1 Release 3.??.? (?? 2021)
2 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
5 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
6 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
7 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
8 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
10 * ==================== CORE CHANGES ===================
12 * ================== PLATFORM CHANGES =================
18 - Support the miscellaneous-instruction-extensions facility 3 and the
19 vector-enhancements facility 2. This enables programs compiled with
20 "-march=arch13" or "-march=z15" to be executed under Valgrind.
24 - ISA 3.1 support is now complete
25 - ISA 3.0 support for the darn instruction added.
26 - ISA 3.0 support for the vector system call instruction scv added.
27 - ISA 3.0 support for the copy, paste and cpabort instructions added.
29 * ==================== TOOL CHANGES ====================
31 * ==================== FIXED BUGS ====================
33 The following bugs have been fixed or resolved. Note that "n-i-bz"
34 stands for "not in bugzilla" -- that is, a bug that was reported to us
35 but never got a bugzilla entry. We encourage you to file bugs in
36 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
37 than mailing the developers (or mailing lists) directly -- bugs that
38 are not entered into bugzilla tend to get forgotten about or ignored.
40 423963 Error in child thread when CLONE_PIDFD is used
41 429375 PPC ISA 3.1 support is missing, part 9
42 431157 PPC_FEATURE2_SCV needs to be masked in AT_HWCAP2
43 431306 Update demangler to support Rust v0 name mangling
44 433801 PPC ISA 3.1 support is missing, part 10 (ISA 3.1 support complete)
45 433863 s390x: memcheck/tests/s390x/{cds,cs,csg} failures
46 434840 PPC64 darn instruction not supported
47 434296 s390x: False-positive memcheck diagnostics from vector string
49 435665 PPC ISA 3.0 copy, paste, cpabort instructions are not supported
50 438871 unhandled instruction bytes: 0xF3 0x49 0xF 0x6F 0x9C 0x24 0x60 0x2 0x0 0x0
51 435908 valgrind tries to fetch from deubginfod for files which already
52 have debug information
53 439590 glibc-2.34 breaks suppressions against obj:*/lib*/libc-2.*so*
54 440670 unhandled ppc64le-linux syscall: 252 (statfs64) and 253 (fstatfs64)
55 432387 s390x: z15 instructions support
56 440906 Fix impossible constraint issue in P10 testcase.
57 441512 Remove a unneeded / unnecessary prefix check.
58 441534 Update the expected output for test_isa_3_1_VRT.
59 442061 very slow execution under Fedora 34 (readdwarf3)
61 To see details of a given bug, visit
62 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
63 where XXXXXX is the bug number as listed below.
66 Release 3.17.0 (19 Mar 2021)
67 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
69 3.17.0 fixes a number of bugs and adds some functional changes: support for
70 GCC 11, Clang 11, DWARF5 debuginfo, the 'debuginfod' debuginfo server, and
71 some new instructions for Arm64, S390 and POWER. There are also some tool
74 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
75 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
76 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
77 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
78 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
80 * ==================== CORE CHANGES ===================
82 * DWARF version 5 support. Valgrind can now read DWARF version 5 debuginfo as
85 * Valgrind now supports debuginfod, an HTTP server for distributing ELF/DWARF
86 debugging information. When a debuginfo file cannot be found locally,
87 Valgrind is able to query debuginfod servers for the file using its
88 build-id. See the user manual for more information about debuginfod support.
90 * ================== PLATFORM CHANGES =================
94 - Inaccuracies resulting from double-rounding in the simulation of
95 floating-point multiply-add/subtract instructions have been fixed. These
96 should now behave exactly as the hardware does.
98 - Partial support for the ARM v8.2 instruction set. v8.2 support work is
99 ongoing. Support for the half-word variants of at least the following
100 instructions has been added:
102 FABS <Vd>.<T>, <Vn>.<T>
104 FNEG <Vd>.<T>, <Vn>.<T>
106 FSQRT <Vd>.<T>, <Vn>.<T>
111 - Implement the new instructions/features that were added to z/Architecture
112 with the vector-enhancements facility 1. Also cover the instructions from
113 the vector-packed-decimal facility that are defined outside the chapter
114 "Vector Decimal Instructions", but not the ones from that chapter itself.
116 For a detailed list of newly supported instructions see the updates to
117 `docs/internals/s390-opcodes.csv'.
119 Since the miscellaneous instruction extensions facility 2 was already
120 added in Valgrind 3.16.0, this completes the support necessary to run
121 general programs built with `--march=z14' under Valgrind. The
122 vector-packed-decimal facility is currently not exploited by the standard
123 toolchain and libraries.
127 - Various bug fixes. Fix for the sync field to limit setting just two of
128 the two bits in the L-field. Fix the write size for the stxsibx and
129 stxsihx instructions. Fix the modsw and modsd instructions.
131 - Partial support for ISA 3.1 has been added. Support for the VSX PCV mask
132 instructions, bfloat16 GER instructions, and bfloat16 to/from float 32-bit
133 conversion instructions are still missing.
135 * ==================== TOOL CHANGES ====================
137 * General tool changes
139 - All the tools and their vgpreload libraries are now installed under
140 libexec because they cannot be executed directly and should be run through
141 the valgrind executable. This should be an internal, not user visible,
142 change, but might impact valgrind packagers.
144 - The --track-fds option now respects -q, --quiet and won't output anything
145 if no file descriptors are leaked. It also won't report the standard stdin
146 (0), stdout (1) or stderr (2) descriptors as being leaked with
147 --trace-fds=yes anymore. To track whether the standard file descriptors
148 are still open at the end of the program run use --trace-fds=all.
