object to read
2215 from it. Instead, reading is done through a small fixed sized
2216 buffer. This avoids virtual memory usage spikes when Valgrind
2217 reads debuginfo from large shared objects.
2219 - A new experimental remote debug info server. Valgrind can read
2220 debug info from a different machine (typically, a build host)
2221 where debuginfo objects are stored. This can save a lot of time
2222 and hassle when running Valgrind on resource-constrained targets
2223 (phones, tablets) when the full debuginfo objects are stored
2224 somewhere else. This is enabled by the --debuginfo-server=
2227 - Consistency checking between main and debug objects can be
2228 disabled using the --allow-mismatched-debuginfo option.
2230 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2231 scanning can recover stack traces in some cases when the normal
2232 unwind mechanisms fail. Stack scanning is best described as "a
2233 nasty, dangerous and misleading hack" and so is disabled by default.
2234 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2237 * Detection and merging of recursive stack frame cycles. When your
2238 program has recursive algorithms, this limits the memory used by
2239 Valgrind for recorded stack traces and avoids recording
2240 uninteresting repeated calls. This is controlled by the command
2241 line option --merge-recursive-frame and by the monitor command
2242 "v.set merge-recursive-frames".
2244 * File name and line numbers for used suppressions. The list of used
2245 suppressions (shown when the -v option is given) now shows, for each
2246 used suppression, the file name and line number where the suppression
2249 * New and modified GDB server monitor features:
2251 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2252 that can be used to execute gdbserver monitor commands from the
2255 - A new monitor command, "v.info open_fds", that gives the list of
2256 open file descriptors and additional details.
2258 - An optional message in the "v.info n_errs_found" monitor command,
2259 for example "v.info n_errs_found test 1234 finished", allowing a
2260 comment string to be added to the process output, perhaps for the
2261 purpose of separating errors of different tests or test phases.
2263 - A new monitor command "v.info execontext" that shows information
2264 about the stack traces recorded by Valgrind.
2266 - A new monitor command "v.do expensive_sanity_check_general" to run
2267 some internal consistency checks.
2269 * New flag --sigill-diagnostics to control whether a diagnostic
2270 message is printed when the JIT encounters an instruction it can't
2271 translate. The actual behavior -- delivery of SIGILL to the
2272 application -- is unchanged.
2274 * The maximum amount of memory that Valgrind can use on 64 bit targets
2275 has been increased from 32GB to 64GB. This should make it possible
2276 to run applications on Memcheck that natively require up to about 35GB.
2278 * ==================== FIXED BUGS ====================
2280 The following bugs have been fixed or resolved. Note that "n-i-bz"
2281 stands for "not in bugzilla" -- that is, a bug that was reported to us
2282 but never got a bugzilla entry. We encourage you to file bugs in
2283 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2284 than mailing the developers (or mailing lists) directly -- bugs that
2285 are not entered into bugzilla tend to get forgotten about or ignored.
2287 To see details of a given bug, visit
2288 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2289 where XXXXXX is the bug number as listed below.
2291 123837 system call: 4th argument is optional, depending on cmd
2292 135425 memcheck should tell you where Freed blocks were Mallocd
2293 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2294 207815 Adds some of the drm ioctls to syswrap-linux.c
2295 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2296 252955 Impossible to compile with ccache
2297 253519 Memcheck reports auxv pointer accesses as invalid reads.
2298 263034 Crash when loading some PPC64 binaries
2299 269599 Increase deepest backtrace
2300 274695 s390x: Support "compare to/from logical" instructions (z196)
2301 275800 s390x: Autodetect cache info (part 2)
2302 280271 Valgrind reports possible memory leaks on still-reachable std::string
2303 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2304 289578 Backtraces with ARM unwind tables (stack scan flags)
2305 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2306 304832 ppc32: build failure
2307 305431 Use find_buildid shdr fallback for separate .debug files
2308 305728 Add support for AVX2 instructions
2309 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2310 306035 s390x: Fix IR generation for LAAG and friends
2311 306054 s390x: Condition code computation for convert-to-int/logical
2312 306098 s390x: alternate opcode form for convert to/from fixed
2313 306587 Fix cache line detection from auxiliary vector for PPC.
2314 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2315 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2316 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2317 307101 sys_capget second argument can be NULL
2318 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2319 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2320 307113 s390x: DFP support
2321 307141 valgrind does't work in mips-linux system
2322 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2323 307285 x86_amd64 feature test for avx in test suite is wrong
2324 307290 memcheck overlap testcase needs memcpy version filter
2325 307463 Please add "&limit=0" to the "all open bugs" link
2326 307465 --show-possibly-lost=no should reduce the error count / exit code
2327 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2328 307729 pkgconfig support broken valgrind.pc
2329 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2330 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2331 308089 Unhandled syscall on ppc64: prctl
2332 308135 PPC32 MPC8xx has 16 bytes cache size
2333 308321 testsuite memcheck filter interferes with gdb_filter
2335 308341 vgdb should report process exit (or fatal signal)
2336 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2337 308495 Remove build dependency on installed Xen headers
2338 308573 Internal error on 64-bit instruction executed in 32-bit mode
2340 308627 pmovmskb validity bit propagation is imprecise
2341 308644 vgdb command for having the info for the track-fds option
2342 308711 give more info about aspacemgr and arenas in out_of_memory
2343 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2344 308718 ARM implement SMLALBB family of instructions
2345 308886 Missing support for PTRACE_SET/GETREGSET
2346 308930 syscall name_to_handle_at (303 on amd64) not handled
2347 309229 V-bit tester does not report number of tests generated
2348 309323 print unrecognized instuction on MIPS
2349 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2350 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2351 309430 Self hosting ppc64 encounters a vassert error on operand type
2352 309600 valgrind is a bit confused about 0-sized sections
2353 309823 Generate errors for still reachable blocks
2354 309921 PCMPISTRI validity bit propagation is imprecise
2355 309922 none/tests/ppc64/test_dfp5 sometimes fails
2356 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2357 310424 --read-var-info does not properly describe static variables
2358 310792 search additional path for debug symbols
2359 310931 s390x: Message-security assist (MSA) instruction extension [..]
2360 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2361 311318 ARM: "128-bit constant is not implemented" error message
2362 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2363 311690 V crashes because it redirects branches inside of a redirected function
2364 311880 x86_64: make regtest hangs at shell_valid1
2365 311922 WARNING: unhandled syscall: 170
2367 312171 ppc: insn selection for DFP
2368 312571 Rounding mode call wrong for the DFP Iops [..]
2369 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2370 312913 Dangling pointers error should also report the alloc stack trace
2371 312980 Building on Mountain Lion generates some compiler warnings
2372 313267 Adding MIPS64/Linux port to Valgrind
2375 313811 Buffer overflow in assert_fail
2376 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2377 314269 ppc: dead code in insn selection
2378 314718 ARM: implement integer divide instruction (sdiv and udiv)
2379 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2380 315441 sendmsg syscall should ignore unset msghdr msg_flags
2381 315534 msgrcv inside a thread causes valgrind to hang (block)
2382 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2383 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2384 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2385 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2386 316144 valgrind.1 manpage contains unknown ??? strings [..]
2387 316145 callgrind command line options in manpage reference (unknown) [..]
2388 316145 callgrind command line options in manpage reference [..]
2389 316181 drd: Fixed a 4x slowdown for certain applications
2390 316503 Valgrind does not support SSE4 "movntdqa" instruction
2391 316535 Use of |signed int| instead of |size_t| in valgrind messages
2392 316696 fluidanimate program of parsec 2.1 stuck
2393 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2394 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2395 317186 "Impossible happens" when occurs VCVT instruction on ARM
2396 317318 Support for Threading Building Blocks "scalable_malloc"
2397 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2398 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2399 317463 bmi testcase IR SANITY CHECK FAILURE
2400 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2401 318050 libmpiwrap fails to compile with out-of-source build
2402 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2403 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2404 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2405 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2406 318932 Add missing PPC64 and PPC32 system call support
2407 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2408 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2409 319494 VEX Makefile-gcc standalone build update after r2702
2410 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2411 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2412 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2413 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2414 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2415 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2416 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2417 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2418 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2419 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2420 320895 add fanotify support (patch included)
2421 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2422 321065 Valgrind updates for Xen 4.3
2423 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2424 321363 Unhandled instruction: SSAX (ARM + Thumb)
2425 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2426 321466 Unhandled instruction: SHASX (ARM + Thumb)
2427 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2428 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2429 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2430 321620 Unhandled instruction: UASX (ARM + Thumb)
2431 321621 Unhandled instruction: USAX (ARM + Thumb)
2432 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2433 321693 Unhandled instruction: LDRSBT (Thumb)
2434 321694 Unhandled instruction: UQASX (ARM + Thumb)
2435 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2436 321697 Unhandled instruction: UHASX (ARM + Thumb)
2437 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2438 321704 Unhandled instruction: REVSH (ARM + Thumb)
2439 321730 Add cg_diff and cg_merge man pages
2440 321738 Add vgdb and valgrind-listener man pages
2442 321891 Unhandled instruction: LDRHT (Thumb)
2443 321960 pthread_create() then alloca() causing invalid stack write errors
2444 321969 ppc32 and ppc64 don't support [lf]setxattr
2445 322254 Show threadname together with tid if set by application
2446 322294 Add initial support for IBM Power ISA 2.07
2447 322368 Assertion failure in wqthread_hijack under OS X 10.8
2448 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2449 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2450 322851 0bXXX binary literal syntax is not standard
2451 323035 Unhandled instruction: LDRSHT(Thumb)
2452 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2453 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2454 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2455 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2456 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2457 323437 Phase 2 support for IBM Power ISA 2.07
2458 323713 Support mmxext (integer sse) subset on i386 (athlon)
2459 323803 Transactional memory instructions are not supported for Power
2460 323893 SSE3 not available on amd cpus in valgrind
2461 323905 Probable false positive from Valgrind/drd on close()
2462 323912 valgrind.h header isn't compatible for mingw64
2463 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
2464 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
2465 324181 mmap does not handle MAP_32BIT
2466 324227 memcheck false positive leak when a thread calls exit+block [..]
