1 Release 3.17.0 (18 Mar 2021)
2 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4 3.17.0 fixes a number of bugs and adds some functional changes: support for
5 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
6 some new instructions for Arm64, S390 and POWER. There are also some tool
9 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
10 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
11 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
12 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
13 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
15 * ==================== CORE CHANGES ===================
17 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
20 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
21 debugging information. When a debuginfo file cannot be found locally,
22 Valgrind is able to query debuginfod servers for the file using its
23 build-id. See the user manual for more information about debuginfod support.
25 * ================== PLATFORM CHANGES =================
29 - Inaccuracies resulting from double-rounding in the simulation of
30 floating-point multiply-add/subtract instructions have been fixed. These
31 should now behave exactly as the hardware does.
33 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
34 ongoing. Support for the half-word variants of at least the following
35 instructions has been added:
37 FABS <Vd>.<T>, <Vn>.<T>
39 FNEG <Vd>.<T>, <Vn>.<T>
41 FSQRT <Vd>.<T>, <Vn>.<T>
46 - Implement the new instructions/features that were added to z/Architecture
47 with the vector-enhancements facility 1. Also cover the instructions from
48 the vector-packed-decimal facility that are defined outside the chapter
49 "Vector Decimal Instructions", but not the ones from that chapter itself.
51 For a detailed list of newly supported instructions see the updates to
52 `docs/internals/s390-opcodes.csv'.
54 Since the miscellaneous instruction extensions facility 2 was already
55 added in Valgrind 3.16.0, this completes the support necessary to run
56 general programs built with `--march=z14' under Valgrind. The
57 vector-packed-decimal facility is currently not exploited by the standard
58 toolchain and libraries.
62 - Various bug fixes. Fix for the sync field to limit setting just two of
63 the two bits in the L-field. Fix the write size for the stxsibx and
64 stxsihx instructions. Fix the modsw and modsd instructions.
66 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
67 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
68 conversion instructions are still missing.
70 * ==================== TOOL CHANGES ====================
72 * General tool changes
74 - All the tools and their vgpreload libraries are now installed under
75 libexec because they cannot be executed directly and should be run through
76 the valgrind executable. This should be an internal, not user visible,
77 change, but might impact valgrind packagers.
79 - The --track-fds option now respects -q, --quiet and won't output anything
80 if no file descriptors are leaked. It also won't report the standard stdin
81 (0), stdout (1) or stderr (2) descriptors as being leaked with
82 --trace-fds=yes anymore. To track whether the standard file descriptors
83 are still open at the end of the program run use --trace-fds=all.
87 - DHAT has been extended, with two new modes of operation. The new
88 --mode=copy flag triggers copy profiling, which records calls to memcpy,
89 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
90 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
91 the new dhat/dhat.h file. This is useful for learning more about hot code
92 paths. See the user manual for more information about the new modes.
94 - Because of these changes, DHAT's file format has changed. DHAT output
95 files produced with earlier versions of DHAT will not work with this
96 version of DHAT's viewer, and DHAT output files produced with this version
97 of DHAT will not work with earlier versions of DHAT's viewer.
107 * ==================== OTHER CHANGES ====================
109 * ==================== FIXED BUGS ====================
111 The following bugs have been fixed or resolved. Note that "n-i-bz"
112 stands for "not in bugzilla" -- that is, a bug that was reported to us
113 but never got a bugzilla entry. We encourage you to file bugs in
114 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
115 than mailing the developers (or mailing lists) directly -- bugs that
116 are not entered into bugzilla tend to get forgotten about or ignored.
118 To see details of a given bug, visit
119 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
120 where XXXXXX is the bug number as listed below.
122 140178 open("/proc/self/exe", ...); doesn't quite work
123 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
124 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
125 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
126 345077 linux syscall execveat support (linux 3.19)
127 361770 Missing F_ADD_SEALS
128 369029 handle linux syscalls sched_getattr and sched_setattr
129 384729 __libc_freeres inhibits cross-platform valgrind
130 388787 Support for C++17 new/delete
131 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
132 396656 Warnings while reading debug info
133 397605 ioctl FICLONE mishandled
134 401416 Compile failure with openmpi 4.0
135 408663 Suppression file for musl libc
136 404076 s390x: z14 vector instructions not implemented
137 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
138 (actually affected all x86 and nanomips regardless of host bitness)
139 413547 regression test does not check for Arm 64 features.
140 414268 Enable AArch64 feature detection and decoding for v8.x instructions
141 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
142 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
143 422261 platform selection fails for unqualified client name
144 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
145 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
146 423195 PPC ISA 3.1 support is missing, part 1
147 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
148 424012 crash with readv/writev having invalid but not NULL arg2 iovec
149 424298 amd64: Implement RDSEED
150 425232 PPC ISA 3.1 support is missing, part 2
151 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
152 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
153 426123 PPC ISA 3.1 support is missing, part 3
154 426144 Fix "condition variable has not been initialized" on Fedora 33.
155 427400 PPC ISA 3.1 support is missing, part 4
156 427401 PPC ISA 3.1 support is missing, part 5
157 427404 PPC ISA 3.1 support is missing, part 6
158 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
159 427787 Support new faccessat2 linux syscall (439)
160 427969 debuginfo section duplicates a section in the main ELF file
161 428035 drd: Unbreak the musl build
162 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
163 428716 cppcheck detects potential leak in VEX/useful/smchash.c
164 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
165 429352 PPC ISA 3.1 support is missing, part 7
166 429354 PPC ISA 3.1 support is missing, part 8
167 429692 unhandled ppc64le-linux syscall: 147 (getsid)
168 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
170 429952 Errors when building regtest with clang
171 430354 ppc stxsibx and stxsihx instructions write too much data
172 430429 valgrind.h doesn't compile on s390x with clang
173 430485 expr_is_guardable doesn't handle Iex_Qop
174 431556 Complete arm64 FADDP v8.2 instruction support
175 432102 Add support for DWARF5 as produced by GCC11
176 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
177 432381 drd: Process STACK_REGISTER client requests
178 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
179 432672 vg_regtest: test-specific environment variables not reset between tests
180 432809 VEX should support REX.W + POPF
181 432861 PPC modsw and modsd give incorrect results for 1 mod 12
182 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
183 432215 Add debuginfod functionality
184 433323 Use pkglibexecdir as vglibdir
185 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
186 433629 valgrind/README has type "abd" instead of "and"
187 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
188 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
189 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
190 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
191 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
193 (3.17.0.RC1: 13 Mar 2021)
194 (3.17.0.RC2: 17 Mar 2021)
198 Release 3.16.1 (22 June 2020)
199 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
201 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
202 fixes character encoding problems in the documentation HTML.
204 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
205 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
207 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
211 Release 3.16.0 (27 May 2020)
212 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
214 3.16.0 is a feature release with many improvements and the usual collection of
217 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
218 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
219 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
220 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
221 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
223 * ==================== CORE CHANGES ===================
225 * It is now possible to dynamically change the value of many command line
226 options while your program (or its children) are running under Valgrind.
228 To see the list of dynamically changeable options, run
229 "valgrind --help-dyn-options".
231 You can change the options from the shell by using vgdb to launch
232 the monitor command "v.clo <clo option>...".
233 The same monitor command can be used from a gdb connected
234 to the valgrind gdbserver.
235 Your program can also change the dynamically changeable options using
236 the client request VALGRIND_CLO_CHANGE(option).
238 * ================== PLATFORM CHANGES =================
240 * MIPS: preliminary support for nanoMIPS instruction set has been added.
242 * ==================== TOOL CHANGES ====================
246 - The implicit memcpy done by each call to realloc now counts towards the
247 read and write counts of resized heap blocks, making those counts higher
252 - cg_annotate's --auto and --show-percs options now default to 'yes', because
253 they are usually wanted.
257 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
258 because they are usually wanted.
260 - The command option --collect-systime has been enhanced to specify
261 the unit used to record the elapsed time spent during system calls.
262 The command option now accepts the values no|yes|msec|usec|nsec,
263 where yes is a synonym of msec. When giving the value nsec, the
264 system cpu time of system calls is also recorded.
268 - Several memcheck options are now dynamically changeable.
269 Use valgrind --help-dyn-options to list them.
271 - The release 3.15 introduced a backward incompatible change for
272 some suppression entries related to preadv and pwritev syscalls.
273 When reading a suppression entry using the unsupported 3.14 format,
274 valgrind will now produce a warning to say the suppression entry will not
275 work, and suggest the needed change.
277 - Significantly fewer false positive errors on optimised code generated by
278 Clang and GCC. In particular, Memcheck now deals better with the
279 situation where the compiler will transform C-level "A && B" into "B && A"
280 under certain circumstances (in which the transformation is valid).
281 Handling of integer equality/non-equality checks on partially defined
282 values is also improved on some architectures.
286 - The exprimental Stack and Global Array Checking tool has been removed.
287 It only ever worked on x86 and amd64, and even on those it had a
288 high false positive rate and was slow. An alternative for detecting
289 stack and global array overruns is using the AddressSanitizer (ASAN)
290 facility of the GCC and Clang compilers, which require you to rebuild
291 your code with -fsanitize=address.
293 * ==================== OTHER CHANGES ====================
295 * New and modified GDB server monitor features:
297 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
299 - The gdbserver monitor commands that require an address and an optional
300 length argument now accepts the alternate 'C like' syntax "address[length]".
301 For example, the memcheck command "monitor who_points_at 0x12345678 120"
302 can now also be given as "monitor who_points_at 0x12345678[120]".
304 * ==================== FIXED BUGS ====================
306 The following bugs have been fixed or resolved. Note that "n-i-bz"
307 stands for "not in bugzilla" -- that is, a bug that was reported to us
308 but never got a bugzilla entry. We encourage you to file bugs in
309 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
310 than mailing the developers (or mailing lists) directly -- bugs that
311 are not entered into bugzilla tend to get forgotten about or ignored.
313 To see details of a given bug, visit
314 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
315 where XXXXXX is the bug number as listed below.
317 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
318 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
319 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
320 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
321 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
322 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
323 400872 Add nanoMIPS support to Valgrind
324 403212 drd/tests/trylock hangs on FreeBSD
325 404406 s390x: z14 miscellaneous instructions not implemented
326 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
327 406561 mcinfcallWSRU gdbserver_test fails on ppc64
328 406824 Unsupported baseline
329 407218 Add support for the copy_file_range syscall
330 407307 Intercept stpcpy also in ld.so for arm64
331 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
333 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
334 408009 Expose rdrand and f16c even on avx if host cpu supports them
335 408091 Missing pkey syscalls
336 408414 Add support for missing for preadv2 and pwritev2 syscalls
337 409141 Valgrind hangs when SIGKILLed
338 409206 Support for Linux PPS and PTP ioctls
339 409367 exit_group() after signal to thread waiting in futex() causes hangs
340 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
341 409780 References to non-existent configure.in
342 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
343 410599 Non-deterministic behaviour of pth_self_kill_15_other test
344 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
345 411134 Allow the user to change a set of command line options during execution
346 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
347 412344 Problem setting mips flags with specific paths
348 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
349 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
350 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
351 413603 callgrind_annotate/cg_annotate truncate function names at '#'
352 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
353 415136 ARMv8.1 Compare-and-Swap instructions are not supported
354 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
355 416239 valgrind crashes when handling clock_adjtime
356 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
357 416286 DRD reports "conflicting load" error on std::mutex::lock()
358 416301 s390x: "compare and signal" not supported
359 416387 finit_module and bpf syscalls are unhandled on arm64
360 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
361 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
362 416753 new 32bit time syscalls for 2038+
363 417075 pwritev(vector[...]) suppression ignored
364 417075 is not fixed, but incompatible supp entries are detected
365 and a warning is produced for these.
366 417187 [MIPS] Conditional branch problem since 'grail' changes
367 417238 Test memcheck/tests/vbit-test fails on mips64 BE
368 417266 Make memcheck/tests/linux/sigqueue usable with musl
369 417281 s390x: /bin/true segfaults with "grail" enabled
370 417427 commit to fix vki_siginfo_t definition created numerous regression
372 417452 s390_insn_store_emit: dst->tag for HRcVec128
373 417578 Add suppressions for glibc DTV leaks
374 417906 clone with CLONE_VFORK and no CLONE_VM fails
375 418004 Grail code additions break ppc64.
376 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
377 418997 s390x: Support Iex_ITE for float and vector types
378 419503 s390x: Avoid modifying registers returned from isel functions
379 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
380 421570 std_mutex fails on Arm v8.1 h/w
381 434035 vgdb might crash if valgrind is killed
382 n-i-bz Fix minor one time leaks in dhat.
383 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
384 n-i-bz Add support for the Linux io_uring system calls
385 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
386 n-i-bz Fix non-glibc build of test suite with s390x_features
387 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
388 423195 PPC ISA 3.1 support is missing, part 1
390 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
391 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
392 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
396 Release 3.15.0 (12 April 2019)
397 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
399 3.15.0 is a feature release with many improvements and the usual collection of
402 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
403 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
404 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
405 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
406 support for X86/macOS 10.13 and AMD64/macOS 10.13.
408 * ==================== CORE CHANGES ===================
410 * The XTree Massif output format now makes use of the information obtained
411 when specifying --read-inline-info=yes.
413 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
415 * ==================== TOOL CHANGES ====================
419 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
420 it has been promoted from an experimental tool to a regular tool. Run it
421 with --tool=dhat instead of --tool=exp-dhat.
423 - DHAT now prints only minimal data when the program ends, instead writing
424 the bulk of the profiling data to a file. As a result, the --show-top-n
425 and --sort-by options have been removed.
427 - Profile results can be viewed with the new viewer, dh_view.html. When
428 a run ends, a short message is printed, explaining how to view the result.
430 - See the documentation for more details.
434 - cg_annotate has a new option, --show-percs, which prints percentages next
439 - callgrind_annotate has a new option, --show-percs, which prints percentages
440 next to all event counts.
442 - callgrind_annotate now inserts commas in call counts, and
443 sort the caller/callee lists in the call tree.
447 - The default value for --read-inline-info is now "yes" on
448 Linux/Android/Solaris. It is still "no" on other OS.
452 - The option --xtree-leak=yes (to output leak result in xtree format)
453 automatically activates the option --show-leak-kinds=all, as xtree
454 visualisation tools such as kcachegrind can in any case select what kind
455 of leak to visualise.
457 - There has been further work to avoid false positives. In particular,
458 integer equality on partially defined inputs (C == and !=) is now handled
461 * ==================== OTHER CHANGES ====================
463 * The new option --show-error-list=no|yes displays, at the end of the run, the
464 list of detected errors and the used suppressions. Prior to this change,
465 showing this information could only be done by specifying "-v -v", but that
466 also produced a lot of other possibly-non-useful messages. The option -s is
467 equivalent to --show-error-list=yes.
469 * ==================== FIXED BUGS ====================
471 The following bugs have been fixed or resolved. Note that "n-i-bz"
472 stands for "not in bugzilla" -- that is, a bug that was reported to us
473 but never got a bugzilla entry. We encourage you to file bugs in
474 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
475 than mailing the developers (or mailing lists) directly -- bugs that
476 are not entered into bugzilla tend to get forgotten about or ignored.
478 To see details of a given bug, visit
479 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
480 where XXXXXX is the bug number as listed below.
482 385411 s390x: z13 vector floating-point instructions not implemented
483 397187 z13 vector register support for vgdb gdbserver
484 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
485 398870 Please add support for instruction vcvtps2ph
486 399287 amd64 front end: Illegal Instruction vcmptrueps
487 399301 Use inlined frames in Massif XTree output.
488 399322 Improve callgrind_annotate output
489 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
490 400164 helgrind test encounters mips x-compiler warnings and assembler error
491 400490 s390x: VRs allocated as if separate from FPRs
492 400491 s390x: Operand of LOCH treated as unsigned integer
493 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
494 options, which specify a mips64 processor
495 401112 LLVM 5.0 generates comparison against partially initialized data
496 401277 More bugs in z13 support
497 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
498 401578 drd: crashes sometimes on fork()
499 401627 memcheck errors with glibc avx2 optimized wcsncmp
500 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
501 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
502 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
504 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
505 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
506 402123 invalid assembler opcodes for mips32r2
507 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
508 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
509 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
510 402351 mips64 libvexmultiarch_test fails on s390x
512 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
513 402480 Do not use %rsp in clobber list
514 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
515 402515 Implement new option --show-error-list=no|yes / -s
516 402519 POWER 3.0 addex instruction incorrectly implemented
517 402781 Redo the cache used to process indirect branch targets
518 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
519 403552 s390x: wrong facility bit checked for vector facility
520 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
521 404638 Add VG_(replaceIndexXA)
522 404843 s390x: backtrace sometimes ends prematurely
523 404888 autotools cleanup series
524 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
525 405182 Valgrind fails to build with Clang
526 405205 filter_libc: remove the line holding the futex syscall error entirely
527 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
528 the upper and lower 32-bits of the 64-bit result
529 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
530 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
531 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
533 405403 s390x disassembler cannot be used on x86
534 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
535 405458 MIPS mkFormVEC arguments swapped?
536 405716 drd: Fix an integer overflow in the stack margin calculation
537 405722 Support arm64 core dump
538 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
539 of the 64-bit destination field.
540 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
541 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
542 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
544 406256 PPC64, vector floating point instructions don't handle subnormal
545 according to VSCR[NJ] bit setting.
546 406352 cachegrind/callgrind fails ann tests because of missing a.c
547 406354 dhat is broken on x86 (32bit)
548 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
549 406357 gdbserver_tests fails because of gdb output change
550 406360 memcheck/tests/libstdc++.supp needs more supression variants
551 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
552 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
553 407340 PPC64, does not support the vlogefp, vexptefp instructions.
554 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
555 n-i-bz Fix callgrind_annotate non deterministic order for equal total
556 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
557 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
558 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
560 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
561 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
562 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
566 Release 3.14.0 (9 October 2018)
567 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
569 3.14.0 is a feature release with many improvements and the usual collection of
572 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
573 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
574 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
575 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
576 support for X86/macOS 10.13, AMD64/macOS 10.13.
578 * ==================== CORE CHANGES ===================
580 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
581 debug info for unloaded code. This allows saved stack traces (e.g. for
582 memory leaks) to include file/line info for code that has been dlclose'd (or
583 similar). See the user manual for more information and known limitations.
585 * Ability to specify suppressions based on source file name and line number.
587 * Majorly overhauled register allocator. No end-user changes, but the JIT
588 generates code a bit more quickly now.
590 * ================== PLATFORM CHANGES =================
592 * Preliminary support for macOS 10.13 has been added.