152 - DHAT has been extended, with two new modes of operation. The new
153 --mode=copy flag triggers copy profiling, which records calls to memcpy,
154 strcpy, and similar functions. The new --mode=ad-hoc flag triggers ad hoc
155 profiling, which records calls to the DHAT_AD_HOC_EVENT client request in
156 the new dhat/dhat.h file. This is useful for learning more about hot code
157 paths. See the user manual for more information about the new modes.
159 - Because of these changes, DHAT's file format has changed. DHAT output
160 files produced with earlier versions of DHAT will not work with this
161 version of DHAT's viewer, and DHAT output files produced with this version
162 of DHAT will not work with earlier versions of DHAT's viewer.
164 * ==================== FIXED BUGS ====================
166 The following bugs have been fixed or resolved. Note that "n-i-bz"
167 stands for "not in bugzilla" -- that is, a bug that was reported to us
168 but never got a bugzilla entry. We encourage you to file bugs in
169 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
170 than mailing the developers (or mailing lists) directly -- bugs that
171 are not entered into bugzilla tend to get forgotten about or ignored.
173 To see details of a given bug, visit
174 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
175 where XXXXXX is the bug number as listed below.
177 140178 open("/proc/self/exe", ...); doesn't quite work
178 140939 --track-fds reports leakage of stdout/in/err and doesn't respect -q
179 217695 malloc/calloc/realloc/memalign failure doesn't set errno to ENOMEM
180 338633 gdbserver_tests/nlcontrolc.vgtest hangs on arm64
181 345077 linux syscall execveat support (linux 3.19)
182 361770 Missing F_ADD_SEALS
183 369029 handle linux syscalls sched_getattr and sched_setattr
184 384729 __libc_freeres inhibits cross-platform valgrind
185 388787 Support for C++17 new/delete
186 391853 Makefile.all.am:L247 and @SOLARIS_UNDEF_LARGESOURCE@ being empty
187 396656 Warnings while reading debug info
188 397605 ioctl FICLONE mishandled
189 401416 Compile failure with openmpi 4.0
190 408663 Suppression file for musl libc
191 404076 s390x: z14 vector instructions not implemented
192 410743 shmat() calls for 32-bit programs fail when running in 64-bit valgrind
193 (actually affected all x86 and nanomips regardless of host bitness)
194 413547 regression test does not check for Arm 64 features.
195 414268 Enable AArch64 feature detection and decoding for v8.x instructions
196 415293 Incorrect call-graph tracking due to new _dl_runtime_resolve_xsave*
197 422174 unhandled instruction bytes: 0x48 0xE9 (REX prefixed JMP instruction)
198 422261 platform selection fails for unqualified client name
199 422623 epoll_ctl warns for uninitialized padding on non-amd64 64bit arches
200 423021 PPC: Add missing ISA 3.0 documentation link and HWCAPS test.
201 423195 PPC ISA 3.1 support is missing, part 1
202 423361 Adds io_uring support on arm64/aarch64 (and all other arches)
203 424012 crash with readv/writev having invalid but not NULL arg2 iovec
204 424298 amd64: Implement RDSEED
205 425232 PPC ISA 3.1 support is missing, part 2
206 425820 Failure to recognize vpcmpeqq as a dependency breaking idiom.
207 426014 arm64: implement fmadd and fmsub as Iop_MAdd/Sub
208 426123 PPC ISA 3.1 support is missing, part 3
209 426144 Fix "condition variable has not been initialized" on Fedora 33.
210 427400 PPC ISA 3.1 support is missing, part 4
211 427401 PPC ISA 3.1 support is missing, part 5
212 427404 PPC ISA 3.1 support is missing, part 6
213 427870 lmw, lswi and related PowerPC insns aren't allowed on ppc64le
214 427787 Support new faccessat2 linux syscall (439)
215 427969 debuginfo section duplicates a section in the main ELF file
216 428035 drd: Unbreak the musl build
217 428648 s390_emit_load_mem panics due to 20-bit offset for vector load
218 428716 cppcheck detects potential leak in VEX/useful/smchash.c
219 428909 helgrind: need to intercept duplicate libc definitions for Fedora 33
220 429352 PPC ISA 3.1 support is missing, part 7
221 429354 PPC ISA 3.1 support is missing, part 8
222 429692 unhandled ppc64le-linux syscall: 147 (getsid)
223 429864 s390x: C++ atomic test_and_set yields false-positive memcheck
225 429952 Errors when building regtest with clang
226 430354 ppc stxsibx and stxsihx instructions write too much data
227 430429 valgrind.h doesn't compile on s390x with clang
228 430485 expr_is_guardable doesn't handle Iex_Qop
229 431556 Complete arm64 FADDP v8.2 instruction support
230 432102 Add support for DWARF5 as produced by GCC11
231 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT
232 432381 drd: Process STACK_REGISTER client requests
233 432552 [AArch64] invalid error emitted for pre-decremented byte/hword addresses
234 432672 vg_regtest: test-specific environment variables not reset between tests
235 432809 VEX should support REX.W + POPF
236 432861 PPC modsw and modsd give incorrect results for 1 mod 12
237 432870 gdbserver_tests:nlcontrolc hangs with newest glibc2.33 x86-64
238 432215 Add debuginfod functionality
239 433323 Use pkglibexecdir as vglibdir
240 433500 DRD regtest faulures when libstdc++ and libgcc debuginfo are installed
241 433629 valgrind/README has type "abd" instead of "and"
242 433641 Rust std::sys::unix::fs::try_statx Syscall param fstatat(file_name)
243 433898 arm64: Handle sp, lr, fp as DwReg in CfiExpr
244 434193 GCC 9+ inlined strcmp causes "Conditional jump or move [..] value" report
245 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
246 n-i-bz arm64 front end: avoid Memcheck false positives relating to CPUID
248 (3.17.0.RC1: 13 Mar 2021)
249 (3.17.0.RC2: 17 Mar 2021)
250 (3.17.0: 19 Mar 2021)
254 Release 3.16.1 (22 June 2020)
255 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
257 3.16.1 fixes two critical bugs discovered after 3.16.0 was frozen. It also
258 fixes character encoding problems in the documentation HTML.