2467 324421 Support for fanotify API on ARM architecture
2468 324514 gdbserver monitor cmd output behaviour consistency [..]
2469 324518 ppc64: Emulation of dcbt instructions does not handle [..]
2470 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
2471 324582 When access is made to freed memory, report both allocation [..]
2472 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
2473 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
2474 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
2475 324834 Unhandled instructions in Microsoft C run-time for x86_64
2476 324894 Phase 3 support for IBM Power ISA 2.07
2477 326091 drd: Avoid false race reports from optimized strlen() impls
2478 326113 valgrind libvex hwcaps error on AMD64
2479 n-i-bz Some wrong command line options could be ignored
2480 n-i-bz patch to allow fair-sched on android
2481 n-i-bz report error for vgdb snapshot requested before execution
2482 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
2484 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
2488 Release 3.8.1 (19 September 2012)
2489 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2490 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
2491 that occur moderately frequently in real use cases, adds support for
2492 some missing instructions on ARM, and fixes a deadlock condition on
2493 MacOSX. If you package or deliver 3.8.0 for others to use, you might
2494 want to consider upgrading to 3.8.1 instead.
2496 The following bugs have been fixed or resolved. Note that "n-i-bz"
2497 stands for "not in bugzilla" -- that is, a bug that was reported to us
2498 but never got a bugzilla entry. We encourage you to file bugs in
2499 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2500 than mailing the developers (or mailing lists) directly -- bugs that
2501 are not entered into bugzilla tend to get forgotten about or ignored.
2503 To see details of a given bug, visit
2504 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2505 where XXXXXX is the bug number as listed below.
2508 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
2509 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
2510 298281 wcslen causes false(?) uninitialised value warnings
2511 301281 valgrind hangs on OS X when the process calls system()
2512 304035 disInstr(arm): unhandled instruction 0xE1023053
2513 304867 implement MOVBE instruction in x86 mode
2514 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
2515 305042 amd64: implement 0F 7F encoding of movq between two registers
2516 305199 ARM: implement QDADD and QDSUB
2517 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
2518 305513 killed by fatal signal: SIGSEGV
2519 305690 DRD reporting invalid semaphore when sem_trywait fails
2520 305926 Invalid alignment checks for some AVX instructions
2521 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
2522 306310 3.8.0 release tarball missing some files
2523 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
2524 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
2525 n-i-bz shmat of a segment > 4Gb does not work
2526 n-i-bz simulate_control_c script wrong USR1 signal number on mips
2527 n-i-bz vgdb ptrace calls wrong on mips [...]
2528 n-i-bz Fixes for more MPI false positives
2529 n-i-bz exp-sgcheck's memcpy causes programs to segfault
2530 n-i-bz OSX build w/ clang: asserts at startup
2531 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
2532 n-i-bz fix a couple of union tag-vs-field mixups
2533 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
2535 The following bugs were fixed in 3.8.0 but not listed in this NEWS
2538 254088 Valgrind should know about UD2 instruction
2541 304754 NEWS blows TeX's little mind
2543 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
2547 Release 3.8.0 (10 August 2012)
2548 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2549 3.8.0 is a feature release with many improvements and the usual
2550 collection of bug fixes.
2552 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2553 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
2554 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
2555 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
2556 There is initial support for MacOSX 10.8, but it is not usable for
2557 serious work at present.
2559 * ================== PLATFORM CHANGES =================
2561 * Support for MIPS32 platforms running Linux. Valgrind has been
2562 tested on MIPS32 and MIPS32r2 platforms running different Debian
2563 Squeeze and MeeGo distributions. Both little-endian and big-endian
2564 cores are supported. The tools Memcheck, Massif and Lackey have
2565 been tested and are known to work. See README.mips for more details.
2567 * Preliminary support for Android running on x86.
2569 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
2571 * Support for Intel AVX instructions and for AES instructions. This
2572 support is available only for 64 bit code.
2574 * Support for POWER Decimal Floating Point instructions.
2576 * ==================== TOOL CHANGES ====================
2578 * Non-libc malloc implementations are now supported. This is useful
2579 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
2580 Using the new option --soname-synonyms, such tools can be informed
2581 that the malloc implementation is either linked statically into the
2582 executable, or is present in some other shared library different
2583 from libc.so. This makes it possible to process statically linked
2584 programs, and programs using other malloc libraries, for example
2585 TCMalloc or JEMalloc.
2587 * For tools that provide their own replacement for malloc et al, the
2588 option --redzone-size=<number> allows users to specify the size of
2589 the padding blocks (redzones) added before and after each client
2590 allocated block. Smaller redzones decrease the memory needed by
2591 Valgrind. Bigger redzones increase the chance to detect blocks
2592 overrun or underrun. Prior to this change, the redzone size was
2593 hardwired to 16 bytes in Memcheck.
2597 - The leak_check GDB server monitor command now can
2598 control the maximum nr of loss records to output.
2600 - Reduction of memory use for applications allocating
2601 many blocks and/or having many partially defined bytes.
2603 - Addition of GDB server monitor command 'block_list' that lists
2604 the addresses/sizes of the blocks of a leak search loss record.
2606 - Addition of GDB server monitor command 'who_points_at' that lists
2607 the locations pointing at a block.
2609 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2610 detect an invalid access of these redzones, by marking them
2611 noaccess. Similarly, if a redzone size is given for a memory
2612 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2613 This still allows to find some bugs if the user has forgotten to
2614 mark the pool superblock noaccess.
2616 - Performance of memory leak check has been improved, especially in
2617 cases where there are many leaked blocks and/or many suppression
2618 rules used to suppress leak reports.
2620 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2621 more precise analysis, which is important for LLVM/Clang
2622 generated code. This is at the cost of somewhat reduced
2623 performance. Note there is no change to analysis precision or
2624 costs on Linux targets.
2628 - Added even more facilities that can help finding the cause of a data
2629 race, namely the command-line option --ptrace-addr and the macro
2630 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2632 - Fixed a subtle bug that could cause false positive data race reports.
2634 * ==================== OTHER CHANGES ====================
2636 * The C++ demangler has been updated so as to work well with C++
2637 compiled by up to at least g++ 4.6.
2639 * Tool developers can make replacement/wrapping more flexible thanks
2640 to the new option --soname-synonyms. This was reported above, but
2641 in fact is very general and applies to all function
2642 replacement/wrapping, not just to malloc-family functions.
2644 * Round-robin scheduling of threads can be selected, using the new
2645 option --fair-sched= yes. Prior to this change, the pipe-based
2646 thread serialisation mechanism (which is still the default) could
2647 give very unfair scheduling. --fair-sched=yes improves
2648 responsiveness of interactive multithreaded applications, and
2649 improves repeatability of results from the thread checkers Helgrind
2652 * For tool developers: support to run Valgrind on Valgrind has been
2653 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2655 * gdbserver now shows the float shadow registers as integer
2656 rather than float values, as the shadow values are mostly
2657 used as bit patterns.
2659 * Increased limit for the --num-callers command line flag to 500.
2661 * Performance improvements for error matching when there are many
2662 suppression records in use.
2664 * Improved support for DWARF4 debugging information (bug 284184).
2666 * Initial support for DWZ compressed Dwarf debug info.
2668 * Improved control over the IR optimiser's handling of the tradeoff
2669 between performance and precision of exceptions. Specifically,
2670 --vex-iropt-precise-memory-exns has been removed and replaced by
2671 --vex-iropt-register-updates, with extended functionality. This
2672 allows the Valgrind gdbserver to always show up to date register
2675 * Modest performance gains through the use of translation chaining for
2678 * ==================== FIXED BUGS ====================
2680 The following bugs have been fixed or resolved. Note that "n-i-bz"
2681 stands for "not in bugzilla" -- that is, a bug that was reported to us
2682 but never got a bugzilla entry. We encourage you to file bugs in
2683 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2684 than mailing the developers (or mailing lists) directly -- bugs that
2685 are not entered into bugzilla tend to get forgotten about or ignored.
2687 To see details of a given bug, visit
2688 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2689 where XXXXXX is the bug number as listed below.
2691 197914 Building valgrind from svn now requires automake-1.10
2692 203877 increase to 16Mb maximum allowed alignment for memalign et al
2693 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
2694 247386 make perf does not run all performance tests
2695 270006 Valgrind scheduler unfair
2696 270777 Adding MIPS/Linux port to Valgrind
2697 270796 s390x: Removed broken support for the TS insn
2698 271438 Fix configure for proper SSE4.2 detection
2699 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
2700 273475 Add support for AVX instructions
2701 274078 improved configure logic for mpicc
2702 276993 fix mremap 'no thrash checks'
2703 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
2704 281482 memcheck incorrect byte allocation count in realloc() for silly argument
2705 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
2706 283413 Fix wrong sanity check
2707 283671 Robustize alignment computation in LibVEX_Alloc
2708 283961 Adding support for some HCI IOCTLs
2709 284124 parse_type_DIE: confused by: DWARF 4
2710 284864 == 273475 (Add support for AVX instructions)
2711 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
2712 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
2713 285725 == 273475 (Add support for AVX instructions)
2714 286261 add wrapper for linux I2C_RDWR ioctl
2715 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
2716 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
2717 286384 configure fails "checking for a supported version of gcc"
2718 286497 == 273475 (Add support for AVX instructions)
2719 286596 == 273475 (Add support for AVX instructions)
2720 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
2721 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
2722 287260 Incorrect conditional jump or move depends on uninitialised value(s)
2723 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
2724 287307 == 273475 (Add support for AVX instructions)
2725 287858 VG_(strerror): unknown error
2726 288298 (MacOSX) unhandled syscall shm_unlink
2727 288995 == 273475 (Add support for AVX instructions)
2728 289470 Loading of large Mach-O thin binaries fails.