594 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
596 * mips: support for MIPS SIMD architecture (MSA) has been added.
598 * mips: support for MIPS N32 ABI has been added.
600 * s390: partial support for vector instructions (integer and string) has been
603 * ==================== TOOL CHANGES ====================
605 * Helgrind: Addition of a flag
606 --delta-stacktrace=no|yes [yes on linux amd64/x86]
607 which specifies how full history stack traces should be computed.
608 Setting this to =yes can speed up Helgrind by 25% when using
609 --history-level=full.
611 * Memcheck: reduced false positive rate for optimised code created by Clang 6
612 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
613 blocks more carefully to determine where it can avoid expensive definedness
614 checks without loss of precision. This is controlled by the flag
615 --expensive-definedness-checks=no|auto|yes [auto].
617 * ==================== OTHER CHANGES ====================
619 * Valgrind is now buildable with link-time optimisation (LTO). A new
620 configure option --enable-lto=yes allows building Valgrind with LTO. If the
621 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
622 Note that if you are doing Valgrind development, --enable-lto=yes massively
623 slows down the build process.
625 * ==================== FIXED BUGS ====================
627 The following bugs have been fixed or resolved. Note that "n-i-bz"
628 stands for "not in bugzilla" -- that is, a bug that was reported to us
629 but never got a bugzilla entry. We encourage you to file bugs in
630 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
631 than mailing the developers (or mailing lists) directly -- bugs that
632 are not entered into bugzilla tend to get forgotten about or ignored.
634 To see details of a given bug, visit
635 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
636 where XXXXXX is the bug number as listed below.
638 79362 Debug info is lost for .so files when they are dlclose'd
639 208052 strlcpy error when n = 0
640 255603 exp-sgcheck Assertion '!already_present' failed
641 338252 building valgrind with -flto (link time optimisation) fails
642 345763 MIPS N32 ABI support
643 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
644 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
645 372347 Replacement problem of the additional c++14/c++17 new/delete operators
646 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
647 376257 helgrind history full speed up using a cached stack
648 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
650 379748 Fix missing pselect syscall (OS X 10.11)
651 379754 Fix missing syscall ulock_wait (OS X 10.12)
652 380397 s390x: __GI_strcspn() replacemenet needed
653 381162 possible array overrun in VEX register allocator
654 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
655 381274 powerpc too chatty even with --sigill-diagnostics=no
656 381289 epoll_pwait can have a NULL sigmask
657 381553 VEX register allocator v3
658 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
659 381769 Use ucontext_t instead of struct ucontext
660 381805 arm32 needs ld.so index hardwire for new glibc security fixes
661 382256 gz compiler flag test doesn't work for gold
662 382407 vg_perf needs "--terse" command line option
663 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
664 382563 MIPS MSA ASE support
665 382998 xml-socket doesn't work
666 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
667 383723 Fix missing kevent_qos syscall (macOS 10.11)
668 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
669 384096 Mention AddrCheck at Memcheck's command line option [..]
670 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
671 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
672 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
673 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
674 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
675 384337 performance improvements to VEX register allocator v2 and v3
676 384526 reduce number of spill insns generated by VEX register allocator v3
677 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
678 384631 Sanitise client args as printed with -v
679 384633 Add a simple progress-reporting facility
680 384987 VEX regalloc: allocate caller-save registers for short lived vregs
681 385055 PPC VEX temporary storage exhausted
682 385182 PPC64 is missing support for the DSCR
683 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
684 385207 PPC64, generate_store_FPRF() generates too many Iops
685 385208 PPC64, xxperm instruction exhausts temporary memory
686 385210 PPC64, vpermr instruction could exhaust temporary memory
687 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
688 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
689 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
690 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
691 385408 s390x: z13 vector "support" instructions not implemented
692 385409 s390x: z13 vector integer instructions not implemented
693 385410 s390x: z13 vector string instructions not implemented
694 385412 s390x: new non-vector z13 instructions not implemented
695 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
696 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
697 385939 Optionally exit on the first error
698 386318 valgrind.org/info/tools.html is missing SGCheck
699 386425 running valgrind + wine on armv7l gives illegal opcode
700 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
701 387410 MIPSr6 support
702 387664 Memcheck: make expensive-definedness-checks be the default
703 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
704 387766 asm shifts cause false positive "Conditional jump or move depends
705 on uninitialised value"
706 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
707 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
708 388786 Support bpf syscall in amd64 Linux
709 388862 Add replacements for wmemchr and wcsnlen on Linux
710 389065 valgrind meets gcc flag -Wlogical-op
711 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
712 390471 suppression by specification of source-file line number
713 390723 make xtree dump files world wide readable, similar to log files
714 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
715 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
716 392118 unhandled amd64-linux syscall: 332 (statx)
717 392449 callgrind not clearing the number of calls properly
718 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
719 lxssp, stxssp and stxvl instructions.
720 393023 callgrind_control risks using the wrong vgdb
721 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
722 393099 posix_memalign() invalid write if alignment == 0
723 393146 failing assert "is_DebugInfo_active(di)"
724 395709 PPC64 is missing support for the xvnegsp instruction
725 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
727 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
728 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
729 396839 s390x: Trap instructions not implemented
730 396887 arch_prctl should return EINVAL on unknown option
731 == 397286 crash before launching binary (Unsupported arch_prctl option)
732 == 397393 valgrind: the 'impossible' happened: (Archlinux)
733 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
734 396906 compile tests failure on mips32-linux: broken inline asm in tests on
736 397012 glibc ld.so uses arch_prctl on i386
737 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
738 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
739 397424 glibc 2.27 and gdb_server tests
740 398028 Assertion `cfsi_fits` failing in simple C program
741 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
743 n-i-bz Fix missing workq_ops operations (macOS)
744 n-i-bz fix bug in strspn replacement
745 n-i-bz Add support for the Linux BLKFLSBUF ioctl
746 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
747 n-i-bz Fix possible stack trashing by semctl syscall wrapping
748 n-i-bz Add support for the Linux membarrier() system call
749 n-i-bz x86 front end: recognise and handle UD2 correctly
750 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
751 correctly aligned before entering the handler.
753 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
754 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
755 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
759 Release 3.13.0 (15 June 2017)
760 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
762 3.13.0 is a feature release with many improvements and the usual collection of
765 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
766 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
767 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
768 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
770 * ==================== CORE CHANGES ===================
772 * The translation cache size has been increased to keep up with the demands of
773 large applications. The maximum number of sectors has increased from 24 to
774 48. The default number of sectors has increased from 16 to 32 on all
775 targets except Android, where the increase is from 6 to 12.
777 * The amount of memory that Valgrind can use has been increased from 64GB to
778 128GB. In particular this means your application can allocate up to about
779 60GB when running on Memcheck.
781 * Valgrind's default load address has been changed from 0x3800'0000 to
782 0x5800'0000, so as to make it possible to load larger executables. This
783 should make it possible to load executables of size at least 1200MB.
785 * A massive spaceleak caused by reading compressed debuginfo files has been
786 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
789 * The C++ demangler has been updated.
791 * Support for demangling Rust symbols has been added.
793 * A new representation of stack traces, the "XTree", has been added. An XTree
794 is a tree of stacktraces with data associated with the stacktraces. This is
795 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
796 consumption of your program. Reporting is controlled by the new options
797 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
799 A report can also be produced on demand using the gdbserver monitor command
800 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
801 format' and 'massif format. The existing visualisers for these formats (e.g.
802 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
803 analyse these reports.
805 Memcheck can also produce XTree leak reports using the Callgrind file
806 format. For more details, see the user manual.
808 * ================== PLATFORM CHANGES =================
810 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
812 * amd64: fixes for JIT failure problems on long AVX2 code blocks
814 * amd64 and x86: support for CET prefixes has been added
816 * arm32: a few missing ARMv8 instructions have been implemented
818 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
819 Store-Conditional instructions has been added. This is to deal with
820 processor implementations that implement the LL/SC specifications strictly
821 and as a result cause Valgrind to hang in certain situations. The
822 alternative implementation is automatically enabled at startup, as required.
823 You can use the option --sim-hints=fallback-llsc to force-enable it if you
826 * Support for OSX 10.12 has been improved.
828 * On Linux, clone handling has been improved to honour CLONE_VFORK that
829 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
830 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
831 CLONE_VM exact semantics will (still) not work.
833 * The TileGX/Linux port has been removed because it appears to be both unused
836 * ==================== TOOL CHANGES ====================
840 - Memcheck should give fewer false positives when running optimised
841 Clang/LLVM generated code.
843 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
845 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
846 to produce the end of execution leak report in a xtree callgrind format
849 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
850 the leak report in an xtree file.
854 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
856 - For some workloads (typically, for big applications), Massif memory
857 consumption and CPU consumption has decreased significantly.
861 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
863 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
864 for Ada gnat compiled applications.
866 * ==================== OTHER CHANGES ====================
868 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
869 append the inner guest stacktrace to the inner host stacktrace. This helps
870 to investigate the errors reported by the outer, when they are caused by the
871 inner guest program (such as an inner regtest). See README_DEVELOPERS for
874 * To allow fast detection of callgrind files by desktop environments and file
875 managers, the format was extended to have an optional first line that
876 uniquely identifies the format ("# callgrind format"). Callgrind creates
877 this line now, as does the new xtree functionality.
879 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
880 have a new %n format letter that is replaced by a sequence number.
882 * "--version -v" now shows the SVN revision numbers from which Valgrind was
885 * ==================== FIXED BUGS ====================
887 The following bugs have been fixed or resolved. Note that "n-i-bz"
888 stands for "not in bugzilla" -- that is, a bug that was reported to us
889 but never got a bugzilla entry. We encourage you to file bugs in
890 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
891 than mailing the developers (or mailing lists) directly -- bugs that
892 are not entered into bugzilla tend to get forgotten about or ignored.
894 To see details of a given bug, visit
895 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
896 where XXXXXX is the bug number as listed below.
898 162848 --log-file output isn't split when a program forks
899 340777 Illegal instruction on mips (ar71xx)
900 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
901 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
902 to a different stack.
903 344139 x86 stack-seg overrides, needed by the Wine people
904 344524 store conditional of guest applications always fail - observed on
906 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
907 352395 Please provide SVN revision info in --version -v
908 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
909 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
910 INVALID_POSIX_THREADID' failed
911 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
912 barrier implementation
913 358697 valgrind.h: Some code remains even when defining NVALGRIND
914 359202 Add musl libc configure/compile
915 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
916 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
917 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
918 362223 assertion failed when .valgrindrc is a directory instead of a file
919 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
920 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
921 368507 can't malloc chunks larger than about 34GB
922 368529 Android arm target link error, missing atexit and pthread_atfork
923 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
924 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
925 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
926 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
927 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
928 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
929 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
930 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
931 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
932 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
933 370028 Reduce the number of compiler warnings on MIPS platforms
934 370635 arm64 missing syscall getcpu
935 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
936 371227 Clean AArch64 syscall table
937 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
938 371471 Valgrind complains about non legit memory leaks on placement new (C++)
939 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
940 371503 disInstr(arm64): unhandled instruction 0xF89F0000
941 371869 support '%' in symbol Z-encoding
942 371916 execution tree xtree concept
943 372120 c++ demangler demangles symbols which are not c++
944 372185 Support of valgrind on ARMv8 with 32 bit executable
945 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
946 372195 Power PC, xxsel instruction is not always recognized.
947 372504 Hanging on exit_group
948 372600 process loops forever when fatal signals are arriving quickly
949 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
950 373046 Stacks registered by core are never deregistered
951 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
952 373086 Implement additional Xen hypercalls
953 373192 Calling posix_spawn in glibc 2.24 completely broken
954 373488 Support for fanotify API on ARM64 architecture
955 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
956 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
957 373938 const IRExpr arguments for matchIRExpr()
958 374719 some spelling fixes
959 374963 increase valgrind's load address to prevent mmap failure
960 375514 valgrind_get_tls_addr() does not work in case of static TLS
961 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
962 for ML_(find_rx_mapping)()
963 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
964 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
965 == 377159 "vex: the `impossible' happened" still present
966 == 375150 Assertion 'tres.status == VexTransOK' failed
967 == 378068 valgrind crashes on AVX2 function in FFmpeg
968 376142 Segfaults on MIPS Cavium Octeon boards
969 376279 disInstr(arm64): unhandled instruction 0xD50320FF
970 376455 Solaris: unhandled syscall lgrpsys(180)
971 376518 Solaris: unhandled fast trap getlgrp(6)
972 376611 ppc64 and arm64 don't know about prlimit64 syscall
973 376729 PPC64, remove R2 from the clobber list
975 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
976 to be wrongly marked as addressable
977 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
978 PIE enabled by default
979 377376 memcheck/tests/linux/getregset fails with glibc2.24
980 377427 PPC64, lxv instruction failing on odd destination register
981 377478 PPC64: ISA 3.0 setup fixes
982 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
983 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
984 and 2 args for FUTEX_TRYLOCK_PI
985 377717 Fix massive space leak when reading compressed debuginfo sections
986 377891 Update Xen 4.6 domctl wrappers
987 377930 fcntl syscall wrapper is missing flock structure check
988 378524 libvexmultiarch_test regression on s390x and ppc64
989 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
990 378673 Update libiberty demangler
991 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
992 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
993 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
994 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
995 (task_register_dyld_image_infos)
996 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
997 (task_register_dyld_shared_cache_image_info)
998 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
999 379473 MIPS: add support for rdhwr cycle counter register
1000 379504 remove TileGX/Linux port
1001 379525 Support more x86 nop opcodes
1002 379838 disAMode(x86): not an addr!
1003 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1004 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1005 379895 clock_gettime does not execute POST syscall wrapper
1006 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1007 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1008 380200 xtree generated callgrind files refer to files without directory name
1009 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1010 380397 s390x: __GI_strcspn() replacement needed
1011 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1013 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1014 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1015 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1019 Release 3.12.0 (20 October 2016)
1020 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1022 3.12.0 is a feature release with many improvements and the usual
1023 collection of bug fixes.
1025 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1026 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1027 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1028 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1029 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1030 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1032 * ================== PLATFORM CHANGES =================
1034 * POWER: Support for ISA 3.0 has been added
1036 * mips: support for O32 FPXX ABI has been added.
1037 * mips: improved recognition of different processors
1038 * mips: determination of page size now done at run time
1040 * amd64: Partial support for AMD FMA4 instructions.
1042 * arm, arm64: Support for v8 crypto and CRC instructions.
1044 * Improvements and robustification of the Solaris port.
1046 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1048 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1049 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1050 where possible. Valgrind's support for 32-bit x86 has stagnated in
1051 recent years and has fallen far behind that for 64-bit x86
1052 instructions. By contrast 64-bit x86 is well supported, up to and
1055 * ==================== TOOL CHANGES ====================
1059 - Added meta mempool support for describing a custom allocator which:
1060 - Auto-frees all chunks assuming that destroying a pool destroys all
1062 - Uses itself to allocate other memory blocks
1064 - New flag --ignore-range-below-sp to ignore memory accesses below
1065 the stack pointer, if you really have to. The related flag
1066 --workaround-gcc296-bugs=yes is now deprecated. Use
1067 --ignore-range-below-sp=1024-1 as a replacement.
1071 - Improved thread startup time significantly on non-Linux platforms.
1075 - Added collection of the metric "tot-blocks-allocd"
1077 * ==================== OTHER CHANGES ====================
1079 * Replacement/wrapping of malloc/new related functions is now done not just
1080 for system libraries by default, but for any globally defined malloc/new
1081 related function (both in shared libraries and statically linked alternative
1082 malloc implementations). The dynamic (runtime) linker is excluded, though.
1083 To only intercept malloc/new related functions in
1084 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1085 "nouserintercepts" can be any non-existing library name).
1086 This new functionality is not implemented for MacOS X.
1088 * The maximum number of callers in a suppression entry is now equal to
1089 the maximum size for --num-callers (500).
1090 Note that --gen-suppressions=yes|all similarly generates suppressions
1091 containing up to --num-callers frames.
1093 * New and modified GDB server monitor features:
1095 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1096 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1099 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1100 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1103 * Valgrind is able to read compressed debuginfo sections in two formats:
1104 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1105 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1107 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1108 for the most common use case (x86_64-linux, Memcheck) has been
1111 * Improved performance for programs that do a lot of discarding of
1112 instruction address ranges of 8KB or less.
1114 * The C++ symbol demangler has been updated.
1116 * More robustness against invalid syscall parameters on Linux.
1118 * ==================== FIXED BUGS ====================
1120 The following bugs have been fixed or resolved. Note that "n-i-bz"
1121 stands for "not in bugzilla" -- that is, a bug that was reported to us
1122 but never got a bugzilla entry. We encourage you to file bugs in
1123 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1124 than mailing the developers (or mailing lists) directly -- bugs that
1125 are not entered into bugzilla tend to get forgotten about or ignored.
1127 To see details of a given bug, visit
1128 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1129 where XXXXXX is the bug number as listed below.
1131 191069 Exiting due to signal not reported in XML output
1132 199468 Suppressions: stack size limited to 25
1133 while --num-callers allows more frames
1134 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1135 278744 cvtps2pd with redundant RexW
1136 303877 valgrind doesn't support compressed debuginfo sections.