260 422677 PPC sync instruction L field should only be 2 bits in ISA 3.0
261 422715 32-bit x86: vex: the `impossible' happened: expr_is_guardable: unhandled expr
263 (3.16.1, 22 June 2020, 36d6727e1d768333a536f274491e5879cab2c2f7)
267 Release 3.16.0 (27 May 2020)
268 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
270 3.16.0 is a feature release with many improvements and the usual collection of
273 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
274 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
275 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
276 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
277 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
279 * ==================== CORE CHANGES ===================
281 * It is now possible to dynamically change the value of many command line
282 options while your program (or its children) are running under Valgrind.
284 To see the list of dynamically changeable options, run
285 "valgrind --help-dyn-options".
287 You can change the options from the shell by using vgdb to launch
288 the monitor command "v.clo <clo option>...".
289 The same monitor command can be used from a gdb connected
290 to the valgrind gdbserver.
291 Your program can also change the dynamically changeable options using
292 the client request VALGRIND_CLO_CHANGE(option).
294 * ================== PLATFORM CHANGES =================
296 * MIPS: preliminary support for nanoMIPS instruction set has been added.
298 * ==================== TOOL CHANGES ====================
302 - The implicit memcpy done by each call to realloc now counts towards the
303 read and write counts of resized heap blocks, making those counts higher
308 - cg_annotate's --auto and --show-percs options now default to 'yes', because
309 they are usually wanted.
313 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
314 because they are usually wanted.
316 - The command option --collect-systime has been enhanced to specify
317 the unit used to record the elapsed time spent during system calls.
318 The command option now accepts the values no|yes|msec|usec|nsec,
319 where yes is a synonym of msec. When giving the value nsec, the
320 system cpu time of system calls is also recorded.
324 - Several memcheck options are now dynamically changeable.
325 Use valgrind --help-dyn-options to list them.
327 - The release 3.15 introduced a backward incompatible change for
328 some suppression entries related to preadv and pwritev syscalls.
329 When reading a suppression entry using the unsupported 3.14 format,
330 valgrind will now produce a warning to say the suppression entry will not
331 work, and suggest the needed change.
333 - Significantly fewer false positive errors on optimised code generated by
334 Clang and GCC. In particular, Memcheck now deals better with the
335 situation where the compiler will transform C-level "A && B" into "B && A"
336 under certain circumstances (in which the transformation is valid).
337 Handling of integer equality/non-equality checks on partially defined
338 values is also improved on some architectures.
342 - The exprimental Stack and Global Array Checking tool has been removed.
343 It only ever worked on x86 and amd64, and even on those it had a
344 high false positive rate and was slow. An alternative for detecting
345 stack and global array overruns is using the AddressSanitizer (ASAN)
346 facility of the GCC and Clang compilers, which require you to rebuild
347 your code with -fsanitize=address.
349 * ==================== OTHER CHANGES ====================
351 * New and modified GDB server monitor features:
353 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
355 - The gdbserver monitor commands that require an address and an optional
356 length argument now accepts the alternate 'C like' syntax "address[length]".
357 For example, the memcheck command "monitor who_points_at 0x12345678 120"
358 can now also be given as "monitor who_points_at 0x12345678[120]".
360 * ==================== FIXED BUGS ====================
362 The following bugs have been fixed or resolved. Note that "n-i-bz"
363 stands for "not in bugzilla" -- that is, a bug that was reported to us
364 but never got a bugzilla entry. We encourage you to file bugs in
365 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
366 than mailing the developers (or mailing lists) directly -- bugs that
367 are not entered into bugzilla tend to get forgotten about or ignored.
369 To see details of a given bug, visit
370 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
371 where XXXXXX is the bug number as listed below.
373 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
374 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
375 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
376 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
377 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
378 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
379 400872 Add nanoMIPS support to Valgrind
380 403212 drd/tests/trylock hangs on FreeBSD
381 404406 s390x: z14 miscellaneous instructions not implemented
382 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
383 406561 mcinfcallWSRU gdbserver_test fails on ppc64
384 406824 Unsupported baseline
385 407218 Add support for the copy_file_range syscall
386 407307 Intercept stpcpy also in ld.so for arm64
387 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
389 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
390 408009 Expose rdrand and f16c even on avx if host cpu supports them
391 408091 Missing pkey syscalls
392 408414 Add support for missing for preadv2 and pwritev2 syscalls
393 409141 Valgrind hangs when SIGKILLed
394 409206 Support for Linux PPS and PTP ioctls
395 409367 exit_group() after signal to thread waiting in futex() causes hangs
396 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
397 409780 References to non-existent configure.in
398 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
399 410599 Non-deterministic behaviour of pth_self_kill_15_other test
400 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
401 411134 Allow the user to change a set of command line options during execution
402 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
403 412344 Problem setting mips flags with specific paths
404 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
405 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
406 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
407 413603 callgrind_annotate/cg_annotate truncate function names at '#'
408 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
409 415136 ARMv8.1 Compare-and-Swap instructions are not supported
410 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
411 416239 valgrind crashes when handling clock_adjtime
412 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
413 416286 DRD reports "conflicting load" error on std::mutex::lock()
414 416301 s390x: "compare and signal" not supported
415 416387 finit_module and bpf syscalls are unhandled on arm64
416 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
417 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
418 416753 new 32bit time syscalls for 2038+
419 417075 pwritev(vector[...]) suppression ignored
420 417075 is not fixed, but incompatible supp entries are detected
421 and a warning is produced for these.