2729 289656 == 273475 (Add support for AVX instructions)
2730 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
2731 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
2732 289839 s390x: Provide support for unicode conversion instructions
2733 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
2734 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
2735 290655 Add support for AESKEYGENASSIST instruction
2736 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
2737 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
2738 291253 ES register not initialised in valgrind simulation
2739 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
2740 291865 s390x: Support the "Compare Double and Swap" family of instructions
2741 292300 == 273475 (Add support for AVX instructions)
2742 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
2743 292493 == 273475 (Add support for AVX instructions)
2744 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
2745 292627 Missing support for some SCSI ioctls
2746 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
2747 292841 == 273475 (Add support for AVX instructions)
2748 292993 implement the getcpu syscall on amd64-linux
2749 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
2750 293088 Add some VEX sanity checks for ppc64 unhandled instructions
2751 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
2752 293754 PCMPxSTRx not implemented for 16-bit characters
2753 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
2754 293808 CLFLUSH not supported by latest VEX for amd64
2755 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
2756 294048 MPSADBW instruction not implemented
2757 294055 regtest none/tests/shell fails when locale is not set to C
2758 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
2759 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
2760 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
2761 294260 disInstr_AMD64: disInstr miscalculated next %rip
2762 294523 --partial-loads-ok=yes causes false negatives
2763 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
2764 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
2765 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
2766 295089 can not annotate source for both helgrind and drd
2767 295221 POWER Processor decimal floating point instruction support missing
2768 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
2769 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
2770 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
2771 295617 ARM - Add some missing syscalls
2772 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
2773 296229 Linux user input device ioctls missing wrappers
2774 296318 ELF Debug info improvements (more than one rx/rw mapping)
2775 296422 Add translation chaining support
2776 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
2777 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
2778 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
2779 297078 gdbserver signal handling problems [..]
2780 297147 drd false positives on newly allocated memory
2781 297329 disallow decoding of IBM Power DFP insns on some machines
2782 297497 POWER Processor decimal floating point instruction support missing
2783 297701 Another alias for strncasecmp_l in libc-2.13.so
2784 297911 'invalid write' not reported when using APIs for custom mem allocators.
2785 297976 s390x: revisit EX implementation
2786 297991 Valgrind interferes with mmap()+ftell()
2787 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
2788 297993 Fix compilation of valgrind with gcc -g3.
2789 298080 POWER Processor DFP support missing, part 3
2790 298227 == 273475 (Add support for AVX instructions)
2791 298335 == 273475 (Add support for AVX instructions)
2792 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
2793 298394 s390x: Don't bail out on an unknown machine model. [..]
2794 298421 accept4() syscall (366) support is missing for ARM
2795 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
2796 298732 valgrind installation problem in ubuntu with kernel version 3.x
2797 298862 POWER Processor DFP instruction support missing, part 4
2798 298864 DWARF reader mis-parses DW_FORM_ref_addr
2799 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
2800 299053 Support DWARF4 DW_AT_high_pc constant form
2801 299104 == 273475 (Add support for AVX instructions)
2802 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
2803 299629 dup3() syscall (358) support is missing for ARM
2804 299694 POWER Processor DFP instruction support missing, part 5
2805 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
2806 299803 == 273475 (Add support for AVX instructions)
2807 299804 == 273475 (Add support for AVX instructions)
2808 299805 == 273475 (Add support for AVX instructions)
2809 300140 ARM - Missing (T1) SMMUL
2810 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
2811 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
2812 300414 FCOM and FCOMP unimplemented for amd64 guest
2813 301204 infinite loop in canonicaliseSymtab with ifunc symbol
2814 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
2815 301265 add x86 support to Android build
2816 301984 configure script doesn't detect certain versions of clang
2817 302205 Fix compiler warnings for POWER VEX code and POWER test cases
2818 302287 Unhandled movbe instruction on Atom processors
2819 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
2820 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
2821 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
2822 302656 == 273475 (Add support for AVX instructions)
2823 302709 valgrind for ARM needs extra tls support for android emulator [..]
2824 302827 add wrapper for CDROM_GET_CAPABILITY
2825 302901 Valgrind crashes with dwz optimized debuginfo
2826 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
2827 303116 Add support for the POWER instruction popcntb
2828 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
2829 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
2830 303466 == 273475 (Add support for AVX instructions)
2831 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
2832 303963 strstr() function produces wrong results under valgrind callgrind
2833 304054 CALL_FN_xx macros need to enforce stack alignment
2834 304561 tee system call not supported
2835 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
2836 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
2837 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
2838 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
2839 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
2840 n-i-bz s390x: Shadow registers can now be examined using vgdb
2842 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
2843 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
2847 Release 3.7.0 (5 November 2011)
2848 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2849 3.7.0 is a feature release with many significant improvements and the
2850 usual collection of bug fixes.
2852 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2853 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
2854 Support for recent distros and toolchain components (glibc 2.14, gcc
2855 4.6, MacOSX 10.7) has been added.
2857 * ================== PLATFORM CHANGES =================
2859 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
2860 analyse 64-bit programs running on z/Architecture. Most user space
2861 instructions up to and including z10 are supported. Valgrind has
2862 been tested extensively on z9, z10, and z196 machines running SLES
2863 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
2864 known to work well. Callgrind, Helgrind, and DRD work reasonably
2865 well on z9 and later models. See README.s390 for more details.
2867 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
2868 64-bit processes are supported. Some complex threaded applications
2869 (Firefox) are observed to hang when run as 32 bit applications,
2870 whereas 64-bit versions run OK. The cause is unknown. Memcheck
2871 will likely report some false errors. In general, expect some rough
2872 spots. This release also supports MacOSX 10.6, but drops support
2875 * Preliminary support for Android (on ARM). Valgrind can now run
2876 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
2877 README.android for more details, plus instructions on how to get
2880 * Support for the IBM Power ISA 2.06 (Power7 instructions)
2882 * General correctness and performance improvements for ARM/Linux, and,
2883 by extension, ARM/Android.
2885 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
2886 instruction set support is under development but is not available in
2889 * Support for AIX5 has been removed.
2891 * ==================== TOOL CHANGES ====================
2893 * Memcheck: some incremental changes:
2895 - reduction of memory use in some circumstances
2897 - improved handling of freed memory, which in some circumstances
2898 can cause detection of use-after-free that would previously have
2901 - fix of a longstanding bug that could cause false negatives (missed
2902 errors) in programs doing vector saturated narrowing instructions.
2904 * Helgrind: performance improvements and major memory use reductions,
2905 particularly for large, long running applications which perform many
2906 synchronisation (lock, unlock, etc) events. Plus many smaller
2909 - display of locksets for both threads involved in a race
2911 - general improvements in formatting/clarity of error messages
2913 - addition of facilities and documentation regarding annotation
2914 of thread safe reference counted C++ classes
2916 - new flag --check-stack-refs=no|yes [yes], to disable race checking
2917 on thread stacks (a performance hack)
2919 - new flag --free-is-write=no|yes [no], to enable detection of races
2920 where one thread accesses heap memory but another one frees it,
2921 without any coordinating synchronisation event
2923 * DRD: enabled XML output; added support for delayed thread deletion
2924 in order to detect races that occur close to the end of a thread
2925 (--join-list-vol); fixed a memory leak triggered by repeated client
2926 memory allocatation and deallocation; improved Darwin support.
2928 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
2930 * exp-sgcheck: this tool has been reduced in scope so as to improve
2931 performance and remove checking that Memcheck does better.
2932 Specifically, the ability to check for overruns for stack and global
2933 arrays is unchanged, but the ability to check for overruns of heap
2934 blocks has been removed. The tool has accordingly been renamed to
2935 exp-sgcheck ("Stack and Global Array Checking").
2937 * ==================== OTHER CHANGES ====================
2939 * GDB server: Valgrind now has an embedded GDB server. That means it
2940 is possible to control a Valgrind run from GDB, doing all the usual
2941 things that GDB can do (single stepping, breakpoints, examining
2942 data, etc). Tool-specific functionality is also available. For
2943 example, it is possible to query the definedness state of variables
2944 or memory from within GDB when running Memcheck; arbitrarily large
2945 memory watchpoints are supported, etc. To use the GDB server, start
2946 Valgrind with the flag --vgdb-error=0 and follow the on-screen
2949 * Improved support for unfriendly self-modifying code: a new option
2950 --smc-check=all-non-file is available. This adds the relevant
2951 consistency checks only to code that originates in non-file-backed
2952 mappings. In effect this confines the consistency checking only to
2953 code that is or might be JIT generated, and avoids checks on code
2954 that must have been compiled ahead of time. This significantly
2955 improves performance on applications that generate code at run time.