1137 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1138 348345 Assertion fails for negative lineno
1139 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1140 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1141 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1142 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1143 352197 mips: mmap2() not wrapped correctly for page size > 4096
1144 353083 arm64 doesn't implement various xattr system calls
1145 353084 arm64 doesn't support sigpending system call
1146 353137 www: update info for Supported Platforms
1147 353138 www: update "The Valgrind Developers" page
1148 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1151 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1152 353398 WARNING: unhandled amd64-solaris syscall: 207
1153 353660 XML in auxwhat tag not escaping reserved symbols properly
1154 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1155 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1156 353802 ELF debug info reader confused with multiple .rodata sections
1157 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1158 353917 unhandled amd64-solaris syscall fchdir(120)
1159 353920 unhandled amd64-solaris syscall: 170
1160 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1161 354392 unhandled amd64-solaris syscall: 171
1162 354797 Vbit test does not include Iops for Power 8 instruction support
1163 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1167 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1168 355188 valgrind should intercept all malloc related global functions
1169 355454 do not intercept malloc related symbols from the runtime linker
1170 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1171 356044 Dwarf line info reader misinterprets is_stmt register
1172 356112 mips: replace addi with addiu
1173 356393 valgrind (vex) crashes because isZeroU happened
1176 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1177 356678 arm64-linux: unhandled syscall 232 (mincore)
1178 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1179 356823 Unsupported ARM instruction: stlex
1180 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1181 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1182 357673 crash if I try to run valgrind with a binary link with libcurl
1183 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1184 357871 pthread_spin_destroy not properly wrapped
1185 357887 Calls to VG_(fclose) do not close the file descriptor
1186 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1187 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1188 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1189 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1190 359181 Buffer Overflow during Demangling
1191 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1192 359289 s390x: popcnt (B9E1) not implemented
1193 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1194 359503 Add missing syscalls for aarch64 (arm64)
1195 359645 "You need libc6-dbg" help message could be more helpful
1196 359703 s390: wire up separate socketcalls system calls
1197 359724 getsockname might crash - deref_UInt should call safe_to_deref
1198 359733 amd64 implement ld.so strchr/index override like x86
1199 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1200 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1202 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1203 359871 Incorrect mask handling in ppoll
1204 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1205 360008 Contents of Power vr registers contents is not printed correctly when
1206 the --vgdb-shadow-registers=yes option is used
1207 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1208 non-zero shadow bits
1209 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1210 360425 arm64 unsupported instruction ldpsw
1212 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1213 360571 Error about the Android Runtime reading below the stack pointer on ARM
1214 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1215 360749 kludge for multiple .rodata sections on Solaris no longer needed
1216 360752 raise the number of reserved fds in m_main.c from 10 to 12
1217 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1218 361226 s390x: risbgn (EC59) not implemented
1219 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1220 361354 ppc64[le]: wire up separate socketcalls system calls
1221 361615 Inconsistent termination for multithreaded process terminated by signal
1222 361926 Unhandled Solaris syscall: sysfs(84)
1223 362009 V dumps core on unimplemented functionality before threads are created
1224 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1225 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1226 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1227 362953 Request for an update to the Valgrind Developers page
1228 363680 add renameat2() support
1229 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1230 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1231 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1232 364058 clarify in manual limitations of array overruns detections
1233 364413 pselect sycallwrapper mishandles NULL sigmask
1234 364728 Power PC, missing support for several HW registers in
1235 get_otrack_shadow_offset_wrk()
1236 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1237 365273 Invalid write to stack location reported after signal handler runs
1238 365912 ppc64BE segfault during jm-insns test (RELRO)
1239 366079 FPXX Support for MIPS32 Valgrind
1240 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1241 366344 Multiple unhandled instruction for Aarch64
1242 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1243 367995 Integration of memcheck with custom memory allocator
1244 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1245 368412 False positive result for altivec capability check
1246 368416 Add tc06_two_races_xml.exp output for ppc64
1247 368419 Perf Events ioctls not implemented
1248 368461 mmapunmap test fails on ppc64
1249 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1250 369000 AMD64 fma4 instructions unsupported.
1251 369169 ppc64 fails jm_int_isa_2_07 test
1252 369175 jm_vec_isa_2_07 test crashes on ppc64
1253 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1254 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1255 369359 msghdr_foreachfield can crash when handling bad iovec
1256 369360 Bad sigprocmask old or new sets can crash valgrind
1257 369361 vmsplice syscall wrapper crashes on bad iovec
1258 369362 Bad sigaction arguments crash valgrind
1259 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1260 369402 Bad set/get_thread_area pointer crashes valgrind
1261 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1262 369446 valgrind crashes on unknown fcntl command
1263 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1264 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1265 370265 ISA 3.0 HW cap stuff needs updating
1266 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1267 372195 Power PC, xxsel instruction is not always recognized
1269 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1270 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1271 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1272 n-i-bz Fix ppoll_alarm exclusion on OS X
1273 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1274 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1275 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1276 n-i-bz amd64: memcheck false positive with shr %edx
1277 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1278 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1279 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1280 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1282 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1283 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1284 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1288 Release 3.11.0 (22 September 2015)
1289 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1291 3.11.0 is a feature release with many improvements and the usual
1292 collection of bug fixes.
1294 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1295 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1296 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1297 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1298 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1299 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1301 * ================== PLATFORM CHANGES =================
1303 * Support for Solaris/x86 and Solaris/amd64 has been added.
1305 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1307 * Preliminary support for the Tilera TileGX architecture has been added.
1309 * s390x: It is now required for the host to have the "long displacement"
1310 facility. The oldest supported machine model is z990.
1312 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1313 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1315 * The JIT's register allocator is significantly faster, making the JIT
1316 as a whole somewhat faster, so JIT-intensive activities, for example
1317 program startup, are modestly faster, around 5%.
1319 * There have been changes to the default settings of several command
1320 line flags, as detailed below.
1322 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1323 capable hosts, the simulated CPUID will now indicate AVX2 support.
1325 * ==================== TOOL CHANGES ====================
1329 - The default value for --leak-check-heuristics has been changed from
1330 "none" to "all". This helps to reduce the number of possibly
1331 lost blocks, in particular for C++ applications.
1333 - The default value for --keep-stacktraces has been changed from
1334 "malloc-then-free" to "malloc-and-free". This has a small cost in
1335 memory (one word per malloc-ed block) but allows Memcheck to show the
1336 3 stacktraces of a dangling reference: where the block was allocated,
1337 where it was freed, and where it is acccessed after being freed.
1339 - The default value for --partial-loads-ok has been changed from "no" to
1340 "yes", so as to avoid false positive errors resulting from some kinds
1341 of vectorised loops.
1343 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1344 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1345 than get_vbits when you need to associate byte data value with
1346 their corresponding validity bits.
1348 - The 'block_list' monitor command has been enhanced:
1349 o it can print a range of loss records
1350 o it now accepts an optional argument 'limited <max_blocks>'
1351 to control the number of blocks printed.
1352 o if a block has been found using a heuristic, then
1353 'block_list' now shows the heuristic after the block size.
1354 o the loss records/blocks to print can be limited to the blocks
1355 found via specified heuristics.
1357 - The C helper functions used to instrument loads on
1358 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1359 replaced by handwritten assembly sequences. This gives speedups
1360 in the region of 0% to 7% for those targets only.
1362 - A new command line option, --expensive-definedness-checks=yes|no,
1363 has been added. This is useful for avoiding occasional invalid
1364 uninitialised-value errors in optimised code. Watch out for
1365 runtime degradation, as this can be up to 25%. As always, though,
1366 the slowdown is highly application specific. The default setting
1371 - A new monitor command 'all_snapshots <filename>' dumps all
1372 snapshots taken so far.
1376 - Significant memory reduction and moderate speedups for
1377 --history-level=full for applications accessing a lot of memory
1378 with many different stacktraces.
1380 - The default value for --conflict-cache-size=N has been doubled to
1381 2000000. Users that were not using the default value should
1382 preferably also double the value they give.
1384 The default was changed due to the changes in the "full history"
1385 implementation. Doubling the value gives on average a slightly more
1386 complete history and uses similar memory (or significantly less memory
1387 in the worst case) than the previous implementation.
1389 - The Helgrind monitor command 'info locks' now accepts an optional
1390 argument 'lock_addr', which shows information about the lock at the
1393 - When using --history-level=full, the new Helgrind monitor command
1394 'accesshistory <addr> [<len>]' will show the recorded accesses for
1395 <len> (or 1) bytes at <addr>.
1397 * ==================== OTHER CHANGES ====================
1399 * The default value for the --smc-check option has been changed from
1400 "stack" to "all-non-file" on targets that provide automatic D-I
1401 cache coherence (x86, amd64 and s390x). The result is to provide,
1402 by default, transparent support for JIT generated and self-modifying
1403 code on all targets.
1405 * Mac OS X only: the default value for the --dsymutil option has been
1406 changed from "no" to "yes", since any serious usage on Mac OS X
1407 always required it to be "yes".
1409 * The command line options --db-attach and --db-command have been removed.
1410 They were deprecated in 3.10.0.
1412 * When a process dies due to a signal, Valgrind now shows the signal
1413 and the stacktrace at default verbosity (i.e. verbosity 1).
1415 * The address description logic used by Memcheck and Helgrind now
1416 describes addresses in anonymous segments, file mmap-ed segments,
1417 shared memory segments and the brk data segment.
1419 * The new option --error-markers=<begin>,<end> can be used to mark the
1420 begin/end of errors in textual output mode, to facilitate
1421 searching/extracting errors in output files that mix valgrind errors
1422 with program output.
1424 * The new option --max-threads=<number> can be used to change the number
1425 of threads valgrind can handle. The default is 500 threads which
1426 should be more than enough for most applications.
1428 * The new option --valgrind-stacksize=<number> can be used to change the
1429 size of the private thread stacks used by Valgrind. This is useful
1430 for reducing memory use or increasing the stack size if Valgrind
1431 segfaults due to stack overflow.
1433 * The new option --avg-transtab-entry-size=<number> can be used to specify
1434 the expected instrumented block size, either to reduce memory use or
1435 to avoid excessive retranslation.
1437 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1439 * New and modified GDB server monitor features:
1441 - When a signal is reported in GDB, you can now use the GDB convenience
1442 variable $_siginfo to examine detailed signal information.
1444 - Valgrind's gdbserver now allows the user to change the signal
1445 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1446 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1447 continue without passing the signal to the process.
1449 - With GDB >= 7.10, the command 'target remote'
1450 will automatically load the executable file of the process running
1451 under Valgrind. This means you do not need to specify the executable
1452 file yourself, GDB will discover it itself. See GDB documentation about
1453 'qXfer:exec-file:read' packet for more info.
1455 * ==================== FIXED BUGS ====================
1457 The following bugs have been fixed or resolved. Note that "n-i-bz"
1458 stands for "not in bugzilla" -- that is, a bug that was reported to us
1459 but never got a bugzilla entry. We encourage you to file bugs in
1460 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1461 than mailing the developers (or mailing lists) directly -- bugs that
1462 are not entered into bugzilla tend to get forgotten about or ignored.
1464 To see details of a given bug, visit
1465 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1466 where XXXXXX is the bug number as listed below.
1468 116002 VG_(printf): Problems with justification of strings and integers
1469 155125 avoid cutting away file:lineno after long function name
1470 197259 Unsupported arch_prtctl PR_SET_GS option
1471 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
1472 201216 Fix Valgrind does not support pthread_sigmask() on OS X
1473 201435 Fix Darwin: -v does not show kernel version
1474 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
1475 211256 Fixed an outdated comment regarding the default platform.
1476 211529 Incomplete call stacks for code compiled by newer versions of MSVC
1477 211926 Avoid compilation warnings in valgrind.h with -pedantic
1478 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
1480 226609 Crediting upstream authors in man page
1481 231257 Valgrind omits path when executing script from shebang line
1482 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
1483 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
1484 269360 s390x: Fix addressing mode selection for compare-and-swap
1485 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
1487 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
1488 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
1489 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
1490 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
1491 330147 libmpiwrap PMPI_Get_count returns undefined value
1492 333051 mmap of huge pages fails due to incorrect alignment
1494 334802 valgrind does not always explain why a given option is bad
1495 335618 mov.w rN, pc/sp (ARM32)
1496 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
1501 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
1502 338602 AVX2 bit in CPUID missing
1503 338606 Strange message for scripts with invalid interpreter
1504 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
1505 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
1506 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
1508 339156 gdbsrv not called for fatal signal
1509 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
1510 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
1511 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
1512 339442 Fix testsuite build failure on OS X 10.9
1513 339542 Enable compilation with Intel's ICC compiler
1514 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
1515 339688 Mac-specific ASM does not support .version directive (cpuid,
1516 tronical and pushfpopf tests)
1517 339745 Valgrind crash when check Marmalade app (partial fix)
1518 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
1519 339778 Linux/TileGx platform support for Valgrind
1520 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
1521 339789 Fix none/tests/execve test on Mac OS X 10.9
1522 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
1523 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
1524 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
1525 340392 Allow user to select more accurate definedness checking in memcheck
1526 to avoid invalid complaints on optimised code
1527 340430 Fix some grammatical weirdness in the manual.
1528 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
1529 341419 Signal handler ucontext_t not filled out correctly on OS X
1530 341539 VG_(describe_addr) should not describe address as belonging to client
1531 segment if it is past the heap end
1532 341613 Enable building of manythreads and thread-exits tests on Mac OS X
1533 341615 Fix none/tests/darwin/access_extended test on Mac OS X
1534 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
1535 341789 aarch64: shmat fails with valgrind on ARMv8
1536 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
1537 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
1538 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
1539 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
1540 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
1541 342221 socket connect false positive uninit memory for unknown af family
1542 342353 Allow dumping full massif output while valgrind is still running
1543 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
1547 342603 Add I2C_SMBUS ioctl support
1548 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
1549 342683 Mark memory past the initial brk limit as unaddressable
1550 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
1551 342795 Internal glibc __GI_mempcpy call should be intercepted
1552 342841 s390x: Support instructions fiebr(a) and fidbr(a)
1553 343012 Unhandled syscall 319 (memfd_create)
1554 343069 Patch updating v4l2 API support
1555 343173 helgrind crash during stack unwind
1556 343219 fix GET_STARTREGS for arm
1557 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
1558 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
1559 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
1560 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
1561 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
1562 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
1563 343597 ppc64le: incorrect use of offseof macro
1564 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
1565 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
1566 343732 Unhandled syscall 144 (setgid) on aarch64
1567 343733 Unhandled syscall 187 (msgctl and related) on aarch64
1568 343802 s390x: False positive "conditional jump or move depends on [..]
1569 343902 --vgdb=yes doesn't break when --xml=yes is used
1570 343967 Don't warn about setuid/setgid/setcap executable for directories
1571 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
1572 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
1573 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
1574 344054 www - update info for Solaris/illumos
1575 344416 'make regtest' does not work cleanly on OS X
1576 344235 Remove duplicate include of pub_core_aspacemgr.h
1577 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
1578 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
1579 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
1580 344314 callgrind_annotate ... warnings about commands containing newlines
1581 344318 socketcall should wrap recvmmsg and sendmmsg
1582 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
1583 344416 Fix 'make regtest' does not work cleanly on OS X
1584 344499 Fix compilation for Linux kernel >= 4.0.0
1585 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
1586 unix:349 (__pthread_fchdir)
1587 344559 Garbage collection of unused segment names in address space manager
1588 344560 Fix stack traces missing penultimate frame on OS X
1589 344621 Fix memcheck/tests/err_disable4 test on OS X
1590 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
1591 344702 Fix missing libobjc suppressions on OS X 10.10
1593 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
1594 344939 Fix memcheck/tests/xml1 on OS X 10.10
1595 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
1596 345079 Fix build problems in VEX/useful/test_main.c
1597 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
1598 345177 arm64: prfm (reg) not implemented
1599 345215 Performance improvements for the register allocator
1600 345248 add support for Solaris OS in valgrind
1601 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
1602 345394 Fix memcheck/tests/strchr on OS X
1603 345637 Fix memcheck/tests/sendmsg on OS X
1604 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
1605 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
1606 345887 Fix an assertion in the address space manager
1607 345928 amd64: callstack only contains current function for small stacks
1608 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1609 345987 MIPS64: Implement cavium LHX instruction
1610 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1611 346185 Fix typo saving altivec register v24
1612 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1613 and LibVEX_GuestPPC64_get_CR()
1614 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1615 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1616 346307 fuse filesystem syscall deadlocks
1617 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1618 346411 MIPS: SysRes::_valEx handling is incorrect
1619 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1620 346474 PPC64 Power 8, spr TEXASRU register not supported
1621 346487 Compiler generates "note" about a future ABI change for PPC64
1622 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1623 and causing spurious "invalid read of size 8" warnings
1624 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1625 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1626 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1627 347322 Power PC regression test cleanup
1628 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1630 347389 unhandled syscall: 373 (Linux ARM syncfs)
1631 347686 Patch set to cleanup PPC64 regtests
1632 347978 Remove bash dependencies where not needed
1633 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1634 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1636 348102 Patch updating v4l2 API support
1637 348247 amd64 front end: jno jumps wrongly when overflow is not set
1638 348269 Improve mmap MAP_HUGETLB support.
1639 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1640 348345 Assertion fails for negative lineno
1641 348377 Unsupported ARM instruction: yield
1642 348565 Fix detection of command line option availability for clang
1643 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1644 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1645 348748 Fix redundant condition
1646 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1647 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1648 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1649 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1650 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1651 349626 Implemented additional Xen hypercalls
1652 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1653 349790 Clean up of the hardware capability checking utilities.
1654 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1655 349874 Fix typos in source code
1656 349879 memcheck: add handwritten assembly for helperc_LOADV*
1657 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1658 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1659 350202 Add limited param to 'monitor block_list'
1660 350290 s390x: Support instructions fixbr(a)
1661 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1662 350809 Fix none/tests/async-sigs for Solaris
1663 350811 Remove reference to --db-attach which has been removed.
1664 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1665 350854 hard-to-understand code in VG_(load_ELF)()
1666 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1667 351386 Solaris: Cannot run ld.so.1 under Valgrind
1668 351474 Fix VG_(iseqsigset) as obvious
1669 351531 Typo in /include/vki/vki-xen-physdev.h header guard
1670 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
1671 351858 ldsoexec support on Solaris
1672 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
1673 352130 helgrind reports false races for printfs using mempcpy on FILE* state
1674 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
1675 352320 arm64 crash on none/tests/nestedfs
1676 352765 Vbit test fails on Power 6
1677 352768 The mbar instruction is missing from the Power PC support
1678 352769 Power PC program priority register (PPR) is not supported
1679 n-i-bz Provide implementations of certain compiler builtins to support
1680 compilers that may not provide those
1681 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
1682 n-i-bz Fix compilation on distros with glibc < 2.5
1683 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
1684 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
1685 n-i-bz mremap did not work properly on shared memory
1686 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
1687 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
1689 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
1690 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
1691 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
1695 Release 3.10.1 (25 November 2014)
1696 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1697 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
1698 and backports fixes for all reported missing AArch64 ARMv8 instructions
1699 and syscalls from the trunk. If you package or deliver 3.10.0 for others
1700 to use, you might want to consider upgrading to 3.10.1 instead.
1702 The following bugs have been fixed or resolved. Note that "n-i-bz"
1703 stands for "not in bugzilla" -- that is, a bug that was reported to us
1704 but never got a bugzilla entry. We encourage you to file bugs in
1705 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1706 than mailing the developers (or mailing lists) directly -- bugs that
1707 are not entered into bugzilla tend to get forgotten about or ignored.
1709 To see details of a given bug, visit
1710 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1711 where XXXXXX is the bug number as listed below.
1713 335440 arm64: ld1 (single structure) is not implemented
1714 335713 arm64: unhanded instruction: prfm (immediate)
1715 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
1716 339182 ppc64: AvSplat ought to load destination vector register with [..]