422 417187 [MIPS] Conditional branch problem since 'grail' changes
423 417238 Test memcheck/tests/vbit-test fails on mips64 BE
424 417266 Make memcheck/tests/linux/sigqueue usable with musl
425 417281 s390x: /bin/true segfaults with "grail" enabled
426 417427 commit to fix vki_siginfo_t definition created numerous regression
428 417452 s390_insn_store_emit: dst->tag for HRcVec128
429 417578 Add suppressions for glibc DTV leaks
430 417906 clone with CLONE_VFORK and no CLONE_VM fails
431 418004 Grail code additions break ppc64.
432 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
433 418997 s390x: Support Iex_ITE for float and vector types
434 419503 s390x: Avoid modifying registers returned from isel functions
435 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
436 421570 std_mutex fails on Arm v8.1 h/w
437 434035 vgdb might crash if valgrind is killed
438 n-i-bz Fix minor one time leaks in dhat.
439 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
440 n-i-bz Add support for the Linux io_uring system calls
441 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
442 n-i-bz Fix non-glibc build of test suite with s390x_features
443 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
444 423195 PPC ISA 3.1 support is missing, part 1
446 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
447 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
448 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
452 Release 3.15.0 (12 April 2019)
453 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
455 3.15.0 is a feature release with many improvements and the usual collection of
458 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
459 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
460 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
461 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
462 support for X86/macOS 10.13 and AMD64/macOS 10.13.
464 * ==================== CORE CHANGES ===================
466 * The XTree Massif output format now makes use of the information obtained
467 when specifying --read-inline-info=yes.
469 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
471 * ==================== TOOL CHANGES ====================
475 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
476 it has been promoted from an experimental tool to a regular tool. Run it
477 with --tool=dhat instead of --tool=exp-dhat.
479 - DHAT now prints only minimal data when the program ends, instead writing
480 the bulk of the profiling data to a file. As a result, the --show-top-n
481 and --sort-by options have been removed.
483 - Profile results can be viewed with the new viewer, dh_view.html. When
484 a run ends, a short message is printed, explaining how to view the result.
486 - See the documentation for more details.
490 - cg_annotate has a new option, --show-percs, which prints percentages next
495 - callgrind_annotate has a new option, --show-percs, which prints percentages
496 next to all event counts.
498 - callgrind_annotate now inserts commas in call counts, and
499 sort the caller/callee lists in the call tree.
503 - The default value for --read-inline-info is now "yes" on
504 Linux/Android/Solaris. It is still "no" on other OS.
508 - The option --xtree-leak=yes (to output leak result in xtree format)
509 automatically activates the option --show-leak-kinds=all, as xtree
510 visualisation tools such as kcachegrind can in any case select what kind
511 of leak to visualise.
513 - There has been further work to avoid false positives. In particular,
514 integer equality on partially defined inputs (C == and !=) is now handled
517 * ==================== OTHER CHANGES ====================
519 * The new option --show-error-list=no|yes displays, at the end of the run, the
520 list of detected errors and the used suppressions. Prior to this change,
521 showing this information could only be done by specifying "-v -v", but that
522 also produced a lot of other possibly-non-useful messages. The option -s is
523 equivalent to --show-error-list=yes.
525 * ==================== FIXED BUGS ====================
527 The following bugs have been fixed or resolved. Note that "n-i-bz"
528 stands for "not in bugzilla" -- that is, a bug that was reported to us
529 but never got a bugzilla entry. We encourage you to file bugs in
530 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
531 than mailing the developers (or mailing lists) directly -- bugs that
532 are not entered into bugzilla tend to get forgotten about or ignored.
534 To see details of a given bug, visit
535 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
536 where XXXXXX is the bug number as listed below.
538 385411 s390x: z13 vector floating-point instructions not implemented
539 397187 z13 vector register support for vgdb gdbserver
540 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
541 398870 Please add support for instruction vcvtps2ph
542 399287 amd64 front end: Illegal Instruction vcmptrueps
543 399301 Use inlined frames in Massif XTree output.
544 399322 Improve callgrind_annotate output
545 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
546 400164 helgrind test encounters mips x-compiler warnings and assembler error
547 400490 s390x: VRs allocated as if separate from FPRs
548 400491 s390x: Operand of LOCH treated as unsigned integer
549 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
550 options, which specify a mips64 processor
551 401112 LLVM 5.0 generates comparison against partially initialized data
552 401277 More bugs in z13 support
553 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
554 401578 drd: crashes sometimes on fork()
555 401627 memcheck errors with glibc avx2 optimized wcsncmp
556 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
557 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
558 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
560 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
561 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
562 402123 invalid assembler opcodes for mips32r2
563 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
564 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
565 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
566 402351 mips64 libvexmultiarch_test fails on s390x
568 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
569 402480 Do not use %rsp in clobber list
570 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
571 402515 Implement new option --show-error-list=no|yes / -s
572 402519 POWER 3.0 addex instruction incorrectly implemented
573 402781 Redo the cache used to process indirect branch targets
574 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
575 403552 s390x: wrong facility bit checked for vector facility
576 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
577 404638 Add VG_(replaceIndexXA)
578 404843 s390x: backtrace sometimes ends prematurely
579 404888 autotools cleanup series
580 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
581 405182 Valgrind fails to build with Clang
582 405205 filter_libc: remove the line holding the futex syscall error entirely
583 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
584 the upper and lower 32-bits of the 64-bit result
585 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
586 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
587 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
589 405403 s390x disassembler cannot be used on x86
590 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
591 405458 MIPS mkFormVEC arguments swapped?