2957 * It is now possible to build a working Valgrind using Clang-2.9 on
2960 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
2961 These enable and disable error reporting on a per-thread, and
2962 nestable, basis. This is useful for hiding errors in particularly
2963 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
2964 now uses this facility.
2966 * Added the --mod-funcname option to cg_diff.
2968 * ==================== FIXED BUGS ====================
2970 The following bugs have been fixed or resolved. Note that "n-i-bz"
2971 stands for "not in bugzilla" -- that is, a bug that was reported to us
2972 but never got a bugzilla entry. We encourage you to file bugs in
2973 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2974 mailing the developers (or mailing lists) directly -- bugs that are
2975 not entered into bugzilla tend to get forgotten about or ignored.
2977 To see details of a given bug, visit
2978 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2979 where XXXXXX is the bug number as listed below.
2981 79311 malloc silly arg warning does not give stack trace
2982 210935 port valgrind.h (not valgrind) to win32 to support client requests
2983 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
2984 243404 Port to zSeries
2985 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
2986 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
2987 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
2988 253206 Some fixes for the faultstatus testcase
2989 255223 capget testcase fails when running as root
2990 256703 xlc_dbl_u32.c testcase broken
2991 256726 Helgrind tests have broken inline asm
2992 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
2993 264800 testcase compile failure on zseries
2994 265762 make public VEX headers compilable by G++ 3.x
2995 265771 assertion in jumps.c (r11523) fails with glibc-2.3
2996 266753 configure script does not give the user the option to not use QtCore
2997 266931 gen_insn_test.pl is broken
2998 266961 ld-linux.so.2 i?86-linux strlen issues
2999 266990 setns instruction causes false positive
3000 267020 Make directory for temporary files configurable at run-time.
3001 267342 == 267997 (segmentation fault on Mac OS 10.6)
3002 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3003 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3004 267488 regtest: darwin support for 64-bit build
3005 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3006 267630 Add support for IBM Power ISA 2.06 -- stage 1
3007 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3008 267819 Add client request for informing the core about reallocation
3009 267925 laog data structure quadratic for a single sequence of lock
3010 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3011 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3012 268513 missed optimizations in fold_Expr
3013 268619 s390x: fpr - gpr transfer facility
3014 268620 s390x: reconsider "long displacement" requirement
3015 268621 s390x: improve IR generation for XC
3016 268715 s390x: FLOGR is not universally available
3017 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3018 268930 s390x: MHY is not universally available
3019 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3020 269079 Support ptrace system call on ARM
3021 269144 missing "Bad option" error message
3022 269209 conditional load and store facility (z196)
3023 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3024 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3025 269736 s390x: minor code generation tweaks
3026 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3027 269863 s390x: remove unused function parameters
3028 269864 s390x: tweak s390_emit_load_cc
3029 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3030 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3031 270115 s390x: rewrite some testcases
3032 270309 == 267997 (valgrind crash on startup)
3033 270320 add support for Linux FIOQSIZE ioctl() call
3034 270326 segfault while trying to sanitize the environment passed to execle
3035 270794 IBM POWER7 support patch causes regression in none/tests
3036 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3037 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3038 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3039 270959 s390x: invalid use of R0 as base register
3040 271042 VSX configure check fails when it should not
3041 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3042 271259 s390x: fix code confusion
3043 271337 == 267997 (Valgrind segfaults on MacOS X)
3044 271385 s390x: Implement Ist_MBE
3045 271501 s390x: misc cleanups
3046 271504 s390x: promote likely and unlikely
3047 271579 ppc: using wrong enum type
3048 271615 unhandled instruction "popcnt" (arch=amd10h)
3049 271730 Fix bug when checking ioctls: duplicate check
3050 271776 s390x: provide STFLE instruction support
3051 271779 s390x: provide clock instructions like STCK
3052 271799 Darwin: ioctls without an arg report a memory error
3053 271820 arm: fix type confusion
3054 271917 pthread_cond_timedwait failure leads to not-locked false positive
3055 272067 s390x: fix DISP20 macro
3056 272615 A typo in debug output in mc_leakcheck.c
3057 272661 callgrind_annotate chokes when run from paths containing regex chars
3058 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3059 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3060 272967 make documentation build-system more robust
3061 272986 Fix gcc-4.6 warnings with valgrind.h
3062 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3063 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3064 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3065 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3066 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3067 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3068 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3069 273778 exp-ptrcheck: unhandled sysno == 259
3070 274089 exp-ptrcheck: unhandled sysno == 208
3071 274378 s390x: Various dispatcher tweaks
3072 274447 WARNING: unhandled syscall: 340
3073 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3074 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3075 274926 valgrind does not build against linux-3
3076 275148 configure FAIL with glibc-2.14
3077 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3078 275168 Make Valgrind work for MacOSX 10.7 Lion
3079 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3080 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3081 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3082 275308 Fix implementation for ppc64 fres instruc
3083 275339 s390x: fix testcase compile warnings
3084 275517 s390x: Provide support for CKSM instruction
3085 275710 s390x: get rid of redundant address mode calculation
3086 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3087 275852 == 250101 (valgrind uses all swap space and is killed)
3088 276784 Add support for IBM Power ISA 2.06 -- stage 3
3089 276987 gdbsrv: fix tests following recent commits
3090 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3091 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3092 277471 Unhandled syscall: 340
3093 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3094 277653 ARM: support Thumb2 PLD instruction
3095 277663 ARM: NEON float VMUL by scalar incorrect
3096 277689 ARM: tests for VSTn with register post-index are broken
3097 277694 ARM: BLX LR instruction broken in ARM mode
3098 277780 ARM: VMOV.F32 (immediate) instruction is broken
3099 278057 fuse filesystem syscall deadlocks
3100 278078 Unimplemented syscall 280 on ppc32
3101 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3102 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3103 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3104 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3105 279027 Support for MVCL and CLCL instruction
3106 279027 s390x: Provide support for CLCL and MVCL instructions
3107 279062 Remove a redundant check in the insn selector for ppc.
3108 279071 JDK creates PTEST with redundant REX.W prefix
3109 279212 gdbsrv: add monitor cmd v.info scheduler.
3110 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3111 279698 memcheck discards valid-bits for packuswb
3112 279795 memcheck reports uninitialised values for mincore on amd64
3113 279994 Add support for IBM Power ISA 2.06 -- stage 3
3114 280083 mempolicy syscall check errors
3115 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3116 280710 s390x: config files for nightly builds
3117 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3118 280965 Valgrind breaks fcntl locks when program does mmap
3119 281138 WARNING: unhandled syscall: 340
3120 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3121 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3122 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3123 281468 s390x: handle do_clone and gcc clones in call traces
3124 281488 ARM: VFP register corruption
3125 281828 == 275284 (false memmove warning: "Source and destination overlap")
3126 281883 s390x: Fix system call wrapper for "clone".
3127 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3128 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3129 282238 SLES10: make check fails
3130 282979 strcasestr needs replacement with recent(>=2.12) glibc
3131 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3132 283243 Regression in ppc64 memcheck tests
3133 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3134 283427 re-connect epoll_pwait syscall on ARM linux
3135 283600 gdbsrv: android: port vgdb.c
3136 283709 none/tests/faultstatus needs to account for page size
3137 284305 filter_gdb needs enhancement to work on ppc64
3138 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3139 284472 Thumb2 ROR.W encoding T2 not implemented
3140 284621 XML-escape process command line in XML output
3141 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3142 that have non-power-of-2 sizes (also AMDs)
3143 n-i-bz don't be spooked by libraries mashed by elfhack
3144 n-i-bz don't be spooked by libxul.so linked with gold
3145 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3147 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3148 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3149 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3153 Release 3.6.1 (16 February 2011)
3154 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3155 3.6.1 is a bug fix release. It adds support for some SSE4
3156 instructions that were omitted in 3.6.0 due to lack of time. Initial
3157 support for glibc-2.13 has been added. A number of bugs causing
3158 crashing or assertion failures have been fixed.
3160 The following bugs have been fixed or resolved. Note that "n-i-bz"
3161 stands for "not in bugzilla" -- that is, a bug that was reported to us
3162 but never got a bugzilla entry. We encourage you to file bugs in
3163 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3164 mailing the developers (or mailing lists) directly -- bugs that are
3165 not entered into bugzilla tend to get forgotten about or ignored.
3167 To see details of a given bug, visit
3168 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3169 where XXXXXX is the bug number as listed below.
3171 188572 Valgrind on Mac should suppress setenv() mem leak
3172 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3173 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3174 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3175 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3176 254420 memory pool tracking broken
3177 254957 Test code failing to compile due to changes in memcheck.h
3178 255009 helgrind/drd: crash on chmod with invalid parameter
3179 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3180 255355 helgrind/drd: crash on threaded programs doing fork
3182 255418 (SSE4.x) rint call compiled with ICC
3183 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3184 255888 closing valgrindoutput tag outputted to log-stream on error
3185 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3186 255966 Slowness when using mempool annotations
3187 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3188 256600 super-optimized strcasecmp() false positive
3189 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3190 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3191 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3192 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3193 257276 Missing case in memcheck --track-origins=yes
3194 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3195 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3196 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3197 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3198 263099 callgrind_annotate counts Ir improperly [...]
3199 263877 undefined coprocessor instruction on ARMv7
3200 265964 configure FAIL with glibc-2.13
3201 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3202 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3203 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3204 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3205 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3207 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3211 Release 3.6.0 (21 October 2010)
3212 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3213 3.6.0 is a feature release with many significant improvements and the
3214 usual collection of bug fixes.
3216 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3217 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3218 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3220 -------------------------
3222 Here are some highlights. Details are shown further down:
3224 * Support for ARM/Linux.