1717 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
1718 339433 ppc64 lxvw4x instruction uses four 32-byte loads
1719 339645 Use correct tag names in sys_getdents/64 wrappers
1720 339706 Fix false positive for ioctl(TIOCSIG) on linux
1721 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
1722 339853 arm64 times syscall unknown
1723 339855 arm64 unhandled getsid/setsid syscalls
1724 339858 arm64 dmb sy not implemented
1725 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
1726 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
1727 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
1729 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
1730 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
1731 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
1732 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
1733 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
1734 340509 arm64: unhandled instruction fcvtas
1735 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
1736 340632 arm64: unhandled instruction fcvtas
1737 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
1738 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
1739 340788 warning: unhandled syscall: 318 (getrandom)
1740 340807 disInstr(arm): unhandled instruction: 0xEE989B20
1741 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
1742 340922 arm64: unhandled getgroups/setgroups syscalls
1743 350251 Fix typo in VEX utility program (test_main.c).
1744 350407 arm64: unhandled instruction ucvtf (vector, integer)
1745 350809 none/tests/async-sigs breaks when run under cron on Solaris
1746 350811 update README.solaris after r15445
1747 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
1748 350854 strange code in VG_(load_ELF)()
1749 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1750 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
1751 n-i-bz Add missing ]] to terminate CDATA.
1752 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
1753 n-i-bz Enable sys_fadvise64_64 on arm32.
1754 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
1755 n-i-bz Add test cases for all known arm64 load/store instructions.
1756 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
1757 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
1758 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
1759 n-i-bz arm64: Implement "fcvtpu w, s".
1760 n-i-bz arm64: implement ADDP and various others
1761 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
1762 n-i-bz arm64: enable FCVT{A,N}S X,S.
1764 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
1768 Release 3.10.0 (10 September 2014)
1769 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1771 3.10.0 is a feature release with many improvements and the usual
1772 collection of bug fixes.
1774 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1775 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1776 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
1777 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
1778 significantly improved relative to the 3.9.0 release.
1780 * ================== PLATFORM CHANGES =================
1782 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
1783 is mostly complete, and is usable, but some SIMD instructions are as
1786 * Support for little-endian variant of the 64-bit POWER architecture.
1788 * Support for Android on MIPS32.
1790 * Support for 64bit FPU on MIPS32 platforms.
1792 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
1794 * Configuration for and running on Android targets has changed.
1795 See README.android in the source tree for details.
1797 * ================== DEPRECATED FEATURES =================
1799 * --db-attach is now deprecated and will be removed in the next
1800 valgrind feature release. The built-in GDB server capabilities are
1801 superior and should be used instead. Learn more here:
1802 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
1804 * ==================== TOOL CHANGES ====================
1808 - Client code can now selectively disable and re-enable reporting of
1809 invalid address errors in specific ranges using the new client
1810 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
1811 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
1813 - Leak checker: there is a new leak check heuristic called
1814 "length64". This is used to detect interior pointers pointing 8
1815 bytes inside a block, on the assumption that the first 8 bytes
1816 holds the value "block size - 8". This is used by
1817 sqlite3MemMalloc, for example.
1819 - Checking of system call parameters: if a syscall parameter
1820 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
1821 several fields not initialised, an error is now reported for each
1822 field. Previously, an error was reported only for the first
1823 uninitialised field.
1825 - Mismatched alloc/free checking: a new flag
1826 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
1827 such checks if necessary.
1831 - Improvements to error messages:
1833 o Race condition error message involving heap allocated blocks also
1834 show the thread number that allocated the raced-on block.
1836 o All locks referenced by an error message are now announced.
1837 Previously, some error messages only showed the lock addresses.
1839 o The message indicating where a lock was first observed now also
1840 describes the address/location of the lock.
1842 - Helgrind now understands the Ada task termination rules and
1843 creates a happens-before relationship between a terminated task
1844 and its master. This avoids some false positives and avoids a big
1845 memory leak when a lot of Ada tasks are created and terminated.
1846 The interceptions are only activated with forthcoming releases of
1847 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
1849 - A new GDB server monitor command "info locks" giving the list of
1850 locks, their location, and their status.
1854 - callgrind_control now supports the --vgdb-prefix argument,
1855 which is needed if valgrind was started with this same argument.
1857 * ==================== OTHER CHANGES ====================
1859 * Unwinding through inlined function calls. Stack unwinding can now
1860 make use of Dwarf3 inlined-unwind information if it is available.
1861 The practical effect is that inlined calls become visible in stack
1862 traces. The suppression matching machinery has been adjusted
1863 accordingly. This is controlled by the new option
1864 --read-inline-info=yes|no. Currently this is enabled by default
1865 only on Linux and Android targets and only for the tools Memcheck,
1868 * Valgrind can now read EXIDX unwind information on 32-bit ARM
1869 targets. If an object contains both CFI and EXIDX unwind
1870 information, Valgrind will prefer the CFI over the EXIDX. This
1871 facilitates unwinding through system libraries on arm-android
1874 * Address description logic has been improved and is now common
1875 between Memcheck and Helgrind, resulting in better address
1876 descriptions for some kinds of error messages.
1878 * Error messages about dubious arguments (eg, to malloc or calloc) are
1879 output like other errors. This means that they can be suppressed
1880 and they have a stack trace.
1882 * The C++ demangler has been updated for better C++11 support.
1884 * New and modified GDB server monitor features:
1886 - Thread local variables/storage (__thread) can now be displayed.
1888 - The GDB server monitor command "v.info location <address>"
1889 displays information about an address. The information produced
1890 depends on the tool and on the options given to valgrind.
1891 Possibly, the following are described: global variables, local
1892 (stack) variables, allocated or freed blocks, ...
1894 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
1895 ask the GDB server to stop at the start of program execution, at
1896 the end of the program execution and on Valgrind internal errors.
1898 - A new monitor command "v.info stats" shows various Valgrind core
1899 and tool statistics.
1901 - A new monitor command "v.set hostvisibility" allows the GDB server
1902 to provide access to Valgrind internal host status/memory.
1904 * A new option "--aspace-minaddr=<address>" can in some situations
1905 allow the use of more memory by decreasing the address above which
1906 Valgrind maps memory. It can also be used to solve address
1907 conflicts with system libraries by increasing the default value.
1908 See user manual for details.
1910 * The amount of memory used by Valgrind to store debug info (unwind
1911 info, line number information and symbol data) has been
1912 significantly reduced, even though Valgrind now reads more
1913 information in order to support unwinding of inlined function calls.
1915 * Dwarf3 handling with --read-var-info=yes has been improved:
1917 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
1919 - Code compiled with
1920 -ffunction-sections -fdata-sections -Wl,--gc-sections
1921 no longer causes assertion failures.
1923 * Improved checking for the --sim-hints= and --kernel-variant=
1924 options. Unknown strings are now detected and reported to the user
1927 * The semantics of stack start/end boundaries in the valgrind.h
1928 VALGRIND_STACK_REGISTER client request has been clarified and
1929 documented. The convention is that start and end are respectively
1930 the lowest and highest addressable bytes of the stack.
1932 * ==================== FIXED BUGS ====================
1934 The following bugs have been fixed or resolved. Note that "n-i-bz"
1935 stands for "not in bugzilla" -- that is, a bug that was reported to us
1936 but never got a bugzilla entry. We encourage you to file bugs in
1937 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1938 than mailing the developers (or mailing lists) directly -- bugs that
1939 are not entered into bugzilla tend to get forgotten about or ignored.
1941 To see details of a given bug, visit
1942 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1943 where XXXXXX is the bug number as listed below.
1945 175819 Support for ipv6 socket reporting with --track-fds
1946 232510 make distcheck fails
1947 249435 Analyzing wine programs with callgrind triggers a crash
1948 278972 support for inlined function calls in stacktraces and suppression
1950 291310 FXSAVE instruction marks memory as undefined on amd64
1951 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
1952 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
1953 315199 vgcore file for threaded app does not show which thread crashed
1954 315952 tun/tap ioctls are not supported
1955 323178 Unhandled instruction: PLDW register (ARM)
1956 323179 Unhandled instruction: PLDW immediate (ARM)
1957 324050 Helgrind: SEGV because of unaligned stack when using movdqa
1958 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
1959 325124 [MIPSEL] Compilation error
1960 325477 Phase 4 support for IBM Power ISA 2.07
1961 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
1962 325628 Phase 5 support for IBM Power ISA 2.07
1963 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
1964 325751 Missing the two privileged Power PC Transactional Memory Instructions
1965 325816 Phase 6 support for IBM Power ISA 2.07
1966 325856 Make SGCheck fail gracefully on unsupported platforms
1967 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
1968 326436 DRD: False positive in libstdc++ std::list::push_back
1969 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
1970 326462 Refactor vgdb to isolate invoker stuff into separate module
1971 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
1972 326623 DRD: false positive conflict report in a field assignment
1973 326724 Valgrind does not compile on OSX 1.9 Mavericks
1974 326816 Intercept for __strncpy_sse2_unaligned missing?
1975 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
1976 326983 Clear direction flag after tests on amd64.
1977 327212 Do not prepend the current directory to absolute path names.
1978 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
1979 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
1980 327284 s390x: Fix translation of the risbg instruction
1981 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
1982 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
1983 327916 DW_TAG_typedef may have no name
1984 327943 s390x: add a redirection for the 'index' function
1985 328100 XABORT not implemented
1986 328205 Implement additional Xen hypercalls
1987 328454 add support Backtraces with ARM unwind tables (EXIDX)
1988 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
1989 328711 valgrind.1 manpage "memcheck options" section is badly generated
1990 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
1991 329612 Incorrect handling of AT_BASE for image execution
1992 329694 clang warns about using uninitialized variable
1993 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
1994 330228 mmap must align to VKI_SHMLBA on mips32
1995 330257 LLVM does not support `-mno-dynamic-no-pic` option
1996 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
1997 330459 --track-fds=yes doesn't track eventfds
1998 330469 Add clock_adjtime syscall support
1999 330594 Missing sysalls on PowerPC / uClibc
2000 330622 Add test to regression suite for POWER instruction: dcbzl
2001 330939 Support for AMD's syscall instruction on x86
2003 330941 Typo in PRE(poll) syscall wrapper
2004 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2005 331254 Fix expected output for memcheck/tests/dw4
2006 331255 Fix race condition in test none/tests/coolo_sigaction
2007 331257 Fix type of jump buffer in test none/tests/faultstatus
2008 331305 configure uses bash specific syntax
2009 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2010 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2011 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2012 331829 Unexpected ioctl opcode sign extension
2013 331830 ppc64: WARNING: unhandled syscall: 96/97
2014 331839 drd/tests/sem_open specifies invalid semaphore name
2015 331847 outcome of drd/tests/thread_name is nondeterministic
2016 332037 Valgrind cannot handle Thumb "add pc, reg"
2017 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2018 consistency checks enabled
2019 332263 intercepts for pthread_rwlock_timedrdlock and
2020 pthread_rwlock_timedwrlock are incorrect
2021 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2023 332276 Implement additional Xen hypercalls
2024 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2025 332765 Fix ms_print to create temporary files in a proper directory
2026 333072 drd: Add semaphore annotations
2027 333145 Tests for missaligned PC+#imm access for arm
2028 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2029 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2030 333248 WARNING: unhandled syscall: unix:443
2031 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2032 333501 cachegrind: assertion: Cache set count is not a power of two.
2035 333666 Recognize MPX instructions and bnd prefix.
2036 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2037 333817 Valgrind reports the memory areas written to by the SG_IO
2039 334049 lzcnt fails silently (x86_32)
2040 334384 Valgrind does not have support Little Endian support for
2042 334585 recvmmsg unhandled (+patch) (arm)
2043 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2044 334727 Build fails with -Werror=format-security
2045 334788 clarify doc about --log-file initial program directory
2046 334834 PPC64 Little Endian support, patch 2
2047 334836 PPC64 Little Endian support, patch 3 testcase fixes
2048 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2049 335034 Unhandled ioctl: HCIGETDEVLIST
2050 335155 vgdb, fix error print statement.
2051 335262 arm64: movi 8bit version is not supported
2052 335263 arm64: dmb instruction is not implemented
2053 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2054 335496 arm64: sbc/abc instructions are not implemented
2055 335554 arm64: unhandled instruction: abs
2056 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2057 335735 arm64: unhandled instruction: cnt
2058 335736 arm64: unhandled instruction: uaddlv
2059 335848 arm64: unhandled instruction: {s,u}cvtf
2060 335902 arm64: unhandled instruction: sli
2061 335903 arm64: unhandled instruction: umull (vector)
2062 336055 arm64: unhandled instruction: mov (element)
2063 336062 arm64: unhandled instruction: shrn{,2}
2064 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2065 336189 arm64: unhandled Instruction: mvn
2066 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2067 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2068 336772 Make moans about unknown ioctls more informative
2069 336957 Add a section about the Solaris/illumos port on the webpage
2070 337094 ifunc wrapper is broken on ppc64
2071 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2072 337528 leak check heuristic for block prefixed by length as 64bit number
2073 337740 Implement additional Xen hypercalls
2074 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2075 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2076 337871 deprecate --db-attach
2077 338023 Add support for all V4L2/media ioctls
2078 338024 inlined functions are not shown if DW_AT_ranges is used
2079 338106 Add support for 'kcmp' syscall
2080 338115 DRD: computed conflict set differs from actual after fork
2081 338160 implement display of thread local storage in gdbsrv
2082 338205 configure.ac and check for -Wno-tautological-compare
2083 338300 coredumps are missing one byte of every segment
2084 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2085 338499 --sim-hints parsing broken due to wrong order in tokens
2086 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2087 338681 Unable to unwind through clone thread created on i386-linux
2088 338698 race condition between gdbsrv and vgdb on startup
2089 338703 helgrind on arm-linux gets false positives in dynamic loader
2090 338791 alt dwz files can be relative of debug/main file
2091 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2092 338932 build V-trunk with gcc-trunk
2093 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2094 345079 Fix build problems in VEX/useful/test_main.c
2095 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2096 n-i-bz s390x: Fix memory corruption for multithreaded applications
2097 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2098 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2099 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2100 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2101 n-i-bz Add command-line option --default-suppressions
2102 n-i-bz Add support for BLKDISCARDZEROES ioctl
2103 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2104 n-i-bz Add support for sys_pivot_root and sys_unshare
2106 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2107 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2108 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2112 Release 3.9.0 (31 October 2013)
2113 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2114 3.9.0 is a feature release with many improvements and the usual
2115 collection of bug fixes.
2117 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2118 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2119 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2120 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2122 * ================== PLATFORM CHANGES =================
2124 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2125 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2127 * Support for MIPS DSP ASE on MIPS32 platforms.
2129 * Support for s390x Decimal Floating Point instructions on hosts that
2130 have the DFP facility installed.
2132 * Support for POWER8 (Power ISA 2.07) instructions
2134 * Support for Intel AVX2 instructions. This is available only on 64
2137 * Initial support for Intel Transactional Synchronization Extensions,
2140 * Initial support for Hardware Transactional Memory on POWER.
2142 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2143 run large GUI apps tolerably well.
2145 * ==================== TOOL CHANGES ====================
2149 - Improvements in handling of vectorised code, leading to
2150 significantly fewer false error reports. You need to use the flag
2151 --partial-loads-ok=yes to get the benefits of these changes.
2153 - Better control over the leak checker. It is now possible to
2154 specify which leak kinds (definite/indirect/possible/reachable)
2155 should be displayed, which should be regarded as errors, and which
2156 should be suppressed by a given leak suppression. This is done
2157 using the options --show-leak-kinds=kind1,kind2,..,
2158 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2159 "match-leak-kinds:" line in suppression entries, respectively.
2161 Note that generated leak suppressions contain this new line and
2162 are therefore more specific than in previous releases. To get the
2163 same behaviour as previous releases, remove the "match-leak-kinds:"
2164 line from generated suppressions before using them.
2166 - Reduced "possible leak" reports from the leak checker by the use
2167 of better heuristics. The available heuristics provide detection
2168 of valid interior pointers to std::stdstring, to new[] allocated
2169 arrays with elements having destructors and to interior pointers
2170 pointing to an inner part of a C++ object using multiple
2171 inheritance. They can be selected individually using the
2172 option --leak-check-heuristics=heur1,heur2,...
2174 - Better control of stacktrace acquisition for heap-allocated
2175 blocks. Using the --keep-stacktraces option, it is possible to
2176 control independently whether a stack trace is acquired for each
2177 allocation and deallocation. This can be used to create better
2178 "use after free" errors or to decrease Valgrind's resource
2179 consumption by recording less information.
2181 - Better reporting of leak suppression usage. The list of used
2182 suppressions (shown when the -v option is given) now shows, for
2183 each leak suppressions, how many blocks and bytes it suppressed
2184 during the last leak search.
2188 - False errors resulting from the use of statically initialised
2189 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2192 - False errors resulting from the use of pthread_cond_waits that
2193 timeout, have been removed.
2195 * ==================== OTHER CHANGES ====================
2197 * Some attempt to tune Valgrind's space requirements to the expected
2198 capabilities of the target:
2200 - The default size of the translation cache has been reduced from 8
2201 sectors to 6 on Android platforms, since each sector occupies
2202 about 40MB when using Memcheck.
2204 - The default size of the translation cache has been increased to 16
2205 sectors on all other platforms, reflecting the fact that large
2206 applications require instrumentation and storage of huge amounts
2207 of code. For similar reasons, the number of memory mapped
2208 segments that can be tracked has been increased by a factor of 6.
2210 - In all cases, the maximum number of sectors in the translation
2211 cache can be controlled by the new flag --num-transtab-sectors.
2213 * Changes in how debug info (line numbers, etc) is read:
2215 - Valgrind no longer temporarily mmaps the entire object to read
2216 from it. Instead, reading is done through a small fixed sized
2217 buffer. This avoids virtual memory usage spikes when Valgrind
2218 reads debuginfo from large shared objects.
2220 - A new experimental remote debug info server. Valgrind can read
2221 debug info from a different machine (typically, a build host)
2222 where debuginfo objects are stored. This can save a lot of time
2223 and hassle when running Valgrind on resource-constrained targets
2224 (phones, tablets) when the full debuginfo objects are stored
2225 somewhere else. This is enabled by the --debuginfo-server=
2228 - Consistency checking between main and debug objects can be
2229 disabled using the --allow-mismatched-debuginfo option.
2231 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2232 scanning can recover stack traces in some cases when the normal
2233 unwind mechanisms fail. Stack scanning is best described as "a
2234 nasty, dangerous and misleading hack" and so is disabled by default.