592 405716 drd: Fix an integer overflow in the stack margin calculation
593 405722 Support arm64 core dump
594 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
595 of the 64-bit destination field.
596 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
597 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
598 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
600 406256 PPC64, vector floating point instructions don't handle subnormal
601 according to VSCR[NJ] bit setting.
602 406352 cachegrind/callgrind fails ann tests because of missing a.c
603 406354 dhat is broken on x86 (32bit)
604 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
605 406357 gdbserver_tests fails because of gdb output change
606 406360 memcheck/tests/libstdc++.supp needs more supression variants
607 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
608 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
609 407340 PPC64, does not support the vlogefp, vexptefp instructions.
610 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
611 n-i-bz Fix callgrind_annotate non deterministic order for equal total
612 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
613 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
614 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
616 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
617 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
618 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
622 Release 3.14.0 (9 October 2018)
623 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
625 3.14.0 is a feature release with many improvements and the usual collection of
628 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
629 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
630 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
631 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
632 support for X86/macOS 10.13, AMD64/macOS 10.13.
634 * ==================== CORE CHANGES ===================
636 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
637 debug info for unloaded code. This allows saved stack traces (e.g. for
638 memory leaks) to include file/line info for code that has been dlclose'd (or
639 similar). See the user manual for more information and known limitations.
641 * Ability to specify suppressions based on source file name and line number.
643 * Majorly overhauled register allocator. No end-user changes, but the JIT
644 generates code a bit more quickly now.
646 * ================== PLATFORM CHANGES =================
648 * Preliminary support for macOS 10.13 has been added.
650 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
652 * mips: support for MIPS SIMD architecture (MSA) has been added.
654 * mips: support for MIPS N32 ABI has been added.
656 * s390: partial support for vector instructions (integer and string) has been
659 * ==================== TOOL CHANGES ====================
661 * Helgrind: Addition of a flag
662 --delta-stacktrace=no|yes [yes on linux amd64/x86]
663 which specifies how full history stack traces should be computed.
664 Setting this to =yes can speed up Helgrind by 25% when using
665 --history-level=full.
667 * Memcheck: reduced false positive rate for optimised code created by Clang 6
668 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
669 blocks more carefully to determine where it can avoid expensive definedness
670 checks without loss of precision. This is controlled by the flag
671 --expensive-definedness-checks=no|auto|yes [auto].
673 * ==================== OTHER CHANGES ====================
675 * Valgrind is now buildable with link-time optimisation (LTO). A new
676 configure option --enable-lto=yes allows building Valgrind with LTO. If the
677 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
678 Note that if you are doing Valgrind development, --enable-lto=yes massively
679 slows down the build process.
681 * ==================== FIXED BUGS ====================
683 The following bugs have been fixed or resolved. Note that "n-i-bz"
684 stands for "not in bugzilla" -- that is, a bug that was reported to us
685 but never got a bugzilla entry. We encourage you to file bugs in
686 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
687 than mailing the developers (or mailing lists) directly -- bugs that
688 are not entered into bugzilla tend to get forgotten about or ignored.
690 To see details of a given bug, visit
691 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
692 where XXXXXX is the bug number as listed below.
694 79362 Debug info is lost for .so files when they are dlclose'd
695 208052 strlcpy error when n = 0
696 255603 exp-sgcheck Assertion '!already_present' failed
697 338252 building valgrind with -flto (link time optimisation) fails
698 345763 MIPS N32 ABI support
699 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
700 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
701 372347 Replacement problem of the additional c++14/c++17 new/delete operators
702 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
703 376257 helgrind history full speed up using a cached stack
704 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
706 379748 Fix missing pselect syscall (OS X 10.11)
707 379754 Fix missing syscall ulock_wait (OS X 10.12)
708 380397 s390x: __GI_strcspn() replacemenet needed
709 381162 possible array overrun in VEX register allocator
710 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
711 381274 powerpc too chatty even with --sigill-diagnostics=no
712 381289 epoll_pwait can have a NULL sigmask
713 381553 VEX register allocator v3
714 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
715 381769 Use ucontext_t instead of struct ucontext
716 381805 arm32 needs ld.so index hardwire for new glibc security fixes
717 382256 gz compiler flag test doesn't work for gold
718 382407 vg_perf needs "--terse" command line option
719 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
720 382563 MIPS MSA ASE support
721 382998 xml-socket doesn't work
722 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
723 383723 Fix missing kevent_qos syscall (macOS 10.11)
724 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
725 384096 Mention AddrCheck at Memcheck's command line option [..]