3226 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3228 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3230 * Support for the SSE4.2 instruction set.
3232 * Enhancements to the Callgrind profiler, including the ability to
3233 handle CPUs with three levels of cache.
3235 * A new experimental heap profiler, DHAT.
3237 * A huge number of bug fixes and small enhancements.
3239 -------------------------
3241 Here are details of the above changes, together with descriptions of
3242 many other changes, and a list of fixed bugs.
3244 * ================== PLATFORM CHANGES =================
3246 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3247 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3248 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3250 This requires a CPU capable of running the ARMv7-A instruction set
3251 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3252 of the user space instruction set, including ARM and Thumb integer
3253 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3254 Cachegrind and Massif tools work properly; other tools work to
3257 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3258 with support for recent releases of the underlying toolchain
3259 components, notably gcc-4.5 and glibc-2.12.
3261 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3262 support also works much better on OS X 10.5, and is as solid as
3265 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3266 64-bit mode. In 32-bit mode, support is only available up to and
3267 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3268 supported in 64-bit mode, and 32-bit mode does in fact support the
3269 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3270 10.6 on 32-bit targets.
3272 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3273 and including version 2.05 is supported.
3275 * ==================== TOOL CHANGES ====================
3277 * Cachegrind has a new processing script, cg_diff, which finds the
3278 difference between two profiles. It's very useful for evaluating
3279 the performance effects of a change in a program.
3281 Related to this change, the meaning of cg_annotate's (rarely-used)
3282 --threshold option has changed; this is unlikely to affect many
3283 people, if you do use it please see the user manual for details.
3285 * Callgrind now can do branch prediction simulation, similar to
3286 Cachegrind. In addition, it optionally can count the number of
3287 executed global bus events. Both can be used for a better
3288 approximation of a "Cycle Estimation" as derived event (you need to
3289 update the event formula in KCachegrind yourself).
3291 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3292 rather than the L2 cache. This is to accommodate machines with
3293 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3294 cache configuration of such a machine it will run the simulation as
3295 if the L2 cache isn't present. This means the results are less
3296 likely to match the true result for the machine, but
3297 Cachegrind/Callgrind's results are already only approximate, and
3298 should not be considered authoritative. The results are still
3299 useful for giving a general idea about a program's locality.
3301 * Massif has a new option, --pages-as-heap, which is disabled by
3302 default. When enabled, instead of tracking allocations at the level
3303 of heap blocks (as allocated with malloc/new/new[]), it instead
3304 tracks memory allocations at the level of memory pages (as mapped by
3305 mmap, brk, etc). Each mapped page is treated as its own block.
3306 Interpreting the page-level output is harder than the heap-level
3307 output, but this option is useful if you want to account for every
3308 byte of memory used by a program.
3310 * DRD has two new command-line options: --free-is-write and
3311 --trace-alloc. The former allows to detect reading from already freed
3312 memory, and the latter allows tracing of all memory allocations and
3315 * DRD has several new annotations. Custom barrier implementations can
3316 now be annotated, as well as benign races on static variables.
3318 * DRD's happens before / happens after annotations have been made more
3319 powerful, so that they can now also be used to annotate e.g. a smart
3320 pointer implementation.
3322 * Helgrind's annotation set has also been drastically improved, so as
3323 to provide to users a general set of annotations to describe locks,
3324 semaphores, barriers and condition variables. Annotations to
3325 describe thread-safe reference counted heap objects have also been
3328 * Memcheck has a new command-line option, --show-possibly-lost, which
3329 is enabled by default. When disabled, the leak detector will not
3330 show possibly-lost blocks.
3332 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3333 has been added. DHAT keeps track of allocated heap blocks, and also
3334 inspects every memory reference to see which block (if any) is being
3335 accessed. This gives a lot of insight into block lifetimes,
3336 utilisation, turnover, liveness, and the location of hot and cold
3337 fields. You can use DHAT to do hot-field profiling.
3339 * ==================== OTHER CHANGES ====================
3341 * Improved support for unfriendly self-modifying code: the extra
3342 overhead incurred by --smc-check=all has been reduced by
3343 approximately a factor of 5 as compared with 3.5.0.
3345 * Ability to show directory names for source files in error messages.
3346 This is combined with a flexible mechanism for specifying which
3347 parts of the paths should be shown. This is enabled by the new flag
3350 * A new flag, --require-text-symbol, which will stop the run if a
3351 specified symbol is not found it a given shared object when it is
3352 loaded into the process. This makes advanced working with function
3353 intercepting and wrapping safer and more reliable.
3355 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3356 and control of Valgrind is now available for the tools Memcheck and
3357 Helgrind. XML output from Valgrind is available for Memcheck,
3358 Helgrind and exp-Ptrcheck.
3360 * More reliable stack unwinding on amd64-linux, particularly in the
3361 presence of function wrappers, and with gcc-4.5 compiled code.
3363 * Modest scalability (performance improvements) for massive
3364 long-running applications, particularly for those with huge amounts
3367 * Support for analyzing programs running under Wine with has been
3368 improved. The header files <valgrind/valgrind.h>,
3369 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3370 Windows-programs compiled with MinGW or one of the Microsoft Visual
3373 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3374 The 32-bit simulator was not affected. This did not occur often,
3375 but when it did would usually crash the program under test.
3378 * A large number of bugs were fixed. These are shown below.
3380 * A number of bugs were investigated, and were candidates for fixing,
3381 but are not fixed in 3.6.0, due to lack of developer time. They may
3382 get fixed in later releases. They are:
3384 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3385 212419 false positive "lock order violated" (A+B vs A)
3386 213685 Undefined value propagates past dependency breaking instruction
3387 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3388 237920 valgrind segfault on fork failure
3389 242137 support for code compiled by LLVM-2.8
3390 242423 Another unknown Intel cache config value
3391 243232 Inconsistent Lock Orderings report with trylock
3392 243483 ppc: callgrind triggers VEX assertion failure
3393 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3394 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3396 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3397 249435 Analyzing wine programs with callgrind triggers a crash
3398 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3399 250065 Handling large allocations
3400 250101 huge "free" memory usage due to m_mallocfree.c
3401 "superblocks fragmentation"
3402 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3403 252091 Callgrind on ARM does not detect function returns correctly
3404 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3405 254420 memory pool tracking broken
3406 n-i-bz support for adding symbols for JIT generated code
3409 The following bugs have been fixed or resolved. Note that "n-i-bz"
3410 stands for "not in bugzilla" -- that is, a bug that was reported to us
3411 but never got a bugzilla entry. We encourage you to file bugs in
3412 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3413 mailing the developers (or mailing lists) directly -- bugs that are
3414 not entered into bugzilla tend to get forgotten about or ignored.
3416 To see details of a given bug, visit
3417 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3418 where XXXXXX is the bug number as listed below.
3420 135264 dcbzl instruction missing
3422 153699 Valgrind should report unaligned reads with movdqa
3424 190429 Valgrind reports lost of errors in ld.so
3425 with x86_64 2.9.90 glibc
3426 197266 valgrind appears to choke on the xmms instruction
3428 197988 Crash when demangling very large symbol names
3429 202315 unhandled syscall: 332 (inotify_init1)
3430 203256 Add page-level profiling to Massif
3431 205093 dsymutil=yes needs quotes, locking (partial fix)
3432 205241 Snow Leopard 10.6 support (partial fix)
3433 206600 Leak checker fails to upgrade indirect blocks when their
3434 parent becomes reachable
3435 210935 port valgrind.h (not valgrind) to win32 so apps run under
3436 wine can make client requests
3437 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3438 within Linux ip-stack checksum functions
3439 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3441 213685 Undefined value propagates past dependency breaking instruction
3443 215914 Valgrind inserts bogus empty environment variable
3445 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3446 222545 shmat fails under valgind on some arm targets
3447 222560 ARM NEON support
3450 232509 Docs build fails with formatting inside <title></title> elements
3452 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3453 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3454 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3455 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3456 237485 symlink (syscall 57) is not supported on Mac OS
3457 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3459 238208 is_just_below_ESP doesn't take into account red-zone
3460 238345 valgrind passes wrong $0 when executing a shell script
3461 238679 mq_timedreceive syscall doesn't flag the reception buffer
3463 238696 fcntl command F_DUPFD_CLOEXEC not supported
3464 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3465 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3466 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
3468 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
3474 242606 unhandled syscall: setegid (in Ptrcheck)
3475 242814 Helgrind "Impossible has happened" during
3476 QApplication::initInstance();
3477 243064 Valgrind attempting to read debug information from iso
3478 243270 Make stack unwinding in Valgrind wrappers more reliable
3479 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
3480 sysno = 277 (mq_open)
3481 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
3482 244493 ARM VFP d16-d31 registers support
3483 244670 add support for audit_session_self syscall on Mac OS 10.6
3484 244921 The xml report of helgrind tool is not well format
3485 244923 In the xml report file, the <preamble> not escape the
3486 xml char, eg '<','&','>'
3487 245535 print full path names in plain text reports
3488 245925 x86-64 red zone handling problem
3489 246258 Valgrind not catching integer underruns + new [] s
3490 246311 reg/reg cmpxchg doesn't work on amd64
3491 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
3492 246888 Improve Makefile.vex.am
3493 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
3494 to [f]chmod_extended
3495 247526 IBM POWER6 (ISA 2.05) support is incomplete
3496 247561 Some leak testcases fails due to reachable addresses in
3498 247875 sizeofIRType to handle Ity_I128
3499 247894 [PATCH] unhandled syscall sys_readahead
3500 247980 Doesn't honor CFLAGS passed to configure
3501 248373 darwin10.supp is empty in the trunk
3502 248822 Linux FIBMAP ioctl has int parameter instead of long
3503 248893 [PATCH] make readdwarf.c big endianess safe to enable
3504 unwinding on big endian systems
3505 249224 Syscall 336 not supported (SYS_proc_info)
3507 249775 Incorrect scheme for detecting NEON capabilities of host CPU
3508 249943 jni JVM init fails when using valgrind
3509 249991 Valgrind incorrectly declares AESKEYGENASSIST support
3511 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
3512 250799 frexp$fenv_access_off function generates SIGILL
3513 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
3514 251251 support pclmulqdq insn
3515 251362 valgrind: ARM: attach to debugger either fails or provokes
3517 251674 Unhandled syscall 294
3520 254257 Add support for debugfiles found by build-id
3521 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
3522 254646 Wrapped functions cause stack misalignment on OS X
3523 (and possibly Linux)
3524 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
3526 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
3530 Release 3.5.0 (19 August 2009)
3531 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3532 3.5.0 is a feature release with many significant improvements and the
3533 usual collection of bug fixes. The main improvement is that Valgrind
3534 now works on Mac OS X.