2235 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2238 * Detection and merging of recursive stack frame cycles. When your
2239 program has recursive algorithms, this limits the memory used by
2240 Valgrind for recorded stack traces and avoids recording
2241 uninteresting repeated calls. This is controlled by the command
2242 line option --merge-recursive-frame and by the monitor command
2243 "v.set merge-recursive-frames".
2245 * File name and line numbers for used suppressions. The list of used
2246 suppressions (shown when the -v option is given) now shows, for each
2247 used suppression, the file name and line number where the suppression
2250 * New and modified GDB server monitor features:
2252 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2253 that can be used to execute gdbserver monitor commands from the
2256 - A new monitor command, "v.info open_fds", that gives the list of
2257 open file descriptors and additional details.
2259 - An optional message in the "v.info n_errs_found" monitor command,
2260 for example "v.info n_errs_found test 1234 finished", allowing a
2261 comment string to be added to the process output, perhaps for the
2262 purpose of separating errors of different tests or test phases.
2264 - A new monitor command "v.info execontext" that shows information
2265 about the stack traces recorded by Valgrind.
2267 - A new monitor command "v.do expensive_sanity_check_general" to run
2268 some internal consistency checks.
2270 * New flag --sigill-diagnostics to control whether a diagnostic
2271 message is printed when the JIT encounters an instruction it can't
2272 translate. The actual behavior -- delivery of SIGILL to the
2273 application -- is unchanged.
2275 * The maximum amount of memory that Valgrind can use on 64 bit targets
2276 has been increased from 32GB to 64GB. This should make it possible
2277 to run applications on Memcheck that natively require up to about 35GB.
2279 * ==================== FIXED BUGS ====================
2281 The following bugs have been fixed or resolved. Note that "n-i-bz"
2282 stands for "not in bugzilla" -- that is, a bug that was reported to us
2283 but never got a bugzilla entry. We encourage you to file bugs in
2284 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2285 than mailing the developers (or mailing lists) directly -- bugs that
2286 are not entered into bugzilla tend to get forgotten about or ignored.
2288 To see details of a given bug, visit
2289 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2290 where XXXXXX is the bug number as listed below.
2292 123837 system call: 4th argument is optional, depending on cmd
2293 135425 memcheck should tell you where Freed blocks were Mallocd
2294 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2295 207815 Adds some of the drm ioctls to syswrap-linux.c
2296 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2297 252955 Impossible to compile with ccache
2298 253519 Memcheck reports auxv pointer accesses as invalid reads.
2299 263034 Crash when loading some PPC64 binaries
2300 269599 Increase deepest backtrace
2301 274695 s390x: Support "compare to/from logical" instructions (z196)
2302 275800 s390x: Autodetect cache info (part 2)
2303 280271 Valgrind reports possible memory leaks on still-reachable std::string
2304 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2305 289578 Backtraces with ARM unwind tables (stack scan flags)
2306 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2307 304832 ppc32: build failure
2308 305431 Use find_buildid shdr fallback for separate .debug files
2309 305728 Add support for AVX2 instructions
2310 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2311 306035 s390x: Fix IR generation for LAAG and friends
2312 306054 s390x: Condition code computation for convert-to-int/logical
2313 306098 s390x: alternate opcode form for convert to/from fixed
2314 306587 Fix cache line detection from auxiliary vector for PPC.
2315 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2316 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2317 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2318 307101 sys_capget second argument can be NULL
2319 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2320 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2321 307113 s390x: DFP support
2322 307141 valgrind does't work in mips-linux system
2323 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2324 307285 x86_amd64 feature test for avx in test suite is wrong
2325 307290 memcheck overlap testcase needs memcpy version filter
2326 307463 Please add "&limit=0" to the "all open bugs" link
2327 307465 --show-possibly-lost=no should reduce the error count / exit code
2328 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2329 307729 pkgconfig support broken valgrind.pc
2330 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2331 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2332 308089 Unhandled syscall on ppc64: prctl
2333 308135 PPC32 MPC8xx has 16 bytes cache size
2334 308321 testsuite memcheck filter interferes with gdb_filter
2336 308341 vgdb should report process exit (or fatal signal)
2337 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2338 308495 Remove build dependency on installed Xen headers
2339 308573 Internal error on 64-bit instruction executed in 32-bit mode
2341 308627 pmovmskb validity bit propagation is imprecise
2342 308644 vgdb command for having the info for the track-fds option
2343 308711 give more info about aspacemgr and arenas in out_of_memory
2344 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2345 308718 ARM implement SMLALBB family of instructions
2346 308886 Missing support for PTRACE_SET/GETREGSET
2347 308930 syscall name_to_handle_at (303 on amd64) not handled
2348 309229 V-bit tester does not report number of tests generated
2349 309323 print unrecognized instuction on MIPS
2350 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2351 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2352 309430 Self hosting ppc64 encounters a vassert error on operand type
2353 309600 valgrind is a bit confused about 0-sized sections
2354 309823 Generate errors for still reachable blocks
2355 309921 PCMPISTRI validity bit propagation is imprecise
2356 309922 none/tests/ppc64/test_dfp5 sometimes fails
2357 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2358 310424 --read-var-info does not properly describe static variables
2359 310792 search additional path for debug symbols
2360 310931 s390x: Message-security assist (MSA) instruction extension [..]
2361 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2362 311318 ARM: "128-bit constant is not implemented" error message
2363 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2364 311690 V crashes because it redirects branches inside of a redirected function
2365 311880 x86_64: make regtest hangs at shell_valid1
2366 311922 WARNING: unhandled syscall: 170
2368 312171 ppc: insn selection for DFP
2369 312571 Rounding mode call wrong for the DFP Iops [..]
2370 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2371 312913 Dangling pointers error should also report the alloc stack trace
2372 312980 Building on Mountain Lion generates some compiler warnings
2373 313267 Adding MIPS64/Linux port to Valgrind
2376 313811 Buffer overflow in assert_fail
2377 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2378 314269 ppc: dead code in insn selection
2379 314718 ARM: implement integer divide instruction (sdiv and udiv)
2380 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2381 315441 sendmsg syscall should ignore unset msghdr msg_flags
2382 315534 msgrcv inside a thread causes valgrind to hang (block)
2383 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2384 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2385 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2386 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2387 316144 valgrind.1 manpage contains unknown ??? strings [..]
2388 316145 callgrind command line options in manpage reference (unknown) [..]
2389 316145 callgrind command line options in manpage reference [..]
2390 316181 drd: Fixed a 4x slowdown for certain applications
2391 316503 Valgrind does not support SSE4 "movntdqa" instruction
2392 316535 Use of |signed int| instead of |size_t| in valgrind messages
2393 316696 fluidanimate program of parsec 2.1 stuck
2394 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2395 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2396 317186 "Impossible happens" when occurs VCVT instruction on ARM
2397 317318 Support for Threading Building Blocks "scalable_malloc"
2398 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2399 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2400 317463 bmi testcase IR SANITY CHECK FAILURE
2401 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2402 318050 libmpiwrap fails to compile with out-of-source build
2403 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2404 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2405 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2406 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2407 318932 Add missing PPC64 and PPC32 system call support
2408 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2409 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2410 319494 VEX Makefile-gcc standalone build update after r2702
2411 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2412 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2413 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2414 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2415 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2416 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2417 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2418 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2419 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2420 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2421 320895 add fanotify support (patch included)
2422 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2423 321065 Valgrind updates for Xen 4.3
2424 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2425 321363 Unhandled instruction: SSAX (ARM + Thumb)
2426 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2427 321466 Unhandled instruction: SHASX (ARM + Thumb)
2428 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2429 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2430 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2431 321620 Unhandled instruction: UASX (ARM + Thumb)
2432 321621 Unhandled instruction: USAX (ARM + Thumb)
2433 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2434 321693 Unhandled instruction: LDRSBT (Thumb)
2435 321694 Unhandled instruction: UQASX (ARM + Thumb)
2436 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2437 321697 Unhandled instruction: UHASX (ARM + Thumb)
2438 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2439 321704 Unhandled instruction: REVSH (ARM + Thumb)
2440 321730 Add cg_diff and cg_merge man pages
2441 321738 Add vgdb and valgrind-listener man pages
2443 321891 Unhandled instruction: LDRHT (Thumb)
2444 321960 pthread_create() then alloca() causing invalid stack write errors
2445 321969 ppc32 and ppc64 don't support [lf]setxattr
2446 322254 Show threadname together with tid if set by application
2447 322294 Add initial support for IBM Power ISA 2.07
2448 322368 Assertion failure in wqthread_hijack under OS X 10.8
2449 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2450 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2451 322851 0bXXX binary literal syntax is not standard
2452 323035 Unhandled instruction: LDRSHT(Thumb)
2453 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2454 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2455 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2456 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2457 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2458 323437 Phase 2 support for IBM Power ISA 2.07
2459 323713 Support mmxext (integer sse) subset on i386 (athlon)
2460 323803 Transactional memory instructions are not supported for Power
2461 323893 SSE3 not available on amd cpus in valgrind
2462 323905 Probable false positive from Valgrind/drd on close()
2463 323912 valgrind.h header isn't compatible for mingw64
2464 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
2465 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
2466 324181 mmap does not handle MAP_32BIT
2467 324227 memcheck false positive leak when a thread calls exit+block [..]
2468 324421 Support for fanotify API on ARM architecture
2469 324514 gdbserver monitor cmd output behaviour consistency [..]
2470 324518 ppc64: Emulation of dcbt instructions does not handle [..]
2471 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
2472 324582 When access is made to freed memory, report both allocation [..]
2473 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
2474 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
2475 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
2476 324834 Unhandled instructions in Microsoft C run-time for x86_64
2477 324894 Phase 3 support for IBM Power ISA 2.07
2478 326091 drd: Avoid false race reports from optimized strlen() impls
2479 326113 valgrind libvex hwcaps error on AMD64
2480 n-i-bz Some wrong command line options could be ignored
2481 n-i-bz patch to allow fair-sched on android
2482 n-i-bz report error for vgdb snapshot requested before execution
2483 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
2485 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
2489 Release 3.8.1 (19 September 2012)
2490 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2491 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
2492 that occur moderately frequently in real use cases, adds support for
2493 some missing instructions on ARM, and fixes a deadlock condition on
2494 MacOSX. If you package or deliver 3.8.0 for others to use, you might
2495 want to consider upgrading to 3.8.1 instead.
2497 The following bugs have been fixed or resolved. Note that "n-i-bz"
2498 stands for "not in bugzilla" -- that is, a bug that was reported to us
2499 but never got a bugzilla entry. We encourage you to file bugs in
2500 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2501 than mailing the developers (or mailing lists) directly -- bugs that
2502 are not entered into bugzilla tend to get forgotten about or ignored.
2504 To see details of a given bug, visit
2505 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2506 where XXXXXX is the bug number as listed below.
2509 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
2510 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
2511 298281 wcslen causes false(?) uninitialised value warnings
2512 301281 valgrind hangs on OS X when the process calls system()
2513 304035 disInstr(arm): unhandled instruction 0xE1023053
2514 304867 implement MOVBE instruction in x86 mode
2515 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
2516 305042 amd64: implement 0F 7F encoding of movq between two registers
2517 305199 ARM: implement QDADD and QDSUB
2518 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
2519 305513 killed by fatal signal: SIGSEGV
2520 305690 DRD reporting invalid semaphore when sem_trywait fails
2521 305926 Invalid alignment checks for some AVX instructions
2522 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
2523 306310 3.8.0 release tarball missing some files
2524 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
2525 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
2526 n-i-bz shmat of a segment > 4Gb does not work
2527 n-i-bz simulate_control_c script wrong USR1 signal number on mips
2528 n-i-bz vgdb ptrace calls wrong on mips [...]
2529 n-i-bz Fixes for more MPI false positives
2530 n-i-bz exp-sgcheck's memcpy causes programs to segfault
2531 n-i-bz OSX build w/ clang: asserts at startup
2532 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
2533 n-i-bz fix a couple of union tag-vs-field mixups
2534 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
2536 The following bugs were fixed in 3.8.0 but not listed in this NEWS
2539 254088 Valgrind should know about UD2 instruction
2542 304754 NEWS blows TeX's little mind
2544 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
2548 Release 3.8.0 (10 August 2012)
2549 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2550 3.8.0 is a feature release with many improvements and the usual
2551 collection of bug fixes.
2553 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2554 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
2555 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
2556 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
2557 There is initial support for MacOSX 10.8, but it is not usable for
2558 serious work at present.
2560 * ================== PLATFORM CHANGES =================
2562 * Support for MIPS32 platforms running Linux. Valgrind has been
2563 tested on MIPS32 and MIPS32r2 platforms running different Debian
2564 Squeeze and MeeGo distributions. Both little-endian and big-endian
2565 cores are supported. The tools Memcheck, Massif and Lackey have
2566 been tested and are known to work. See README.mips for more details.
2568 * Preliminary support for Android running on x86.
2570 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
2572 * Support for Intel AVX instructions and for AES instructions. This
2573 support is available only for 64 bit code.
2575 * Support for POWER Decimal Floating Point instructions.
2577 * ==================== TOOL CHANGES ====================
2579 * Non-libc malloc implementations are now supported. This is useful
2580 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
2581 Using the new option --soname-synonyms, such tools can be informed
2582 that the malloc implementation is either linked statically into the
2583 executable, or is present in some other shared library different
2584 from libc.so. This makes it possible to process statically linked
2585 programs, and programs using other malloc libraries, for example
2586 TCMalloc or JEMalloc.
2588 * For tools that provide their own replacement for malloc et al, the
2589 option --redzone-size=<number> allows users to specify the size of
2590 the padding blocks (redzones) added before and after each client
2591 allocated block. Smaller redzones decrease the memory needed by
2592 Valgrind. Bigger redzones increase the chance to detect blocks
2593 overrun or underrun. Prior to this change, the redzone size was
2594 hardwired to 16 bytes in Memcheck.
2598 - The leak_check GDB server monitor command now can
2599 control the maximum nr of loss records to output.
2601 - Reduction of memory use for applications allocating
2602 many blocks and/or having many partially defined bytes.
2604 - Addition of GDB server monitor command 'block_list' that lists
2605 the addresses/sizes of the blocks of a leak search loss record.
2607 - Addition of GDB server monitor command 'who_points_at' that lists
2608 the locations pointing at a block.
2610 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2611 detect an invalid access of these redzones, by marking them
2612 noaccess. Similarly, if a redzone size is given for a memory
2613 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2614 This still allows to find some bugs if the user has forgotten to
2615 mark the pool superblock noaccess.
2617 - Performance of memory leak check has been improved, especially in
2618 cases where there are many leaked blocks and/or many suppression
2619 rules used to suppress leak reports.
2621 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2622 more precise analysis, which is important for LLVM/Clang
2623 generated code. This is at the cost of somewhat reduced
2624 performance. Note there is no change to analysis precision or
2625 costs on Linux targets.
2629 - Added even more facilities that can help finding the cause of a data
2630 race, namely the command-line option --ptrace-addr and the macro
2631 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2633 - Fixed a subtle bug that could cause false positive data race reports.
2635 * ==================== OTHER CHANGES ====================
2637 * The C++ demangler has been updated so as to work well with C++
2638 compiled by up to at least g++ 4.6.
2640 * Tool developers can make replacement/wrapping more flexible thanks
2641 to the new option --soname-synonyms. This was reported above, but
2642 in fact is very general and applies to all function
2643 replacement/wrapping, not just to malloc-family functions.
2645 * Round-robin scheduling of threads can be selected, using the new
2646 option --fair-sched= yes. Prior to this change, the pipe-based
2647 thread serialisation mechanism (which is still the default) could
2648 give very unfair scheduling. --fair-sched=yes improves
2649 responsiveness of interactive multithreaded applications, and
2650 improves repeatability of results from the thread checkers Helgrind
2653 * For tool developers: support to run Valgrind on Valgrind has been
2654 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2656 * gdbserver now shows the float shadow registers as integer
2657 rather than float values, as the shadow values are mostly
2658 used as bit patterns.
2660 * Increased limit for the --num-callers command line flag to 500.
2662 * Performance improvements for error matching when there are many
2663 suppression records in use.
2665 * Improved support for DWARF4 debugging information (bug 284184).
2667 * Initial support for DWZ compressed Dwarf debug info.
2669 * Improved control over the IR optimiser's handling of the tradeoff
2670 between performance and precision of exceptions. Specifically,
2671 --vex-iropt-precise-memory-exns has been removed and replaced by
2672 --vex-iropt-register-updates, with extended functionality. This
2673 allows the Valgrind gdbserver to always show up to date register
2676 * Modest performance gains through the use of translation chaining for
2679 * ==================== FIXED BUGS ====================
2681 The following bugs have been fixed or resolved. Note that "n-i-bz"
2682 stands for "not in bugzilla" -- that is, a bug that was reported to us
2683 but never got a bugzilla entry. We encourage you to file bugs in
2684 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2685 than mailing the developers (or mailing lists) directly -- bugs that
2686 are not entered into bugzilla tend to get forgotten about or ignored.
2688 To see details of a given bug, visit
2689 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2690 where XXXXXX is the bug number as listed below.
2692 197914 Building valgrind from svn now requires automake-1.10
2693 203877 increase to 16Mb maximum allowed alignment for memalign et al
2694 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
2695 247386 make perf does not run all performance tests
2696 270006 Valgrind scheduler unfair
2697 270777 Adding MIPS/Linux port to Valgrind
2698 270796 s390x: Removed broken support for the TS insn
2699 271438 Fix configure for proper SSE4.2 detection
2700 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
2701 273475 Add support for AVX instructions
2702 274078 improved configure logic for mpicc
2703 276993 fix mremap 'no thrash checks'
2704 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
2705 281482 memcheck incorrect byte allocation count in realloc() for silly argument
2706 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
2707 283413 Fix wrong sanity check
2708 283671 Robustize alignment computation in LibVEX_Alloc
2709 283961 Adding support for some HCI IOCTLs
2710 284124 parse_type_DIE: confused by: DWARF 4
2711 284864 == 273475 (Add support for AVX instructions)
2712 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
2713 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
2714 285725 == 273475 (Add support for AVX instructions)
2715 286261 add wrapper for linux I2C_RDWR ioctl
2716 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
2717 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
2718 286384 configure fails "checking for a supported version of gcc"
2719 286497 == 273475 (Add support for AVX instructions)
2720 286596 == 273475 (Add support for AVX instructions)
2721 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
2722 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
2723 287260 Incorrect conditional jump or move depends on uninitialised value(s)
2724 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
2725 287307 == 273475 (Add support for AVX instructions)
2726 287858 VG_(strerror): unknown error
2727 288298 (MacOSX) unhandled syscall shm_unlink
2728 288995 == 273475 (Add support for AVX instructions)
2729 289470 Loading of large Mach-O thin binaries fails.