726 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
727 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
728 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
729 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
730 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
731 384337 performance improvements to VEX register allocator v2 and v3
732 384526 reduce number of spill insns generated by VEX register allocator v3
733 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
734 384631 Sanitise client args as printed with -v
735 384633 Add a simple progress-reporting facility
736 384987 VEX regalloc: allocate caller-save registers for short lived vregs
737 385055 PPC VEX temporary storage exhausted
738 385182 PPC64 is missing support for the DSCR
739 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
740 385207 PPC64, generate_store_FPRF() generates too many Iops
741 385208 PPC64, xxperm instruction exhausts temporary memory
742 385210 PPC64, vpermr instruction could exhaust temporary memory
743 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
744 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
745 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
746 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
747 385408 s390x: z13 vector "support" instructions not implemented
748 385409 s390x: z13 vector integer instructions not implemented
749 385410 s390x: z13 vector string instructions not implemented
750 385412 s390x: new non-vector z13 instructions not implemented
751 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
752 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
753 385939 Optionally exit on the first error
754 386318 valgrind.org/info/tools.html is missing SGCheck
755 386425 running valgrind + wine on armv7l gives illegal opcode
756 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
757 387410 MIPSr6 support
758 387664 Memcheck: make expensive-definedness-checks be the default
759 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
760 387766 asm shifts cause false positive "Conditional jump or move depends
761 on uninitialised value"
762 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
763 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
764 388786 Support bpf syscall in amd64 Linux
765 388862 Add replacements for wmemchr and wcsnlen on Linux
766 389065 valgrind meets gcc flag -Wlogical-op
767 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
768 390471 suppression by specification of source-file line number
769 390723 make xtree dump files world wide readable, similar to log files
770 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
771 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
772 392118 unhandled amd64-linux syscall: 332 (statx)
773 392449 callgrind not clearing the number of calls properly
774 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
775 lxssp, stxssp and stxvl instructions.
776 393023 callgrind_control risks using the wrong vgdb
777 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
778 393099 posix_memalign() invalid write if alignment == 0
779 393146 failing assert "is_DebugInfo_active(di)"
780 395709 PPC64 is missing support for the xvnegsp instruction
781 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
783 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
784 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
785 396839 s390x: Trap instructions not implemented
786 396887 arch_prctl should return EINVAL on unknown option
787 == 397286 crash before launching binary (Unsupported arch_prctl option)
788 == 397393 valgrind: the 'impossible' happened: (Archlinux)
789 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
790 396906 compile tests failure on mips32-linux: broken inline asm in tests on
792 397012 glibc ld.so uses arch_prctl on i386
793 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
794 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
795 397424 glibc 2.27 and gdb_server tests
796 398028 Assertion `cfsi_fits` failing in simple C program
797 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
799 n-i-bz Fix missing workq_ops operations (macOS)
800 n-i-bz fix bug in strspn replacement
801 n-i-bz Add support for the Linux BLKFLSBUF ioctl
802 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
803 n-i-bz Fix possible stack trashing by semctl syscall wrapping
804 n-i-bz Add support for the Linux membarrier() system call
805 n-i-bz x86 front end: recognise and handle UD2 correctly
806 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
807 correctly aligned before entering the handler.
809 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
810 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
811 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
815 Release 3.13.0 (15 June 2017)
816 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
818 3.13.0 is a feature release with many improvements and the usual collection of
821 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
822 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
823 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
824 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
826 * ==================== CORE CHANGES ===================
828 * The translation cache size has been increased to keep up with the demands of
829 large applications. The maximum number of sectors has increased from 24 to
830 48. The default number of sectors has increased from 16 to 32 on all
831 targets except Android, where the increase is from 6 to 12.
833 * The amount of memory that Valgrind can use has been increased from 64GB to
834 128GB. In particular this means your application can allocate up to about
835 60GB when running on Memcheck.
837 * Valgrind's default load address has been changed from 0x3800'0000 to
838 0x5800'0000, so as to make it possible to load larger executables. This
839 should make it possible to load executables of size at least 1200MB.
841 * A massive spaceleak caused by reading compressed debuginfo files has been
842 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
845 * The C++ demangler has been updated.
847 * Support for demangling Rust symbols has been added.
849 * A new representation of stack traces, the "XTree", has been added. An XTree
850 is a tree of stacktraces with data associated with the stacktraces. This is
851 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
852 consumption of your program. Reporting is controlled by the new options
853 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
855 A report can also be produced on demand using the gdbserver monitor command
856 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
857 format' and 'massif format. The existing visualisers for these formats (e.g.
858 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
859 analyse these reports.
861 Memcheck can also produce XTree leak reports using the Callgrind file
862 format. For more details, see the user manual.
864 * ================== PLATFORM CHANGES =================
866 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
868 * amd64: fixes for JIT failure problems on long AVX2 code blocks
870 * amd64 and x86: support for CET prefixes has been added
872 * arm32: a few missing ARMv8 instructions have been implemented
874 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
875 Store-Conditional instructions has been added. This is to deal with
876 processor implementations that implement the LL/SC specifications strictly
877 and as a result cause Valgrind to hang in certain situations. The
878 alternative implementation is automatically enabled at startup, as required.
879 You can use the option --sim-hints=fallback-llsc to force-enable it if you
882 * Support for OSX 10.12 has been improved.
884 * On Linux, clone handling has been improved to honour CLONE_VFORK that
885 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
886 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
887 CLONE_VM exact semantics will (still) not work.
889 * The TileGX/Linux port has been removed because it appears to be both unused
892 * ==================== TOOL CHANGES ====================
896 - Memcheck should give fewer false positives when running optimised
897 Clang/LLVM generated code.
899 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
901 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
902 to produce the end of execution leak report in a xtree callgrind format
905 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
906 the leak report in an xtree file.
910 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
912 - For some workloads (typically, for big applications), Massif memory
913 consumption and CPU consumption has decreased significantly.
917 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
919 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
920 for Ada gnat compiled applications.