3536 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
3537 and X86/Darwin. Support for recent distros and toolchain components
3538 (glibc 2.10, gcc 4.5) has been added.
3540 -------------------------
3542 Here is a short summary of the changes. Details are shown further
3545 * Support for Mac OS X (10.5.x).
3547 * Improvements and simplifications to Memcheck's leak checker.
3549 * Clarification and simplifications in various aspects of Valgrind's
3552 * XML output for Helgrind and Ptrcheck.
3554 * Performance and stability improvements for Helgrind and DRD.
3556 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
3558 * A new experimental tool, BBV, useful for computer architecture
3561 * Improved Wine support, including ability to read Windows PDB
3564 -------------------------
3566 Here are details of the above changes, followed by descriptions of
3567 many other minor changes, and a list of fixed bugs.
3570 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
3571 called "Darwin" because that is the name of the OS core, which is the
3572 level that Valgrind works at.)
3576 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
3577 because it would require work and 10.4 is only becoming less common.
3579 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
3580 fairly well. For 10.5.x, 32-bit programs are the default even on
3581 64-bit machines, so it handles most current programs.
3583 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
3584 officially supported, but simple programs at least will probably work.
3585 However, start-up is slow.
3587 - PowerPC machines are not supported.
3589 Things that don't work:
3591 - The Ptrcheck tool.
3593 - Objective-C garbage collection.
3597 - If you have Rogue Amoeba's "Instant Hijack" program installed,
3598 Valgrind will fail with a SIGTRAP at start-up. See
3599 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
3604 - You will likely find --dsymutil=yes a useful option, as error
3605 messages may be imprecise without it.
3607 - Mac OS X support is new and therefore will be less robust than the
3608 Linux support. Please report any bugs you find.
3610 - Threaded programs may run more slowly than on Linux.
3612 Many thanks to Greg Parker for developing this port over several years.
3615 * Memcheck's leak checker has been improved.
3617 - The results for --leak-check=summary now match the summary results
3618 for --leak-check=full. Previously they could differ because
3619 --leak-check=summary counted "indirectly lost" blocks and
3620 "suppressed" blocks as "definitely lost".
3622 - Blocks that are only reachable via at least one interior-pointer,
3623 but are directly pointed to by a start-pointer, were previously
3624 marked as "still reachable". They are now correctly marked as
3627 - The default value for the --leak-resolution option has been
3628 changed from "low" to "high". In general, this means that more
3629 leak reports will be produced, but each leak report will describe
3630 fewer leaked blocks.
3632 - With --leak-check=full, "definitely lost" and "possibly lost"
3633 leaks are now considered as proper errors, ie. they are counted
3634 for the "ERROR SUMMARY" and affect the behaviour of
3635 --error-exitcode. These leaks are not counted as errors if
3636 --leak-check=summary is specified, however.
3638 - Documentation for the leak checker has been improved.
3641 * Various aspects of Valgrind's text output have changed.
3643 - Valgrind's start-up message has changed. It is shorter but also
3644 includes the command being run, which makes it easier to use
3645 --trace-children=yes. An example:
3647 - Valgrind's shut-down messages have also changed. This is most
3648 noticeable with Memcheck, where the leak summary now occurs before
3649 the error summary. This change was necessary to allow leaks to be
3650 counted as proper errors (see the description of the leak checker
3651 changes above for more details). This was also necessary to fix a
3652 longstanding bug in which uses of suppressions against leaks were
3653 not "counted", leading to difficulties in maintaining suppression
3654 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3656 - Behavior of -v has changed. In previous versions, -v printed out
3657 a mixture of marginally-user-useful information, and tool/core
3658 statistics. The statistics printing has now been moved to its own
3659 flag, --stats=yes. This means -v is less verbose and more likely
3660 to convey useful end-user information.
3662 - The format of some (non-XML) stack trace entries has changed a
3663 little. Previously there were six possible forms:
3665 0x80483BF: really (a.c:20)
3666 0x80483BF: really (in /foo/a.out)
3668 0x80483BF: (within /foo/a.out)
3669 0x80483BF: ??? (a.c:20)
3672 The third and fourth of these forms have been made more consistent
3673 with the others. The six possible forms are now:
3675 0x80483BF: really (a.c:20)
3676 0x80483BF: really (in /foo/a.out)
3677 0x80483BF: really (in ???)
3678 0x80483BF: ??? (in /foo/a.out)
3679 0x80483BF: ??? (a.c:20)
3682 Stack traces produced when --xml=yes is specified are different
3686 * Helgrind and Ptrcheck now support XML output, so they can be used
3687 from GUI tools. Also, the XML output mechanism has been
3690 - The XML format has been overhauled and generalised, so it is more
3691 suitable for error reporting tools in general. The Memcheck
3692 specific aspects of it have been removed. The new format, which
3693 is an evolution of the old format, is described in
3694 docs/internals/xml-output-protocol4.txt.
3696 - Memcheck has been updated to use the new format.
3698 - Helgrind and Ptrcheck are now able to emit output in this format.
3700 - The XML output mechanism has been overhauled. XML is now output
3701 to its own file descriptor, which means that:
3703 * Valgrind can output text and XML independently.
3705 * The longstanding problem of XML output being corrupted by
3706 unexpected un-tagged text messages is solved.
3708 As before, the destination for text output is specified using
3709 --log-file=, --log-fd= or --log-socket=.
3711 As before, XML output for a tool is enabled using --xml=yes.
3713 Because there's a new XML output channel, the XML output
3714 destination is now specified by --xml-file=, --xml-fd= or
3717 Initial feedback has shown this causes some confusion. To
3718 clarify, the two envisaged usage scenarios are:
3720 (1) Normal text output. In this case, do not specify --xml=yes
3721 nor any of --xml-file=, --xml-fd= or --xml-socket=.
3723 (2) XML output. In this case, specify --xml=yes, and one of
3724 --xml-file=, --xml-fd= or --xml-socket= to select the XML
3725 destination, one of --log-file=, --log-fd= or --log-socket=
3726 to select the destination for any remaining text messages,
3727 and, importantly, -q.
3729 -q makes Valgrind completely silent on the text channel,
3730 except in the case of critical failures, such as Valgrind
3731 itself segfaulting, or failing to read debugging information.
3732 Hence, in this scenario, it suffices to check whether or not
3733 any output appeared on the text channel. If yes, then it is
3734 likely to be a critical error which should be brought to the
3735 attention of the user. If no (the text channel produced no
3736 output) then it can be assumed that the run was successful.
3738 This allows GUIs to make the critical distinction they need to
3739 make (did the run fail or not?) without having to search or
3740 filter the text output channel in any way.
3742 It is also recommended to use --child-silent-after-fork=yes in
3746 * Improvements and changes in Helgrind:
3748 - XML output, as described above
3750 - Checks for consistent association between pthread condition
3751 variables and their associated mutexes are now performed.
3753 - pthread_spinlock functions are supported.
3755 - Modest performance improvements.
3757 - Initial (skeletal) support for describing the behaviour of
3758 non-POSIX synchronisation objects through ThreadSanitizer
3759 compatible ANNOTATE_* macros.
3761 - More controllable tradeoffs between performance and the level of
3762 detail of "previous" accesses in a race. There are now three
3765 * --history-level=full. This is the default, and was also the
3766 default in 3.4.x. It shows both stacks involved in a race, but
3767 requires a lot of memory and can be very slow in programs that
3768 do many inter-thread synchronisation events.
3770 * --history-level=none. This only shows the later stack involved
3771 in a race. This can be much faster than --history-level=full,
3772 but makes it much more difficult to find the other access
3773 involved in the race.
3775 The new intermediate setting is
3777 * --history-level=approx
3779 For the earlier (other) access, two stacks are presented. The
3780 earlier access is guaranteed to be somewhere in between the two
3781 program points denoted by those stacks. This is not as useful
3782 as showing the exact stack for the previous access (as per
3783 --history-level=full), but it is better than nothing, and it's
3784 almost as fast as --history-level=none.
3787 * New features and improvements in DRD:
3789 - The error messages printed by DRD are now easier to interpret.
3790 Instead of using two different numbers to identify each thread
3791 (Valgrind thread ID and DRD thread ID), DRD does now identify
3792 threads via a single number (the DRD thread ID). Furthermore
3793 "first observed at" information is now printed for all error
3794 messages related to synchronization objects.