2730 289656 == 273475 (Add support for AVX instructions)
2731 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
2732 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
2733 289839 s390x: Provide support for unicode conversion instructions
2734 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
2735 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
2736 290655 Add support for AESKEYGENASSIST instruction
2737 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
2738 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
2739 291253 ES register not initialised in valgrind simulation
2740 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
2741 291865 s390x: Support the "Compare Double and Swap" family of instructions
2742 292300 == 273475 (Add support for AVX instructions)
2743 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
2744 292493 == 273475 (Add support for AVX instructions)
2745 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
2746 292627 Missing support for some SCSI ioctls
2747 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
2748 292841 == 273475 (Add support for AVX instructions)
2749 292993 implement the getcpu syscall on amd64-linux
2750 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
2751 293088 Add some VEX sanity checks for ppc64 unhandled instructions
2752 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
2753 293754 PCMPxSTRx not implemented for 16-bit characters
2754 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
2755 293808 CLFLUSH not supported by latest VEX for amd64
2756 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
2757 294048 MPSADBW instruction not implemented
2758 294055 regtest none/tests/shell fails when locale is not set to C
2759 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
2760 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
2761 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
2762 294260 disInstr_AMD64: disInstr miscalculated next %rip
2763 294523 --partial-loads-ok=yes causes false negatives
2764 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
2765 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
2766 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
2767 295089 can not annotate source for both helgrind and drd
2768 295221 POWER Processor decimal floating point instruction support missing
2769 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
2770 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
2771 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
2772 295617 ARM - Add some missing syscalls
2773 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
2774 296229 Linux user input device ioctls missing wrappers
2775 296318 ELF Debug info improvements (more than one rx/rw mapping)
2776 296422 Add translation chaining support
2777 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
2778 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
2779 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
2780 297078 gdbserver signal handling problems [..]
2781 297147 drd false positives on newly allocated memory
2782 297329 disallow decoding of IBM Power DFP insns on some machines
2783 297497 POWER Processor decimal floating point instruction support missing
2784 297701 Another alias for strncasecmp_l in libc-2.13.so
2785 297911 'invalid write' not reported when using APIs for custom mem allocators.
2786 297976 s390x: revisit EX implementation
2787 297991 Valgrind interferes with mmap()+ftell()
2788 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
2789 297993 Fix compilation of valgrind with gcc -g3.
2790 298080 POWER Processor DFP support missing, part 3
2791 298227 == 273475 (Add support for AVX instructions)
2792 298335 == 273475 (Add support for AVX instructions)
2793 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
2794 298394 s390x: Don't bail out on an unknown machine model. [..]
2795 298421 accept4() syscall (366) support is missing for ARM
2796 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
2797 298732 valgrind installation problem in ubuntu with kernel version 3.x
2798 298862 POWER Processor DFP instruction support missing, part 4
2799 298864 DWARF reader mis-parses DW_FORM_ref_addr
2800 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
2801 299053 Support DWARF4 DW_AT_high_pc constant form
2802 299104 == 273475 (Add support for AVX instructions)
2803 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
2804 299629 dup3() syscall (358) support is missing for ARM
2805 299694 POWER Processor DFP instruction support missing, part 5
2806 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
2807 299803 == 273475 (Add support for AVX instructions)
2808 299804 == 273475 (Add support for AVX instructions)
2809 299805 == 273475 (Add support for AVX instructions)
2810 300140 ARM - Missing (T1) SMMUL
2811 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
2812 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
2813 300414 FCOM and FCOMP unimplemented for amd64 guest
2814 301204 infinite loop in canonicaliseSymtab with ifunc symbol
2815 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
2816 301265 add x86 support to Android build
2817 301984 configure script doesn't detect certain versions of clang
2818 302205 Fix compiler warnings for POWER VEX code and POWER test cases
2819 302287 Unhandled movbe instruction on Atom processors
2820 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
2821 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
2822 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
2823 302656 == 273475 (Add support for AVX instructions)
2824 302709 valgrind for ARM needs extra tls support for android emulator [..]
2825 302827 add wrapper for CDROM_GET_CAPABILITY
2826 302901 Valgrind crashes with dwz optimized debuginfo
2827 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
2828 303116 Add support for the POWER instruction popcntb
2829 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
2830 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
2831 303466 == 273475 (Add support for AVX instructions)
2832 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
2833 303963 strstr() function produces wrong results under valgrind callgrind
2834 304054 CALL_FN_xx macros need to enforce stack alignment
2835 304561 tee system call not supported
2836 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
2837 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
2838 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
2839 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
2840 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
2841 n-i-bz s390x: Shadow registers can now be examined using vgdb
2843 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
2844 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
2848 Release 3.7.0 (5 November 2011)
2849 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2850 3.7.0 is a feature release with many significant improvements and the
2851 usual collection of bug fixes.
2853 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2854 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
2855 Support for recent distros and toolchain components (glibc 2.14, gcc
2856 4.6, MacOSX 10.7) has been added.
2858 * ================== PLATFORM CHANGES =================
2860 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
2861 analyse 64-bit programs running on z/Architecture. Most user space
2862 instructions up to and including z10 are supported. Valgrind has
2863 been tested extensively on z9, z10, and z196 machines running SLES
2864 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
2865 known to work well. Callgrind, Helgrind, and DRD work reasonably
2866 well on z9 and later models. See README.s390 for more details.
2868 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
2869 64-bit processes are supported. Some complex threaded applications
2870 (Firefox) are observed to hang when run as 32 bit applications,
2871 whereas 64-bit versions run OK. The cause is unknown. Memcheck
2872 will likely report some false errors. In general, expect some rough
2873 spots. This release also supports MacOSX 10.6, but drops support
2876 * Preliminary support for Android (on ARM). Valgrind can now run
2877 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
2878 README.android for more details, plus instructions on how to get
2881 * Support for the IBM Power ISA 2.06 (Power7 instructions)
2883 * General correctness and performance improvements for ARM/Linux, and,
2884 by extension, ARM/Android.
2886 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
2887 instruction set support is under development but is not available in
2890 * Support for AIX5 has been removed.
2892 * ==================== TOOL CHANGES ====================
2894 * Memcheck: some incremental changes:
2896 - reduction of memory use in some circumstances
2898 - improved handling of freed memory, which in some circumstances
2899 can cause detection of use-after-free that would previously have
2902 - fix of a longstanding bug that could cause false negatives (missed
2903 errors) in programs doing vector saturated narrowing instructions.
2905 * Helgrind: performance improvements and major memory use reductions,
2906 particularly for large, long running applications which perform many
2907 synchronisation (lock, unlock, etc) events. Plus many smaller
2910 - display of locksets for both threads involved in a race
2912 - general improvements in formatting/clarity of error messages
2914 - addition of facilities and documentation regarding annotation
2915 of thread safe reference counted C++ classes
2917 - new flag --check-stack-refs=no|yes [yes], to disable race checking
2918 on thread stacks (a performance hack)
2920 - new flag --free-is-write=no|yes [no], to enable detection of races
2921 where one thread accesses heap memory but another one frees it,
2922 without any coordinating synchronisation event
2924 * DRD: enabled XML output; added support for delayed thread deletion
2925 in order to detect races that occur close to the end of a thread
2926 (--join-list-vol); fixed a memory leak triggered by repeated client
2927 memory allocatation and deallocation; improved Darwin support.
2929 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
2931 * exp-sgcheck: this tool has been reduced in scope so as to improve
2932 performance and remove checking that Memcheck does better.
2933 Specifically, the ability to check for overruns for stack and global
2934 arrays is unchanged, but the ability to check for overruns of heap
2935 blocks has been removed. The tool has accordingly been renamed to
2936 exp-sgcheck ("Stack and Global Array Checking").
2938 * ==================== OTHER CHANGES ====================
2940 * GDB server: Valgrind now has an embedded GDB server. That means it
2941 is possible to control a Valgrind run from GDB, doing all the usual
2942 things that GDB can do (single stepping, breakpoints, examining
2943 data, etc). Tool-specific functionality is also available. For
2944 example, it is possible to query the definedness state of variables
2945 or memory from within GDB when running Memcheck; arbitrarily large
2946 memory watchpoints are supported, etc. To use the GDB server, start
2947 Valgrind with the flag --vgdb-error=0 and follow the on-screen
2950 * Improved support for unfriendly self-modifying code: a new option
2951 --smc-check=all-non-file is available. This adds the relevant
2952 consistency checks only to code that originates in non-file-backed
2953 mappings. In effect this confines the consistency checking only to
2954 code that is or might be JIT generated, and avoids checks on code
2955 that must have been compiled ahead of time. This significantly
2956 improves performance on applications that generate code at run time.
2958 * It is now possible to build a working Valgrind using Clang-2.9 on
2961 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
2962 These enable and disable error reporting on a per-thread, and
2963 nestable, basis. This is useful for hiding errors in particularly
2964 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
2965 now uses this facility.
2967 * Added the --mod-funcname option to cg_diff.
2969 * ==================== FIXED BUGS ====================
2971 The following bugs have been fixed or resolved. Note that "n-i-bz"
2972 stands for "not in bugzilla" -- that is, a bug that was reported to us
2973 but never got a bugzilla entry. We encourage you to file bugs in
2974 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2975 mailing the developers (or mailing lists) directly -- bugs that are
2976 not entered into bugzilla tend to get forgotten about or ignored.
2978 To see details of a given bug, visit
2979 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2980 where XXXXXX is the bug number as listed below.
2982 79311 malloc silly arg warning does not give stack trace
2983 210935 port valgrind.h (not valgrind) to win32 to support client requests
2984 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
2985 243404 Port to zSeries
2986 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
2987 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
2988 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
2989 253206 Some fixes for the faultstatus testcase
2990 255223 capget testcase fails when running as root
2991 256703 xlc_dbl_u32.c testcase broken
2992 256726 Helgrind tests have broken inline asm
2993 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
2994 264800 testcase compile failure on zseries
2995 265762 make public VEX headers compilable by G++ 3.x
2996 265771 assertion in jumps.c (r11523) fails with glibc-2.3
2997 266753 configure script does not give the user the option to not use QtCore
2998 266931 gen_insn_test.pl is broken
2999 266961 ld-linux.so.2 i?86-linux strlen issues
3000 266990 setns instruction causes false positive
3001 267020 Make directory for temporary files configurable at run-time.
3002 267342 == 267997 (segmentation fault on Mac OS 10.6)
3003 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3004 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3005 267488 regtest: darwin support for 64-bit build
3006 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3007 267630 Add support for IBM Power ISA 2.06 -- stage 1
3008 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3009 267819 Add client request for informing the core about reallocation
3010 267925 laog data structure quadratic for a single sequence of lock
3011 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3012 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3013 268513 missed optimizations in fold_Expr
3014 268619 s390x: fpr - gpr transfer facility
3015 268620 s390x: reconsider "long displacement" requirement
3016 268621 s390x: improve IR generation for XC
3017 268715 s390x: FLOGR is not universally available
3018 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3019 268930 s390x: MHY is not universally available
3020 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3021 269079 Support ptrace system call on ARM
3022 269144 missing "Bad option" error message
3023 269209 conditional load and store facility (z196)
3024 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3025 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3026 269736 s390x: minor code generation tweaks
3027 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3028 269863 s390x: remove unused function parameters
3029 269864 s390x: tweak s390_emit_load_cc
3030 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3031 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3032 270115 s390x: rewrite some testcases
3033 270309 == 267997 (valgrind crash on startup)
3034 270320 add support for Linux FIOQSIZE ioctl() call
3035 270326 segfault while trying to sanitize the environment passed to execle
3036 270794 IBM POWER7 support patch causes regression in none/tests
3037 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3038 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3039 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3040 270959 s390x: invalid use of R0 as base register
3041 271042 VSX configure check fails when it should not
3042 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3043 271259 s390x: fix code confusion
3044 271337 == 267997 (Valgrind segfaults on MacOS X)
3045 271385 s390x: Implement Ist_MBE
3046 271501 s390x: misc cleanups
3047 271504 s390x: promote likely and unlikely
3048 271579 ppc: using wrong enum type
3049 271615 unhandled instruction "popcnt" (arch=amd10h)
3050 271730 Fix bug when checking ioctls: duplicate check
3051 271776 s390x: provide STFLE instruction support
3052 271779 s390x: provide clock instructions like STCK
3053 271799 Darwin: ioctls without an arg report a memory error
3054 271820 arm: fix type confusion
3055 271917 pthread_cond_timedwait failure leads to not-locked false positive
3056 272067 s390x: fix DISP20 macro
3057 272615 A typo in debug output in mc_leakcheck.c
3058 272661 callgrind_annotate chokes when run from paths containing regex chars
3059 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3060 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3061 272967 make documentation build-system more robust
3062 272986 Fix gcc-4.6 warnings with valgrind.h
3063 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3064 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3065 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3066 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3067 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3068 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3069 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3070 273778 exp-ptrcheck: unhandled sysno == 259
3071 274089 exp-ptrcheck: unhandled sysno == 208
3072 274378 s390x: Various dispatcher tweaks
3073 274447 WARNING: unhandled syscall: 340
3074 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3075 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3076 274926 valgrind does not build against linux-3
3077 275148 configure FAIL with glibc-2.14
3078 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3079 275168 Make Valgrind work for MacOSX 10.7 Lion
3080 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3081 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3082 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3083 275308 Fix implementation for ppc64 fres instruc
3084 275339 s390x: fix testcase compile warnings
3085 275517 s390x: Provide support for CKSM instruction
3086 275710 s390x: get rid of redundant address mode calculation
3087 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3088 275852 == 250101 (valgrind uses all swap space and is killed)
3089 276784 Add support for IBM Power ISA 2.06 -- stage 3
3090 276987 gdbsrv: fix tests following recent commits
3091 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3092 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3093 277471 Unhandled syscall: 340
3094 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3095 277653 ARM: support Thumb2 PLD instruction
3096 277663 ARM: NEON float VMUL by scalar incorrect
3097 277689 ARM: tests for VSTn with register post-index are broken
3098 277694 ARM: BLX LR instruction broken in ARM mode
3099 277780 ARM: VMOV.F32 (immediate) instruction is broken
3100 278057 fuse filesystem syscall deadlocks
3101 278078 Unimplemented syscall 280 on ppc32
3102 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3103 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3104 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3105 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3106 279027 Support for MVCL and CLCL instruction
3107 279027 s390x: Provide support for CLCL and MVCL instructions
3108 279062 Remove a redundant check in the insn selector for ppc.
3109 279071 JDK creates PTEST with redundant REX.W prefix
3110 279212 gdbsrv: add monitor cmd v.info scheduler.
3111 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3112 279698 memcheck discards valid-bits for packuswb
3113 279795 memcheck reports uninitialised values for mincore on amd64
3114 279994 Add support for IBM Power ISA 2.06 -- stage 3
3115 280083 mempolicy syscall check errors
3116 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3117 280710 s390x: config files for nightly builds
3118 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3119 280965 Valgrind breaks fcntl locks when program does mmap
3120 281138 WARNING: unhandled syscall: 340
3121 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3122 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3123 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3124 281468 s390x: handle do_clone and gcc clones in call traces
3125 281488 ARM: VFP register corruption
3126 281828 == 275284 (false memmove warning: "Source and destination overlap")
3127 281883 s390x: Fix system call wrapper for "clone".
3128 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3129 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3130 282238 SLES10: make check fails
3131 282979 strcasestr needs replacement with recent(>=2.12) glibc
3132 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3133 283243 Regression in ppc64 memcheck tests
3134 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3135 283427 re-connect epoll_pwait syscall on ARM linux
3136 283600 gdbsrv: android: port vgdb.c
3137 283709 none/tests/faultstatus needs to account for page size
3138 284305 filter_gdb needs enhancement to work on ppc64
3139 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3140 284472 Thumb2 ROR.W encoding T2 not implemented
3141 284621 XML-escape process command line in XML output
3142 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3143 that have non-power-of-2 sizes (also AMDs)
3144 n-i-bz don't be spooked by libraries mashed by elfhack
3145 n-i-bz don't be spooked by libxul.so linked with gold
3146 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3148 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3149 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3150 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3154 Release 3.6.1 (16 February 2011)
3155 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3156 3.6.1 is a bug fix release. It adds support for some SSE4
3157 instructions that were omitted in 3.6.0 due to lack of time. Initial
3158 support for glibc-2.13 has been added. A number of bugs causing
3159 crashing or assertion failures have been fixed.
3161 The following bugs have been fixed or resolved. Note that "n-i-bz"
3162 stands for "not in bugzilla" -- that is, a bug that was reported to us
3163 but never got a bugzilla entry. We encourage you to file bugs in
3164 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3165 mailing the developers (or mailing lists) directly -- bugs that are
3166 not entered into bugzilla tend to get forgotten about or ignored.
3168 To see details of a given bug, visit
3169 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3170 where XXXXXX is the bug number as listed below.
3172 188572 Valgrind on Mac should suppress setenv() mem leak
3173 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3174 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3175 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3176 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3177 254420 memory pool tracking broken
3178 254957 Test code failing to compile due to changes in memcheck.h
3179 255009 helgrind/drd: crash on chmod with invalid parameter
3180 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3181 255355 helgrind/drd: crash on threaded programs doing fork
3183 255418 (SSE4.x) rint call compiled with ICC
3184 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3185 255888 closing valgrindoutput tag outputted to log-stream on error
3186 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3187 255966 Slowness when using mempool annotations
3188 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3189 256600 super-optimized strcasecmp() false positive
3190 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3191 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3192 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3193 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3194 257276 Missing case in memcheck --track-origins=yes
3195 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3196 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3197 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3198 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3199 263099 callgrind_annotate counts Ir improperly [...]
3200 263877 undefined coprocessor instruction on ARMv7
3201 265964 configure FAIL with glibc-2.13
3202 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3203 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3204 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3205 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3206 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3208 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3212 Release 3.6.0 (21 October 2010)
3213 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3214 3.6.0 is a feature release with many significant improvements and the
3215 usual collection of bug fixes.
3217 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3218 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3219 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3221 -------------------------
3223 Here are some highlights. Details are shown further down:
3225 * Support for ARM/Linux.
3227 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3229 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3231 * Support for the SSE4.2 instruction set.
3233 * Enhancements to the Callgrind profiler, including the ability to
3234 handle CPUs with three levels of cache.
3236 * A new experimental heap profiler, DHAT.
3238 * A huge number of bug fixes and small enhancements.
3240 -------------------------
3242 Here are details of the above changes, together with descriptions of
3243 many other changes, and a list of fixed bugs.