922 * ==================== OTHER CHANGES ====================
924 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
925 append the inner guest stacktrace to the inner host stacktrace. This helps
926 to investigate the errors reported by the outer, when they are caused by the
927 inner guest program (such as an inner regtest). See README_DEVELOPERS for
930 * To allow fast detection of callgrind files by desktop environments and file
931 managers, the format was extended to have an optional first line that
932 uniquely identifies the format ("# callgrind format"). Callgrind creates
933 this line now, as does the new xtree functionality.
935 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
936 have a new %n format letter that is replaced by a sequence number.
938 * "--version -v" now shows the SVN revision numbers from which Valgrind was
941 * ==================== FIXED BUGS ====================
943 The following bugs have been fixed or resolved. Note that "n-i-bz"
944 stands for "not in bugzilla" -- that is, a bug that was reported to us
945 but never got a bugzilla entry. We encourage you to file bugs in
946 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
947 than mailing the developers (or mailing lists) directly -- bugs that
948 are not entered into bugzilla tend to get forgotten about or ignored.
950 To see details of a given bug, visit
951 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
952 where XXXXXX is the bug number as listed below.
954 162848 --log-file output isn't split when a program forks
955 340777 Illegal instruction on mips (ar71xx)
956 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
957 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
958 to a different stack.
959 344139 x86 stack-seg overrides, needed by the Wine people
960 344524 store conditional of guest applications always fail - observed on
962 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
963 352395 Please provide SVN revision info in --version -v
964 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
965 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
966 INVALID_POSIX_THREADID' failed
967 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
968 barrier implementation
969 358697 valgrind.h: Some code remains even when defining NVALGRIND
970 359202 Add musl libc configure/compile
971 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
972 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
973 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
974 362223 assertion failed when .valgrindrc is a directory instead of a file
975 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
976 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
977 368507 can't malloc chunks larger than about 34GB
978 368529 Android arm target link error, missing atexit and pthread_atfork
979 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
980 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
981 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
982 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
983 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
984 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
985 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
986 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
987 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
988 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
989 370028 Reduce the number of compiler warnings on MIPS platforms
990 370635 arm64 missing syscall getcpu
991 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
992 371227 Clean AArch64 syscall table
993 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
994 371471 Valgrind complains about non legit memory leaks on placement new (C++)
995 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
996 371503 disInstr(arm64): unhandled instruction 0xF89F0000
997 371869 support '%' in symbol Z-encoding
998 371916 execution tree xtree concept
999 372120 c++ demangler demangles symbols which are not c++
1000 372185 Support of valgrind on ARMv8 with 32 bit executable
1001 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
1002 372195 Power PC, xxsel instruction is not always recognized.
1003 372504 Hanging on exit_group
1004 372600 process loops forever when fatal signals are arriving quickly
1005 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
1006 373046 Stacks registered by core are never deregistered
1007 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
1008 373086 Implement additional Xen hypercalls
1009 373192 Calling posix_spawn in glibc 2.24 completely broken
1010 373488 Support for fanotify API on ARM64 architecture
1011 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
1012 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
1013 373938 const IRExpr arguments for matchIRExpr()
1014 374719 some spelling fixes
1015 374963 increase valgrind's load address to prevent mmap failure
1016 375514 valgrind_get_tls_addr() does not work in case of static TLS
1017 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
1018 for ML_(find_rx_mapping)()
1019 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
1020 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
1021 == 377159 "vex: the `impossible' happened" still present
1022 == 375150 Assertion 'tres.status == VexTransOK' failed
1023 == 378068 valgrind crashes on AVX2 function in FFmpeg
1024 376142 Segfaults on MIPS Cavium Octeon boards
1025 376279 disInstr(arm64): unhandled instruction 0xD50320FF
1026 376455 Solaris: unhandled syscall lgrpsys(180)
1027 376518 Solaris: unhandled fast trap getlgrp(6)
1028 376611 ppc64 and arm64 don't know about prlimit64 syscall
1029 376729 PPC64, remove R2 from the clobber list
1031 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
1032 to be wrongly marked as addressable
1033 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
1034 PIE enabled by default
1035 377376 memcheck/tests/linux/getregset fails with glibc2.24
1036 377427 PPC64, lxv instruction failing on odd destination register
1037 377478 PPC64: ISA 3.0 setup fixes
1038 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
1039 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
1040 and 2 args for FUTEX_TRYLOCK_PI
1041 377717 Fix massive space leak when reading compressed debuginfo sections
1042 377891 Update Xen 4.6 domctl wrappers
1043 377930 fcntl syscall wrapper is missing flock structure check
1044 378524 libvexmultiarch_test regression on s390x and ppc64
1045 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
1046 378673 Update libiberty demangler
1047 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
1048 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
1049 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
1050 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
1051 (task_register_dyld_image_infos)
1052 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
1053 (task_register_dyld_shared_cache_image_info)
1054 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
1055 379473 MIPS: add support for rdhwr cycle counter register
1056 379504 remove TileGX/Linux port
1057 379525 Support more x86 nop opcodes
1058 379838 disAMode(x86): not an addr!
1059 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
1060 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
1061 379895 clock_gettime does not execute POST syscall wrapper
1062 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
1063 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
1064 380200 xtree generated callgrind files refer to files without directory name
1065 380202 Assertion failure for cache line size (cls == 64) on aarch64.
1066 380397 s390x: __GI_strcspn() replacement needed
1067 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
1069 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
1070 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
1071 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
1075 Release 3.12.0 (20 October 2016)
1076 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1078 3.12.0 is a feature release with many improvements and the usual
1079 collection of bug fixes.
1081 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1082 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1083 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1084 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1085 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1086 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
1088 * ================== PLATFORM CHANGES =================
1090 * POWER: Support for ISA 3.0 has been added
1092 * mips: support for O32 FPXX ABI has been added.