3796 - Added support for named semaphores (sem_open() and sem_close()).
3798 - Race conditions between pthread_barrier_wait() and
3799 pthread_barrier_destroy() calls are now reported.
3801 - Added support for custom allocators through the macros
3802 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
3803 in <valgrind/valgrind.h>). An alternative for these two macros is
3804 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
3807 - Added support for annotating non-POSIX synchronization objects
3808 through several new ANNOTATE_*() macros.
3810 - OpenMP: added support for the OpenMP runtime (libgomp) included
3811 with gcc versions 4.3.0 and 4.4.0.
3815 - Added two new command-line options (--first-race-only and
3816 --segment-merging-interval).
3819 * Genuinely atomic support for x86/amd64/ppc atomic instructions
3821 Valgrind will now preserve (memory-access) atomicity of LOCK-
3822 prefixed x86/amd64 instructions, and any others implying a global
3823 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
3825 This means that Valgrinded processes will "play nicely" in
3826 situations where communication with other processes, or the kernel,
3827 is done through shared memory and coordinated with such atomic
3828 instructions. Prior to this change, such arrangements usually
3829 resulted in hangs, races or other synchronisation failures, because
3830 Valgrind did not honour atomicity of such instructions.
3833 * A new experimental tool, BBV, has been added. BBV generates basic
3834 block vectors for use with the SimPoint analysis tool, which allows
3835 a program's overall behaviour to be approximated by running only a
3836 fraction of it. This is useful for computer architecture
3837 researchers. You can run BBV by specifying --tool=exp-bbv (the
3838 "exp-" prefix is short for "experimental"). BBV was written by
3842 * Modestly improved support for running Windows applications under
3843 Wine. In particular, initial support for reading Windows .PDB debug
3844 information has been added.
3847 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
3848 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
3852 * The Valgrind client requests VALGRIND_PRINTF and
3853 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
3854 the string was always printed immediately on its own line. Now, the
3855 string will be added to a buffer but not printed until a newline is
3856 encountered, or other Valgrind output is printed (note that for
3857 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
3858 "other Valgrind output"). This allows you to use multiple
3859 VALGRIND_PRINTF calls to build up a single output line, and also to
3860 print multiple output lines with a single request (by embedding
3861 multiple newlines in the string).
3864 * The graphs drawn by Massif's ms_print program have changed slightly:
3866 - The half-height chars '.' and ',' are no longer drawn, because
3867 they are confusing. The --y option can be used if the default
3868 y-resolution is not high enough.
3870 - Horizontal lines are now drawn after the top of a snapshot if
3871 there is a gap until the next snapshot. This makes it clear that
3872 the memory usage has not dropped to zero between snapshots.
3875 * Something that happened in 3.4.0, but wasn't clearly announced: the
3876 option --read-var-info=yes can be used by some tools (Memcheck,
3877 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
3878 variable type and location information. This makes those tools
3879 start up more slowly and increases memory consumption, but
3880 descriptions of data addresses in error messages become more
3884 * exp-Omega, an experimental instantaneous leak-detecting tool, was
3885 disabled in 3.4.0 due to a lack of interest and maintenance,
3886 although the source code was still in the distribution. The source
3887 code has now been removed from the distribution. For anyone
3888 interested, the removal occurred in SVN revision r10247.
3891 * Some changes have been made to the build system.
3893 - VEX/ is now integrated properly into the build system. This means
3894 that dependency tracking within VEX/ now works properly, "make
3895 install" will work without requiring "make" before it, and
3896 parallel builds (ie. 'make -j') now work (previously a
3897 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
3898 was effectively ignored).
3900 - The --with-vex configure option has been removed. It was of
3901 little use and removing it simplified the build system.
3903 - The location of some install files has changed. This should not
3904 affect most users. Those who might be affected:
3906 * For people who use Valgrind with MPI programs, the installed
3907 libmpiwrap.so library has moved from
3908 $(INSTALL)/<platform>/libmpiwrap.so to
3909 $(INSTALL)/libmpiwrap-<platform>.so.
3911 * For people who distribute standalone Valgrind tools, the
3912 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
3913 have moved to $(INSTALL)/libcoregrind-<platform>.a.
3915 These changes simplify the build system.
3917 - Previously, all the distributed suppression (*.supp) files were
3918 installed. Now, only default.supp is installed. This should not
3919 affect users as the other installed suppression files were not
3920 read; the fact that they were installed was a mistake.
3923 * KNOWN LIMITATIONS:
3925 - Memcheck is unusable with the Intel compiler suite version 11.1,
3926 when it generates code for SSE2-and-above capable targets. This
3927 is because of icc's use of highly optimised inlined strlen
3928 implementations. It causes Memcheck to report huge numbers of
3929 false errors even in simple programs. Helgrind and DRD may also
3932 Versions 11.0 and earlier may be OK, but this has not been
3936 The following bugs have been fixed or resolved. Note that "n-i-bz"
3937 stands for "not in bugzilla" -- that is, a bug that was reported to us
3938 but never got a bugzilla entry. We encourage you to file bugs in
3939 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3940 mailing the developers (or mailing lists) directly -- bugs that are
3941 not entered into bugzilla tend to get forgotten about or ignored.
3943 To see details of a given bug, visit
3944 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3945 where XXXXXX is the bug number as listed below.
3947 84303 How about a LockCheck tool?
3948 91633 dereference of null ptr in vgPlain_st_basetype
3949 97452 Valgrind doesn't report any pthreads problems
3950 100628 leak-check gets assertion failure when using
3951 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
3952 108528 NPTL pthread cleanup handlers not called
3953 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
3954 110128 mallinfo is not implemented...
3955 110770 VEX: Generated files not always updated when making valgrind
3956 111102 Memcheck: problems with large (memory footprint) applications
3957 115673 Vex's decoder should never assert
3958 117564 False positive: Syscall param clone(child_tidptr) contains
3959 uninitialised byte(s)
3960 119404 executing ssh from inside valgrind fails
3961 133679 Callgrind does not write path names to sources with dwarf debug
3963 135847 configure.in problem with non gnu compilers (and possible fix)
3964 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
3965 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
3966 136230 memcheck reports "possibly lost", should be "still reachable"
3967 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
3968 137904 Valgrind reports a memory leak when using POSIX threads,
3970 139076 valgrind VT_GETSTATE error
3971 142228 complaint of elf_dynamic_do_rela in trivial usage
3972 145347 spurious warning with USBDEVFS_REAPURB
3973 148441 (wine) can't find memory leak in Wine, win32 binary
3975 148742 Leak-check fails assert on exit
3976 149878 add (proper) check for calloc integer overflow
3977 150606 Call graph is broken when using callgrind control
3978 152393 leak errors produce an exit code of 0. I need some way to
3979 cause leak errors to result in a nonzero exit code.
3980 157154 documentation (leak-resolution doc speaks about num-callers
3981 def=4) + what is a loss record
3982 159501 incorrect handling of ALSA ioctls
3983 162020 Valgrinding an empty/zero-byte file crashes valgrind
3984 162482 ppc: Valgrind crashes while reading stabs information
3985 162718 x86: avoid segment selector 0 in sys_set_thread_area()
3986 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
3987 163560 VEX/test_main.c is missing from valgrind-3.3.1
3988 164353 malloc_usable_size() doesn't return a usable size
3989 165468 Inconsistent formatting in memcheck manual -- please fix
3990 169505 main.c:286 (endOfInstr):
3991 Assertion 'ii->cost_offset == *cost_offset' failed
3992 177206 Generate default.supp during compile instead of configure
3993 177209 Configure valt_load_address based on arch+os
3994 177305 eventfd / syscall 323 patch lost
3995 179731 Tests fail to build because of inlining of non-local asm labels
3996 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
3997 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
3998 181594 Bogus warning for empty text segment
3999 181707 dwarf doesn't require enumerations to have name
4000 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4001 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4002 Assertion '!already_present' failed.
4003 185359 exp-ptrcheck: unhandled syscall getresuid()
4004 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4005 185816 Valgrind is unable to handle debug info for files with split
4006 debug info that are prelinked afterwards
4007 185980 [darwin] unhandled syscall: sem_open
4008 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4009 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4010 186790 Suppression pattern used for leaks are not reported
4011 186796 Symbols with length>200 in suppression files are ignored
4012 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4013 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4014 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4015 188046 bashisms in the configure script
4016 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4017 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4018 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4019 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4020 assertion fail "!lock->heldBy"
4021 188427 Add support for epoll_create1 (with patch)
4022 188530 Support for SIOCGSTAMPNS
4023 188560 Include valgrind.spec in the tarball
4024 188572 Valgrind on Mac should suppress setenv() mem leak
4025 189054 Valgrind fails to build because of duplicate non-local asm labels
4026 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4027 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4028 189763 drd assertion failure: s_threadinfo[tid].is_recording
4029 190219 unhandled syscall: 328 (x86-linux)
4030 190391 dup of 181394; see above
4031 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4032 190820 No debug information on powerpc-linux
4033 191095 PATCH: Improve usbdevfs ioctl handling
4034 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4036 191189 --xml=yes should obey --gen-suppressions=all
4037 191192 syslog() needs a suppression on macosx
4038 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4039 191761 getrlimit on MacOSX
4040 191992 multiple --fn-skip only works sometimes; dependent on order
4041 192634 V. reports "aspacem sync_check_mapping_callback:
4042 segment mismatch" on Darwin
4043 192954 __extension__ missing on 2 client requests
4044 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4045 194474 "INSTALL" file has different build instructions than "README"
4046 194671 Unhandled syscall (sem_wait?) from mac valgrind
4047 195069 memcheck: reports leak (memory still reachable) for
4049 195169 drd: (vgDrd_barrier_post_wait):
4050 Assertion 'r->sg[p->post_iteration]' failed.
4051 195268 valgrind --log-file doesn't accept ~/...