3245 * ================== PLATFORM CHANGES =================
3247 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3248 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3249 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3251 This requires a CPU capable of running the ARMv7-A instruction set
3252 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3253 of the user space instruction set, including ARM and Thumb integer
3254 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3255 Cachegrind and Massif tools work properly; other tools work to
3258 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3259 with support for recent releases of the underlying toolchain
3260 components, notably gcc-4.5 and glibc-2.12.
3262 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3263 support also works much better on OS X 10.5, and is as solid as
3266 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3267 64-bit mode. In 32-bit mode, support is only available up to and
3268 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3269 supported in 64-bit mode, and 32-bit mode does in fact support the
3270 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3271 10.6 on 32-bit targets.
3273 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3274 and including version 2.05 is supported.
3276 * ==================== TOOL CHANGES ====================
3278 * Cachegrind has a new processing script, cg_diff, which finds the
3279 difference between two profiles. It's very useful for evaluating
3280 the performance effects of a change in a program.
3282 Related to this change, the meaning of cg_annotate's (rarely-used)
3283 --threshold option has changed; this is unlikely to affect many
3284 people, if you do use it please see the user manual for details.
3286 * Callgrind now can do branch prediction simulation, similar to
3287 Cachegrind. In addition, it optionally can count the number of
3288 executed global bus events. Both can be used for a better
3289 approximation of a "Cycle Estimation" as derived event (you need to
3290 update the event formula in KCachegrind yourself).
3292 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3293 rather than the L2 cache. This is to accommodate machines with
3294 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3295 cache configuration of such a machine it will run the simulation as
3296 if the L2 cache isn't present. This means the results are less
3297 likely to match the true result for the machine, but
3298 Cachegrind/Callgrind's results are already only approximate, and
3299 should not be considered authoritative. The results are still
3300 useful for giving a general idea about a program's locality.
3302 * Massif has a new option, --pages-as-heap, which is disabled by
3303 default. When enabled, instead of tracking allocations at the level
3304 of heap blocks (as allocated with malloc/new/new[]), it instead
3305 tracks memory allocations at the level of memory pages (as mapped by
3306 mmap, brk, etc). Each mapped page is treated as its own block.
3307 Interpreting the page-level output is harder than the heap-level
3308 output, but this option is useful if you want to account for every
3309 byte of memory used by a program.
3311 * DRD has two new command-line options: --free-is-write and
3312 --trace-alloc. The former allows to detect reading from already freed
3313 memory, and the latter allows tracing of all memory allocations and
3316 * DRD has several new annotations. Custom barrier implementations can
3317 now be annotated, as well as benign races on static variables.
3319 * DRD's happens before / happens after annotations have been made more
3320 powerful, so that they can now also be used to annotate e.g. a smart
3321 pointer implementation.
3323 * Helgrind's annotation set has also been drastically improved, so as
3324 to provide to users a general set of annotations to describe locks,
3325 semaphores, barriers and condition variables. Annotations to
3326 describe thread-safe reference counted heap objects have also been
3329 * Memcheck has a new command-line option, --show-possibly-lost, which
3330 is enabled by default. When disabled, the leak detector will not
3331 show possibly-lost blocks.
3333 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3334 has been added. DHAT keeps track of allocated heap blocks, and also
3335 inspects every memory reference to see which block (if any) is being
3336 accessed. This gives a lot of insight into block lifetimes,
3337 utilisation, turnover, liveness, and the location of hot and cold
3338 fields. You can use DHAT to do hot-field profiling.
3340 * ==================== OTHER CHANGES ====================
3342 * Improved support for unfriendly self-modifying code: the extra
3343 overhead incurred by --smc-check=all has been reduced by
3344 approximately a factor of 5 as compared with 3.5.0.
3346 * Ability to show directory names for source files in error messages.
3347 This is combined with a flexible mechanism for specifying which
3348 parts of the paths should be shown. This is enabled by the new flag
3351 * A new flag, --require-text-symbol, which will stop the run if a
3352 specified symbol is not found it a given shared object when it is
3353 loaded into the process. This makes advanced working with function
3354 intercepting and wrapping safer and more reliable.
3356 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3357 and control of Valgrind is now available for the tools Memcheck and
3358 Helgrind. XML output from Valgrind is available for Memcheck,
3359 Helgrind and exp-Ptrcheck.
3361 * More reliable stack unwinding on amd64-linux, particularly in the
3362 presence of function wrappers, and with gcc-4.5 compiled code.
3364 * Modest scalability (performance improvements) for massive
3365 long-running applications, particularly for those with huge amounts
3368 * Support for analyzing programs running under Wine with has been
3369 improved. The header files <valgrind/valgrind.h>,
3370 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3371 Windows-programs compiled with MinGW or one of the Microsoft Visual
3374 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3375 The 32-bit simulator was not affected. This did not occur often,
3376 but when it did would usually crash the program under test.
3379 * A large number of bugs were fixed. These are shown below.
3381 * A number of bugs were investigated, and were candidates for fixing,
3382 but are not fixed in 3.6.0, due to lack of developer time. They may
3383 get fixed in later releases. They are:
3385 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3386 212419 false positive "lock order violated" (A+B vs A)
3387 213685 Undefined value propagates past dependency breaking instruction
3388 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3389 237920 valgrind segfault on fork failure
3390 242137 support for code compiled by LLVM-2.8
3391 242423 Another unknown Intel cache config value
3392 243232 Inconsistent Lock Orderings report with trylock
3393 243483 ppc: callgrind triggers VEX assertion failure
3394 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3395 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3397 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3398 249435 Analyzing wine programs with callgrind triggers a crash
3399 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3400 250065 Handling large allocations
3401 250101 huge "free" memory usage due to m_mallocfree.c
3402 "superblocks fragmentation"
3403 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3404 252091 Callgrind on ARM does not detect function returns correctly
3405 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3406 254420 memory pool tracking broken
3407 n-i-bz support for adding symbols for JIT generated code
3410 The following bugs have been fixed or resolved. Note that "n-i-bz"
3411 stands for "not in bugzilla" -- that is, a bug that was reported to us
3412 but never got a bugzilla entry. We encourage you to file bugs in
3413 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3414 mailing the developers (or mailing lists) directly -- bugs that are
3415 not entered into bugzilla tend to get forgotten about or ignored.
3417 To see details of a given bug, visit
3418 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3419 where XXXXXX is the bug number as listed below.
3421 135264 dcbzl instruction missing
3423 153699 Valgrind should report unaligned reads with movdqa
3425 190429 Valgrind reports lost of errors in ld.so
3426 with x86_64 2.9.90 glibc
3427 197266 valgrind appears to choke on the xmms instruction
3429 197988 Crash when demangling very large symbol names
3430 202315 unhandled syscall: 332 (inotify_init1)
3431 203256 Add page-level profiling to Massif
3432 205093 dsymutil=yes needs quotes, locking (partial fix)
3433 205241 Snow Leopard 10.6 support (partial fix)
3434 206600 Leak checker fails to upgrade indirect blocks when their
3435 parent becomes reachable
3436 210935 port valgrind.h (not valgrind) to win32 so apps run under
3437 wine can make client requests
3438 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3439 within Linux ip-stack checksum functions
3440 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3442 213685 Undefined value propagates past dependency breaking instruction
3444 215914 Valgrind inserts bogus empty environment variable
3446 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3447 222545 shmat fails under valgind on some arm targets
3448 222560 ARM NEON support
3451 232509 Docs build fails with formatting inside <title></title> elements
3453 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3454 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3455 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3456 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3457 237485 symlink (syscall 57) is not supported on Mac OS
3458 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3460 238208 is_just_below_ESP doesn't take into account red-zone
3461 238345 valgrind passes wrong $0 when executing a shell script
3462 238679 mq_timedreceive syscall doesn't flag the reception buffer
3464 238696 fcntl command F_DUPFD_CLOEXEC not supported
3465 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3466 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3467 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
3469 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
3475 242606 unhandled syscall: setegid (in Ptrcheck)
3476 242814 Helgrind "Impossible has happened" during
3477 QApplication::initInstance();
3478 243064 Valgrind attempting to read debug information from iso
3479 243270 Make stack unwinding in Valgrind wrappers more reliable
3480 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
3481 sysno = 277 (mq_open)
3482 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
3483 244493 ARM VFP d16-d31 registers support
3484 244670 add support for audit_session_self syscall on Mac OS 10.6
3485 244921 The xml report of helgrind tool is not well format
3486 244923 In the xml report file, the <preamble> not escape the
3487 xml char, eg '<','&','>'
3488 245535 print full path names in plain text reports
3489 245925 x86-64 red zone handling problem
3490 246258 Valgrind not catching integer underruns + new [] s
3491 246311 reg/reg cmpxchg doesn't work on amd64
3492 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
3493 246888 Improve Makefile.vex.am
3494 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
3495 to [f]chmod_extended
3496 247526 IBM POWER6 (ISA 2.05) support is incomplete
3497 247561 Some leak testcases fails due to reachable addresses in
3499 247875 sizeofIRType to handle Ity_I128
3500 247894 [PATCH] unhandled syscall sys_readahead
3501 247980 Doesn't honor CFLAGS passed to configure
3502 248373 darwin10.supp is empty in the trunk
3503 248822 Linux FIBMAP ioctl has int parameter instead of long
3504 248893 [PATCH] make readdwarf.c big endianess safe to enable
3505 unwinding on big endian systems
3506 249224 Syscall 336 not supported (SYS_proc_info)
3508 249775 Incorrect scheme for detecting NEON capabilities of host CPU
3509 249943 jni JVM init fails when using valgrind
3510 249991 Valgrind incorrectly declares AESKEYGENASSIST support
3512 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
3513 250799 frexp$fenv_access_off function generates SIGILL
3514 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
3515 251251 support pclmulqdq insn
3516 251362 valgrind: ARM: attach to debugger either fails or provokes
3518 251674 Unhandled syscall 294
3521 254257 Add support for debugfiles found by build-id
3522 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
3523 254646 Wrapped functions cause stack misalignment on OS X
3524 (and possibly Linux)
3525 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
3527 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
3531 Release 3.5.0 (19 August 2009)
3532 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3533 3.5.0 is a feature release with many significant improvements and the
3534 usual collection of bug fixes. The main improvement is that Valgrind
3535 now works on Mac OS X.
3537 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
3538 and X86/Darwin. Support for recent distros and toolchain components
3539 (glibc 2.10, gcc 4.5) has been added.
3541 -------------------------
3543 Here is a short summary of the changes. Details are shown further
3546 * Support for Mac OS X (10.5.x).
3548 * Improvements and simplifications to Memcheck's leak checker.
3550 * Clarification and simplifications in various aspects of Valgrind's
3553 * XML output for Helgrind and Ptrcheck.
3555 * Performance and stability improvements for Helgrind and DRD.
3557 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
3559 * A new experimental tool, BBV, useful for computer architecture
3562 * Improved Wine support, including ability to read Windows PDB
3565 -------------------------
3567 Here are details of the above changes, followed by descriptions of
3568 many other minor changes, and a list of fixed bugs.
3571 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
3572 called "Darwin" because that is the name of the OS core, which is the
3573 level that Valgrind works at.)
3577 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
3578 because it would require work and 10.4 is only becoming less common.
3580 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
3581 fairly well. For 10.5.x, 32-bit programs are the default even on
3582 64-bit machines, so it handles most current programs.
3584 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
3585 officially supported, but simple programs at least will probably work.
3586 However, start-up is slow.
3588 - PowerPC machines are not supported.
3590 Things that don't work:
3592 - The Ptrcheck tool.
3594 - Objective-C garbage collection.
3598 - If you have Rogue Amoeba's "Instant Hijack" program installed,
3599 Valgrind will fail with a SIGTRAP at start-up. See
3600 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
3605 - You will likely find --dsymutil=yes a useful option, as error
3606 messages may be imprecise without it.
3608 - Mac OS X support is new and therefore will be less robust than the
3609 Linux support. Please report any bugs you find.
3611 - Threaded programs may run more slowly than on Linux.
3613 Many thanks to Greg Parker for developing this port over several years.
3616 * Memcheck's leak checker has been improved.
3618 - The results for --leak-check=summary now match the summary results
3619 for --leak-check=full. Previously they could differ because
3620 --leak-check=summary counted "indirectly lost" blocks and
3621 "suppressed" blocks as "definitely lost".
3623 - Blocks that are only reachable via at least one interior-pointer,
3624 but are directly pointed to by a start-pointer, were previously
3625 marked as "still reachable". They are now correctly marked as
3628 - The default value for the --leak-resolution option has been
3629 changed from "low" to "high". In general, this means that more
3630 leak reports will be produced, but each leak report will describe
3631 fewer leaked blocks.
3633 - With --leak-check=full, "definitely lost" and "possibly lost"
3634 leaks are now considered as proper errors, ie. they are counted
3635 for the "ERROR SUMMARY" and affect the behaviour of
3636 --error-exitcode. These leaks are not counted as errors if
3637 --leak-check=summary is specified, however.
3639 - Documentation for the leak checker has been improved.
3642 * Various aspects of Valgrind's text output have changed.
3644 - Valgrind's start-up message has changed. It is shorter but also
3645 includes the command being run, which makes it easier to use
3646 --trace-children=yes. An example:
3648 - Valgrind's shut-down messages have also changed. This is most
3649 noticeable with Memcheck, where the leak summary now occurs before
3650 the error summary. This change was necessary to allow leaks to be
3651 counted as proper errors (see the description of the leak checker
3652 changes above for more details). This was also necessary to fix a
3653 longstanding bug in which uses of suppressions against leaks were
3654 not "counted", leading to difficulties in maintaining suppression
3655 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3657 - Behavior of -v has changed. In previous versions, -v printed out
3658 a mixture of marginally-user-useful information, and tool/core
3659 statistics. The statistics printing has now been moved to its own
3660 flag, --stats=yes. This means -v is less verbose and more likely
3661 to convey useful end-user information.
3663 - The format of some (non-XML) stack trace entries has changed a
3664 little. Previously there were six possible forms:
3666 0x80483BF: really (a.c:20)
3667 0x80483BF: really (in /foo/a.out)
3669 0x80483BF: (within /foo/a.out)
3670 0x80483BF: ??? (a.c:20)
3673 The third and fourth of these forms have been made more consistent
3674 with the others. The six possible forms are now:
3676 0x80483BF: really (a.c:20)
3677 0x80483BF: really (in /foo/a.out)
3678 0x80483BF: really (in ???)
3679 0x80483BF: ??? (in /foo/a.out)
3680 0x80483BF: ??? (a.c:20)
3683 Stack traces produced when --xml=yes is specified are different
3687 * Helgrind and Ptrcheck now support XML output, so they can be used
3688 from GUI tools. Also, the XML output mechanism has been
3691 - The XML format has been overhauled and generalised, so it is more
3692 suitable for error reporting tools in general. The Memcheck
3693 specific aspects of it have been removed. The new format, which
3694 is an evolution of the old format, is described in
3695 docs/internals/xml-output-protocol4.txt.
3697 - Memcheck has been updated to use the new format.
3699 - Helgrind and Ptrcheck are now able to emit output in this format.
3701 - The XML output mechanism has been overhauled. XML is now output
3702 to its own file descriptor, which means that:
3704 * Valgrind can output text and XML independently.
3706 * The longstanding problem of XML output being corrupted by
3707 unexpected un-tagged text messages is solved.
3709 As before, the destination for text output is specified using
3710 --log-file=, --log-fd= or --log-socket=.
3712 As before, XML output for a tool is enabled using --xml=yes.
3714 Because there's a new XML output channel, the XML output
3715 destination is now specified by --xml-file=, --xml-fd= or
3718 Initial feedback has shown this causes some confusion. To
3719 clarify, the two envisaged usage scenarios are:
3721 (1) Normal text output. In this case, do not specify --xml=yes
3722 nor any of --xml-file=, --xml-fd= or --xml-socket=.
3724 (2) XML output. In this case, specify --xml=yes, and one of
3725 --xml-file=, --xml-fd= or --xml-socket= to select the XML
3726 destination, one of --log-file=, --log-fd= or --log-socket=
3727 to select the destination for any remaining text messages,
3728 and, importantly, -q.
3730 -q makes Valgrind completely silent on the text channel,
3731 except in the case of critical failures, such as Valgrind
3732 itself segfaulting, or failing to read debugging information.
3733 Hence, in this scenario, it suffices to check whether or not
3734 any output appeared on the text channel. If yes, then it is
3735 likely to be a critical error which should be brought to the
3736 attention of the user. If no (the text channel produced no
3737 output) then it can be assumed that the run was successful.
3739 This allows GUIs to make the critical distinction they need to
3740 make (did the run fail or not?) without having to search or
3741 filter the text output channel in any way.
3743 It is also recommended to use --child-silent-after-fork=yes in
3747 * Improvements and changes in Helgrind:
3749 - XML output, as described above
3751 - Checks for consistent association between pthread condition
3752 variables and their associated mutexes are now performed.
3754 - pthread_spinlock functions are supported.
3756 - Modest performance improvements.
3758 - Initial (skeletal) support for describing the behaviour of
3759 non-POSIX synchronisation objects through ThreadSanitizer
3760 compatible ANNOTATE_* macros.
3762 - More controllable tradeoffs between performance and the level of
3763 detail of "previous" accesses in a race. There are now three
3766 * --history-level=full. This is the default, and was also the
3767 default in 3.4.x. It shows both stacks involved in a race, but
3768 requires a lot of memory and can be very slow in programs that
3769 do many inter-thread synchronisation events.
3771 * --history-level=none. This only shows the later stack involved
3772 in a race. This can be much faster than --history-level=full,
3773 but makes it much more difficult to find the other access
3774 involved in the race.
3776 The new intermediate setting is
3778 * --history-level=approx
3780 For the earlier (other) access, two stacks are presented. The
3781 earlier access is guaranteed to be somewhere in between the two
3782 program points denoted by those stacks. This is not as useful
3783 as showing the exact stack for the previous access (as per
3784 --history-level=full), but it is better than nothing, and it's
3785 almost as fast as --history-level=none.
3788 * New features and improvements in DRD:
3790 - The error messages printed by DRD are now easier to interpret.
3791 Instead of using two different numbers to identify each thread
3792 (Valgrind thread ID and DRD thread ID), DRD does now identify
3793 threads via a single number (the DRD thread ID). Furthermore
3794 "first observed at" information is now printed for all error
3795 messages related to synchronization objects.
3797 - Added support for named semaphores (sem_open() and sem_close()).
3799 - Race conditions between pthread_barrier_wait() and
3800 pthread_barrier_destroy() calls are now reported.