4052 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4053 195860 WARNING: unhandled syscall: unix:223
4054 196528 need a error suppression for pthread_rwlock_init under os x?
4055 197227 Support aio_* syscalls on Darwin
4056 197456 valgrind should reject --suppressions=(directory)
4057 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4058 197591 unhandled syscall 27 (mincore)
4059 197793 Merge DCAS branch to the trunk == 85756, 142103
4060 197794 Avoid duplicate filenames in Vex
4061 197898 make check fails on current SVN
4062 197901 make check fails also under exp-ptrcheck in current SVN
4063 197929 Make --leak-resolution=high the default
4064 197930 Reduce spacing between leak reports
4065 197933 Print command line of client at start-up, and shorten preamble
4066 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4067 198395 add BBV to the distribution as an experimental tool
4068 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4069 198649 callgrind_annotate doesn't cumulate counters
4070 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4071 199977 Valgrind complains about an unrecognized instruction in the
4072 atomic_incs test program
4073 200029 valgrind isn't able to read Fedora 12 debuginfo
4074 200760 darwin unhandled syscall: unix:284
4075 200827 DRD doesn't work on Mac OS X
4076 200990 VG_(read_millisecond_timer)() does not work correctly
4077 201016 Valgrind does not support pthread_kill() on Mac OS
4078 201169 Document --read-var-info
4079 201323 Pre-3.5.0 performance sanity checking
4080 201384 Review user manual for the 3.5.0 release
4081 201585 mfpvr not implemented on ppc
4082 201708 tests failing because x86 direction flag is left set
4083 201757 Valgrind doesn't handle any recent sys_futex additions
4084 204377 64-bit valgrind can not start a shell script
4085 (with #!/path/to/shell) if the shell is a 32-bit executable
4086 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4087 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4088 about memory allocation events with memory access tracing enabled
4089 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4091 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4095 Release 3.4.1 (28 February 2009)
4096 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4097 3.4.1 is a bug-fix release that fixes some regressions and assertion
4098 failures in debug info reading in 3.4.0, most notably incorrect stack
4099 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4100 other debug info problems are also fixed. A number of bugs in the
4101 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4103 In view of the fact that 3.4.0 contains user-visible regressions
4104 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4105 encouraged to ship 3.4.1 in preference to 3.4.0.
4107 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4108 bugzilla" -- that is, a bug that was reported to us but never got a
4109 bugzilla entry. We encourage you to file bugs in bugzilla
4110 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4111 developers (or mailing lists) directly -- bugs that are not entered
4112 into bugzilla tend to get forgotten about or ignored.
4114 n-i-bz Fix various bugs reading icc-11 generated debug info
4115 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4116 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4117 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4118 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4119 179618 exp-ptrcheck crashed / exit prematurely
4120 179624 helgrind: false positive races with pthread_create and
4121 recv/open/close/read
4122 134207 pkg-config output contains @VG_PLATFORM@
4123 176926 floating point exception at valgrind startup with PPC 440EPX
4124 181594 Bogus warning for empty text segment
4125 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4126 181707 Dwarf3 doesn't require enumerations to have name
4127 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4128 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4129 Assertion '!already_present' failed.
4130 185359 exp-ptrcheck unhandled syscall getresuid()
4132 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4133 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4137 Release 3.4.0 (2 January 2009)
4138 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4139 3.4.0 is a feature release with many significant improvements and the
4140 usual collection of bug fixes. This release supports X86/Linux,
4141 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4142 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4144 3.4.0 brings some significant tool improvements. Memcheck can now
4145 report the origin of uninitialised values, the thread checkers
4146 Helgrind and DRD are much improved, and we have a new experimental
4147 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4148 global arrays. In detail:
4150 * Memcheck is now able to track the origin of uninitialised values.
4151 When it reports an uninitialised value error, it will try to show
4152 the origin of the value, as either a heap or stack allocation.
4153 Origin tracking is expensive and so is not enabled by default. To
4154 use it, specify --track-origins=yes. Memcheck's speed will be
4155 essentially halved, and memory usage will be significantly
4156 increased. Nevertheless it can drastically reduce the effort
4157 required to identify the root cause of uninitialised value errors,
4158 and so is often a programmer productivity win, despite running more
4161 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4162 3.4.0, will be released shortly.
4164 * Helgrind's race detection algorithm has been completely redesigned
4165 and reimplemented, to address usability and scalability concerns:
4167 - The new algorithm has a lower false-error rate: it is much less
4168 likely to report races that do not really exist.
4170 - Helgrind will display full call stacks for both accesses involved
4171 in a race. This makes it easier to identify the root causes of
4174 - Limitations on the size of program that can run have been removed.
4176 - Performance has been modestly improved, although that is very
4179 - Direct support for Qt4 threading has been added.
4181 - pthread_barriers are now directly supported.
4183 - Helgrind works well on all supported Linux targets.
4185 * The DRD thread debugging tool has seen major improvements:
4187 - Greatly improved performance and significantly reduced memory
4190 - Support for several major threading libraries (Boost.Thread, Qt4,
4191 glib, OpenMP) has been added.
4193 - Support for atomic instructions, POSIX semaphores, barriers and
4194 reader-writer locks has been added.
4196 - Works now on PowerPC CPUs too.
4198 - Added support for printing thread stack usage at thread exit time.
4200 - Added support for debugging lock contention.
4202 - Added a manual for Drd.
4204 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4205 checks for misuses of pointers. In that sense it is a bit like
4206 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4207 detect overruns of stack and global arrays, it can detect
4208 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4209 detect accesses heap blocks that have been freed a very long time
4210 ago (millions of blocks in the past).
4212 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4213 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4214 of the main Valgrind documentation. As this is an experimental
4215 tool, we would be particularly interested in hearing about your
4216 experiences with it.
4218 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4219 longer built by default, although the code remains in the repository
4220 and the tarball. This is due to three factors: a perceived lack of
4221 users, a lack of maintenance, and concerns that it may not be
4222 possible to achieve reliable operation using the existing design.
4224 * As usual, support for the latest Linux distros and toolchain
4225 components has been added. It should work well on Fedora Core 10,
4226 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4227 state) is supported, as is glibc-2.9. The C++ demangler has been
4228 updated so as to work well with C++ compiled by even the most recent
4231 * You can now use frame-level wildcards in suppressions. This was a
4232 frequently-requested enhancement. A line "..." in a suppression now
4233 matches zero or more frames. This makes it easier to write
4234 suppressions which are precise yet insensitive to changes in
4237 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4239 * Very basic support for IBM Power6 has been added (64-bit processes only).
4241 * Valgrind is now cross-compilable. For example, it is possible to
4242 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4243 on a ppc32/64-linux target.
4245 * You can set the main thread's stack size at startup using the
4246 new --main-stacksize= flag (subject of course to ulimit settings).
4247 This is useful for running apps that need a lot of stack space.
4249 * The limitation that you can't use --trace-children=yes together
4250 with --db-attach=yes has been removed.
4252 * The following bugs have been fixed. Note that "n-i-bz" stands for
4253 "not in bugzilla" -- that is, a bug that was reported to us but
4254 never got a bugzilla entry. We encourage you to file bugs in
4255 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4256 mailing the developers (or mailing lists) directly.
4258 n-i-bz Make return types for some client requests 64-bit clean
4259 n-i-bz glibc 2.9 support
4260 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4261 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4262 n-i-bz Building in an env without gdb gives bogus gdb attach
4263 92456 Tracing the origin of uninitialised memory
4264 106497 Valgrind does not demangle some C++ template symbols
4266 151612 Suppression with "..." (frame-level wildcards in .supp files)
4267 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4268 159285 unhandled syscall:25 (stime, on x86-linux)
4269 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4270 160954 ppc build of valgrind crashes with illegal instruction (isel)
4271 160956 mallinfo implementation, w/ patch
4272 162092 Valgrind fails to start gnome-system-monitor
4273 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4274 163794 assertion failure with "--track-origins=yes"
4275 163933 sigcontext.err and .trapno must be set together
4276 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4277 164476 Missing kernel module loading system calls
4278 164669 SVN regression: mmap() drops posix file locks
4279 166581 Callgrind output corruption when program forks
4280 167288 Patch file for missing system calls on Cell BE
4281 168943 unsupported scas instruction pentium
4282 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4283 172417 x86->IR: 0x82 ...
4284 172563 amd64->IR: 0xD9 0xF5 - fprem1
4285 173099 .lds linker script generation error
4286 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4287 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4289 174908 --log-file value not expanded correctly for core file
4290 175044 Add lookup_dcookie for amd64
4291 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4293 Developer-visible changes:
4295 * Valgrind's debug-info reading machinery has been majorly overhauled.
4296 It can now correctly establish the addresses for ELF data symbols,
4297 which is something that has never worked properly before now.
4299 Also, Valgrind can now read DWARF3 type and location information for
4300 stack and global variables. This makes it possible to use the
4301 framework to build tools that rely on knowing the type and locations
4302 of stack and global variables, for example exp-Ptrcheck.
4304 Reading of such information is disabled by default, because most
4305 tools don't need it, and because it is expensive in space and time.
4306 However, you can force Valgrind to read it, using the
4307 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4308 make use of such information, if present, to provide source-level
4309 descriptions of data addresses in the error messages they create.
4311 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4312 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).