3802 - Added support for custom allocators through the macros
3803 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
3804 in <valgrind/valgrind.h>). An alternative for these two macros is
3805 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
3808 - Added support for annotating non-POSIX synchronization objects
3809 through several new ANNOTATE_*() macros.
3811 - OpenMP: added support for the OpenMP runtime (libgomp) included
3812 with gcc versions 4.3.0 and 4.4.0.
3816 - Added two new command-line options (--first-race-only and
3817 --segment-merging-interval).
3820 * Genuinely atomic support for x86/amd64/ppc atomic instructions
3822 Valgrind will now preserve (memory-access) atomicity of LOCK-
3823 prefixed x86/amd64 instructions, and any others implying a global
3824 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
3826 This means that Valgrinded processes will "play nicely" in
3827 situations where communication with other processes, or the kernel,
3828 is done through shared memory and coordinated with such atomic
3829 instructions. Prior to this change, such arrangements usually
3830 resulted in hangs, races or other synchronisation failures, because
3831 Valgrind did not honour atomicity of such instructions.
3834 * A new experimental tool, BBV, has been added. BBV generates basic
3835 block vectors for use with the SimPoint analysis tool, which allows
3836 a program's overall behaviour to be approximated by running only a
3837 fraction of it. This is useful for computer architecture
3838 researchers. You can run BBV by specifying --tool=exp-bbv (the
3839 "exp-" prefix is short for "experimental"). BBV was written by
3843 * Modestly improved support for running Windows applications under
3844 Wine. In particular, initial support for reading Windows .PDB debug
3845 information has been added.
3848 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
3849 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
3853 * The Valgrind client requests VALGRIND_PRINTF and
3854 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
3855 the string was always printed immediately on its own line. Now, the
3856 string will be added to a buffer but not printed until a newline is
3857 encountered, or other Valgrind output is printed (note that for
3858 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
3859 "other Valgrind output"). This allows you to use multiple
3860 VALGRIND_PRINTF calls to build up a single output line, and also to
3861 print multiple output lines with a single request (by embedding
3862 multiple newlines in the string).
3865 * The graphs drawn by Massif's ms_print program have changed slightly:
3867 - The half-height chars '.' and ',' are no longer drawn, because
3868 they are confusing. The --y option can be used if the default
3869 y-resolution is not high enough.
3871 - Horizontal lines are now drawn after the top of a snapshot if
3872 there is a gap until the next snapshot. This makes it clear that
3873 the memory usage has not dropped to zero between snapshots.
3876 * Something that happened in 3.4.0, but wasn't clearly announced: the
3877 option --read-var-info=yes can be used by some tools (Memcheck,
3878 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
3879 variable type and location information. This makes those tools
3880 start up more slowly and increases memory consumption, but
3881 descriptions of data addresses in error messages become more
3885 * exp-Omega, an experimental instantaneous leak-detecting tool, was
3886 disabled in 3.4.0 due to a lack of interest and maintenance,
3887 although the source code was still in the distribution. The source
3888 code has now been removed from the distribution. For anyone
3889 interested, the removal occurred in SVN revision r10247.
3892 * Some changes have been made to the build system.
3894 - VEX/ is now integrated properly into the build system. This means
3895 that dependency tracking within VEX/ now works properly, "make
3896 install" will work without requiring "make" before it, and
3897 parallel builds (ie. 'make -j') now work (previously a
3898 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
3899 was effectively ignored).
3901 - The --with-vex configure option has been removed. It was of
3902 little use and removing it simplified the build system.
3904 - The location of some install files has changed. This should not
3905 affect most users. Those who might be affected:
3907 * For people who use Valgrind with MPI programs, the installed
3908 libmpiwrap.so library has moved from
3909 $(INSTALL)/<platform>/libmpiwrap.so to
3910 $(INSTALL)/libmpiwrap-<platform>.so.
3912 * For people who distribute standalone Valgrind tools, the
3913 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
3914 have moved to $(INSTALL)/libcoregrind-<platform>.a.
3916 These changes simplify the build system.
3918 - Previously, all the distributed suppression (*.supp) files were
3919 installed. Now, only default.supp is installed. This should not
3920 affect users as the other installed suppression files were not
3921 read; the fact that they were installed was a mistake.
3924 * KNOWN LIMITATIONS:
3926 - Memcheck is unusable with the Intel compiler suite version 11.1,
3927 when it generates code for SSE2-and-above capable targets. This
3928 is because of icc's use of highly optimised inlined strlen
3929 implementations. It causes Memcheck to report huge numbers of
3930 false errors even in simple programs. Helgrind and DRD may also
3933 Versions 11.0 and earlier may be OK, but this has not been
3937 The following bugs have been fixed or resolved. Note that "n-i-bz"
3938 stands for "not in bugzilla" -- that is, a bug that was reported to us
3939 but never got a bugzilla entry. We encourage you to file bugs in
3940 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3941 mailing the developers (or mailing lists) directly -- bugs that are
3942 not entered into bugzilla tend to get forgotten about or ignored.
3944 To see details of a given bug, visit
3945 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3946 where XXXXXX is the bug number as listed below.
3948 84303 How about a LockCheck tool?
3949 91633 dereference of null ptr in vgPlain_st_basetype
3950 97452 Valgrind doesn't report any pthreads problems
3951 100628 leak-check gets assertion failure when using
3952 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
3953 108528 NPTL pthread cleanup handlers not called
3954 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
3955 110128 mallinfo is not implemented...
3956 110770 VEX: Generated files not always updated when making valgrind
3957 111102 Memcheck: problems with large (memory footprint) applications
3958 115673 Vex's decoder should never assert
3959 117564 False positive: Syscall param clone(child_tidptr) contains
3960 uninitialised byte(s)
3961 119404 executing ssh from inside valgrind fails
3962 133679 Callgrind does not write path names to sources with dwarf debug
3964 135847 configure.in problem with non gnu compilers (and possible fix)
3965 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
3966 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
3967 136230 memcheck reports "possibly lost", should be "still reachable"
3968 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
3969 137904 Valgrind reports a memory leak when using POSIX threads,
3971 139076 valgrind VT_GETSTATE error
3972 142228 complaint of elf_dynamic_do_rela in trivial usage
3973 145347 spurious warning with USBDEVFS_REAPURB
3974 148441 (wine) can't find memory leak in Wine, win32 binary
3976 148742 Leak-check fails assert on exit
3977 149878 add (proper) check for calloc integer overflow
3978 150606 Call graph is broken when using callgrind control
3979 152393 leak errors produce an exit code of 0. I need some way to
3980 cause leak errors to result in a nonzero exit code.
3981 157154 documentation (leak-resolution doc speaks about num-callers
3982 def=4) + what is a loss record
3983 159501 incorrect handling of ALSA ioctls
3984 162020 Valgrinding an empty/zero-byte file crashes valgrind
3985 162482 ppc: Valgrind crashes while reading stabs information
3986 162718 x86: avoid segment selector 0 in sys_set_thread_area()
3987 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
3988 163560 VEX/test_main.c is missing from valgrind-3.3.1
3989 164353 malloc_usable_size() doesn't return a usable size
3990 165468 Inconsistent formatting in memcheck manual -- please fix
3991 169505 main.c:286 (endOfInstr):
3992 Assertion 'ii->cost_offset == *cost_offset' failed
3993 177206 Generate default.supp during compile instead of configure
3994 177209 Configure valt_load_address based on arch+os
3995 177305 eventfd / syscall 323 patch lost
3996 179731 Tests fail to build because of inlining of non-local asm labels
3997 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
3998 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
3999 181594 Bogus warning for empty text segment
4000 181707 dwarf doesn't require enumerations to have name
4001 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4002 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4003 Assertion '!already_present' failed.
4004 185359 exp-ptrcheck: unhandled syscall getresuid()
4005 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4006 185816 Valgrind is unable to handle debug info for files with split
4007 debug info that are prelinked afterwards
4008 185980 [darwin] unhandled syscall: sem_open
4009 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4010 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4011 186790 Suppression pattern used for leaks are not reported
4012 186796 Symbols with length>200 in suppression files are ignored
4013 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4014 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4015 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4016 188046 bashisms in the configure script
4017 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4018 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4019 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4020 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4021 assertion fail "!lock->heldBy"
4022 188427 Add support for epoll_create1 (with patch)
4023 188530 Support for SIOCGSTAMPNS
4024 188560 Include valgrind.spec in the tarball
4025 188572 Valgrind on Mac should suppress setenv() mem leak
4026 189054 Valgrind fails to build because of duplicate non-local asm labels
4027 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4028 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4029 189763 drd assertion failure: s_threadinfo[tid].is_recording
4030 190219 unhandled syscall: 328 (x86-linux)
4031 190391 dup of 181394; see above
4032 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4033 190820 No debug information on powerpc-linux
4034 191095 PATCH: Improve usbdevfs ioctl handling
4035 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4037 191189 --xml=yes should obey --gen-suppressions=all
4038 191192 syslog() needs a suppression on macosx
4039 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4040 191761 getrlimit on MacOSX
4041 191992 multiple --fn-skip only works sometimes; dependent on order
4042 192634 V. reports "aspacem sync_check_mapping_callback:
4043 segment mismatch" on Darwin
4044 192954 __extension__ missing on 2 client requests
4045 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4046 194474 "INSTALL" file has different build instructions than "README"
4047 194671 Unhandled syscall (sem_wait?) from mac valgrind
4048 195069 memcheck: reports leak (memory still reachable) for
4050 195169 drd: (vgDrd_barrier_post_wait):
4051 Assertion 'r->sg[p->post_iteration]' failed.
4052 195268 valgrind --log-file doesn't accept ~/...
4053 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4054 195860 WARNING: unhandled syscall: unix:223
4055 196528 need a error suppression for pthread_rwlock_init under os x?
4056 197227 Support aio_* syscalls on Darwin
4057 197456 valgrind should reject --suppressions=(directory)
4058 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4059 197591 unhandled syscall 27 (mincore)
4060 197793 Merge DCAS branch to the trunk == 85756, 142103
4061 197794 Avoid duplicate filenames in Vex
4062 197898 make check fails on current SVN
4063 197901 make check fails also under exp-ptrcheck in current SVN
4064 197929 Make --leak-resolution=high the default
4065 197930 Reduce spacing between leak reports
4066 197933 Print command line of client at start-up, and shorten preamble
4067 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4068 198395 add BBV to the distribution as an experimental tool
4069 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4070 198649 callgrind_annotate doesn't cumulate counters
4071 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4072 199977 Valgrind complains about an unrecognized instruction in the
4073 atomic_incs test program
4074 200029 valgrind isn't able to read Fedora 12 debuginfo
4075 200760 darwin unhandled syscall: unix:284
4076 200827 DRD doesn't work on Mac OS X
4077 200990 VG_(read_millisecond_timer)() does not work correctly
4078 201016 Valgrind does not support pthread_kill() on Mac OS
4079 201169 Document --read-var-info
4080 201323 Pre-3.5.0 performance sanity checking
4081 201384 Review user manual for the 3.5.0 release
4082 201585 mfpvr not implemented on ppc
4083 201708 tests failing because x86 direction flag is left set
4084 201757 Valgrind doesn't handle any recent sys_futex additions
4085 204377 64-bit valgrind can not start a shell script
4086 (with #!/path/to/shell) if the shell is a 32-bit executable
4087 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4088 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4089 about memory allocation events with memory access tracing enabled
4090 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4092 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4096 Release 3.4.1 (28 February 2009)
4097 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4098 3.4.1 is a bug-fix release that fixes some regressions and assertion
4099 failures in debug info reading in 3.4.0, most notably incorrect stack
4100 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4101 other debug info problems are also fixed. A number of bugs in the
4102 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4104 In view of the fact that 3.4.0 contains user-visible regressions
4105 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4106 encouraged to ship 3.4.1 in preference to 3.4.0.
4108 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4109 bugzilla" -- that is, a bug that was reported to us but never got a
4110 bugzilla entry. We encourage you to file bugs in bugzilla
4111 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4112 developers (or mailing lists) directly -- bugs that are not entered
4113 into bugzilla tend to get forgotten about or ignored.
4115 n-i-bz Fix various bugs reading icc-11 generated debug info
4116 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4117 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4118 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4119 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4120 179618 exp-ptrcheck crashed / exit prematurely
4121 179624 helgrind: false positive races with pthread_create and
4122 recv/open/close/read
4123 134207 pkg-config output contains @VG_PLATFORM@
4124 176926 floating point exception at valgrind startup with PPC 440EPX
4125 181594 Bogus warning for empty text segment
4126 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4127 181707 Dwarf3 doesn't require enumerations to have name
4128 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4129 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4130 Assertion '!already_present' failed.
4131 185359 exp-ptrcheck unhandled syscall getresuid()
4133 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4134 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4138 Release 3.4.0 (2 January 2009)
4139 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4140 3.4.0 is a feature release with many significant improvements and the
4141 usual collection of bug fixes. This release supports X86/Linux,
4142 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4143 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4145 3.4.0 brings some significant tool improvements. Memcheck can now
4146 report the origin of uninitialised values, the thread checkers
4147 Helgrind and DRD are much improved, and we have a new experimental
4148 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4149 global arrays. In detail:
4151 * Memcheck is now able to track the origin of uninitialised values.
4152 When it reports an uninitialised value error, it will try to show
4153 the origin of the value, as either a heap or stack allocation.
4154 Origin tracking is expensive and so is not enabled by default. To
4155 use it, specify --track-origins=yes. Memcheck's speed will be
4156 essentially halved, and memory usage will be significantly
4157 increased. Nevertheless it can drastically reduce the effort
4158 required to identify the root cause of uninitialised value errors,
4159 and so is often a programmer productivity win, despite running more
4162 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4163 3.4.0, will be released shortly.
4165 * Helgrind's race detection algorithm has been completely redesigned
4166 and reimplemented, to address usability and scalability concerns:
4168 - The new algorithm has a lower false-error rate: it is much less
4169 likely to report races that do not really exist.
4171 - Helgrind will display full call stacks for both accesses involved
4172 in a race. This makes it easier to identify the root causes of
4175 - Limitations on the size of program that can run have been removed.
4177 - Performance has been modestly improved, although that is very
4180 - Direct support for Qt4 threading has been added.
4182 - pthread_barriers are now directly supported.
4184 - Helgrind works well on all supported Linux targets.
4186 * The DRD thread debugging tool has seen major improvements:
4188 - Greatly improved performance and significantly reduced memory
4191 - Support for several major threading libraries (Boost.Thread, Qt4,
4192 glib, OpenMP) has been added.
4194 - Support for atomic instructions, POSIX semaphores, barriers and
4195 reader-writer locks has been added.
4197 - Works now on PowerPC CPUs too.
4199 - Added support for printing thread stack usage at thread exit time.
4201 - Added support for debugging lock contention.
4203 - Added a manual for Drd.
4205 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4206 checks for misuses of pointers. In that sense it is a bit like
4207 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4208 detect overruns of stack and global arrays, it can detect
4209 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4210 detect accesses heap blocks that have been freed a very long time
4211 ago (millions of blocks in the past).
4213 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4214 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4215 of the main Valgrind documentation. As this is an experimental
4216 tool, we would be particularly interested in hearing about your
4217 experiences with it.
4219 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4220 longer built by default, although the code remains in the repository
4221 and the tarball. This is due to three factors: a perceived lack of
4222 users, a lack of maintenance, and concerns that it may not be
4223 possible to achieve reliable operation using the existing design.
4225 * As usual, support for the latest Linux distros and toolchain
4226 components has been added. It should work well on Fedora Core 10,
4227 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4228 state) is supported, as is glibc-2.9. The C++ demangler has been
4229 updated so as to work well with C++ compiled by even the most recent
4232 * You can now use frame-level wildcards in suppressions. This was a
4233 frequently-requested enhancement. A line "..." in a suppression now
4234 matches zero or more frames. This makes it easier to write
4235 suppressions which are precise yet insensitive to changes in
4238 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4240 * Very basic support for IBM Power6 has been added (64-bit processes only).
4242 * Valgrind is now cross-compilable. For example, it is possible to
4243 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4244 on a ppc32/64-linux target.
4246 * You can set the main thread's stack size at startup using the
4247 new --main-stacksize= flag (subject of course to ulimit settings).
4248 This is useful for running apps that need a lot of stack space.
4250 * The limitation that you can't use --trace-children=yes together
4251 with --db-attach=yes has been removed.
4253 * The following bugs have been fixed. Note that "n-i-bz" stands for
4254 "not in bugzilla" -- that is, a bug that was reported to us but
4255 never got a bugzilla entry. We encourage you to file bugs in
4256 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4257 mailing the developers (or mailing lists) directly.
4259 n-i-bz Make return types for some client requests 64-bit clean
4260 n-i-bz glibc 2.9 support
4261 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4262 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4263 n-i-bz Building in an env without gdb gives bogus gdb attach
4264 92456 Tracing the origin of uninitialised memory
4265 106497 Valgrind does not demangle some C++ template symbols
4267 151612 Suppression with "..." (frame-level wildcards in .supp files)
4268 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4269 159285 unhandled syscall:25 (stime, on x86-linux)
4270 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4271 160954 ppc build of valgrind crashes with illegal instruction (isel)
4272 160956 mallinfo implementation, w/ patch
4273 162092 Valgrind fails to start gnome-system-monitor
4274 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4275 163794 assertion failure with "--track-origins=yes"
4276 163933 sigcontext.err and .trapno must be set together
4277 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4278 164476 Missing kernel module loading system calls
4279 164669 SVN regression: mmap() drops posix file locks
4280 166581 Callgrind output corruption when program forks
4281 167288 Patch file for missing system calls on Cell BE
4282 168943 unsupported scas instruction pentium
4283 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4284 172417 x86->IR: 0x82 ...
4285 172563 amd64->IR: 0xD9 0xF5 - fprem1
4286 173099 .lds linker script generation error
4287 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4288 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4290 174908 --log-file value not expanded correctly for core file
4291 175044 Add lookup_dcookie for amd64
4292 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4294 Developer-visible changes:
4296 * Valgrind's debug-info reading machinery has been majorly overhauled.
4297 It can now correctly establish the addresses for ELF data symbols,
4298 which is something that has never worked properly before now.
4300 Also, Valgrind can now read DWARF3 type and location information for
4301 stack and global variables. This makes it possible to use the
4302 framework to build tools that rely on knowing the type and locations
4303 of stack and global variables, for example exp-Ptrcheck.
4305 Reading of such information is disabled by default, because most
4306 tools don't need it, and because it is expensive in space and time.
4307 However, you can force Valgrind to read it, using the
4308 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4309 make use of such information, if present, to provide source-level
4310 descriptions of data addresses in the error messages they create.
4312 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4313 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).