1093 * mips: improved recognition of different processors
1094 * mips: determination of page size now done at run time
1096 * amd64: Partial support for AMD FMA4 instructions.
1098 * arm, arm64: Support for v8 crypto and CRC instructions.
1100 * Improvements and robustification of the Solaris port.
1102 * Preliminary support for MacOS 10.12 (Sierra) has been added.
1104 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
1105 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
1106 where possible. Valgrind's support for 32-bit x86 has stagnated in
1107 recent years and has fallen far behind that for 64-bit x86
1108 instructions. By contrast 64-bit x86 is well supported, up to and
1111 * ==================== TOOL CHANGES ====================
1115 - Added meta mempool support for describing a custom allocator which:
1116 - Auto-frees all chunks assuming that destroying a pool destroys all
1118 - Uses itself to allocate other memory blocks
1120 - New flag --ignore-range-below-sp to ignore memory accesses below
1121 the stack pointer, if you really have to. The related flag
1122 --workaround-gcc296-bugs=yes is now deprecated. Use
1123 --ignore-range-below-sp=1024-1 as a replacement.
1127 - Improved thread startup time significantly on non-Linux platforms.
1131 - Added collection of the metric "tot-blocks-allocd"
1133 * ==================== OTHER CHANGES ====================
1135 * Replacement/wrapping of malloc/new related functions is now done not just
1136 for system libraries by default, but for any globally defined malloc/new
1137 related function (both in shared libraries and statically linked alternative
1138 malloc implementations). The dynamic (runtime) linker is excluded, though.
1139 To only intercept malloc/new related functions in
1140 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
1141 "nouserintercepts" can be any non-existing library name).
1142 This new functionality is not implemented for MacOS X.
1144 * The maximum number of callers in a suppression entry is now equal to
1145 the maximum size for --num-callers (500).
1146 Note that --gen-suppressions=yes|all similarly generates suppressions
1147 containing up to --num-callers frames.
1149 * New and modified GDB server monitor features:
1151 - Valgrind's gdbserver now accepts the command 'catch syscall'.
1152 Note that you must have GDB >= 7.11 to use 'catch syscall' with
1155 * New option --run-cxx-freeres=<yes|no> can be used to change whether
1156 __gnu_cxx::__freeres() cleanup function is called or not. Default is
1159 * Valgrind is able to read compressed debuginfo sections in two formats:
1160 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
1161 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
1163 * Modest JIT-cost improvements: the cost of instrumenting code blocks
1164 for the most common use case (x86_64-linux, Memcheck) has been
1167 * Improved performance for programs that do a lot of discarding of
1168 instruction address ranges of 8KB or less.
1170 * The C++ symbol demangler has been updated.
1172 * More robustness against invalid syscall parameters on Linux.
1174 * ==================== FIXED BUGS ====================
1176 The following bugs have been fixed or resolved. Note that "n-i-bz"
1177 stands for "not in bugzilla" -- that is, a bug that was reported to us
1178 but never got a bugzilla entry. We encourage you to file bugs in
1179 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1180 than mailing the developers (or mailing lists) directly -- bugs that
1181 are not entered into bugzilla tend to get forgotten about or ignored.
1183 To see details of a given bug, visit
1184 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1185 where XXXXXX is the bug number as listed below.
1187 191069 Exiting due to signal not reported in XML output
1188 199468 Suppressions: stack size limited to 25
1189 while --num-callers allows more frames
1190 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
1191 278744 cvtps2pd with redundant RexW
1192 303877 valgrind doesn't support compressed debuginfo sections.
1193 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
1194 348345 Assertion fails for negative lineno
1195 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
1196 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
1197 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
1198 351804 Crash on generating suppressions for "printf" call on OS X 10.10
1199 352197 mips: mmap2() not wrapped correctly for page size > 4096
1200 353083 arm64 doesn't implement various xattr system calls
1201 353084 arm64 doesn't support sigpending system call
1202 353137 www: update info for Supported Platforms
1203 353138 www: update "The Valgrind Developers" page
1204 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
1207 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
1208 353398 WARNING: unhandled amd64-solaris syscall: 207
1209 353660 XML in auxwhat tag not escaping reserved symbols properly
1210 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1211 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1212 353802 ELF debug info reader confused with multiple .rodata sections
1213 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1214 353917 unhandled amd64-solaris syscall fchdir(120)
1215 353920 unhandled amd64-solaris syscall: 170
1216 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1217 354392 unhandled amd64-solaris syscall: 171
1218 354797 Vbit test does not include Iops for Power 8 instruction support
1219 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1223 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1224 355188 valgrind should intercept all malloc related global functions
1225 355454 do not intercept malloc related symbols from the runtime linker
1226 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1227 356044 Dwarf line info reader misinterprets is_stmt register
1228 356112 mips: replace addi with addiu
1229 356393 valgrind (vex) crashes because isZeroU happened
1232 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1233 356678 arm64-linux: unhandled syscall 232 (mincore)
1234 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1235 356823 Unsupported ARM instruction: stlex
1236 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1237 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1238 357673 crash if I try to run valgrind with a binary link with libcurl
1239 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1240 357871 pthread_spin_destroy not properly wrapped
1241 357887 Calls to VG_(fclose) do not close the file descriptor
1242 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1243 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1244 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1245 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1246 359181 Buffer Overflow during Demangling
1247 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1248 359289 s390x: popcnt (B9E1) not implemented
1249 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1250 359503 Add missing syscalls for aarch64 (arm64)
1251 359645 "You need libc6-dbg" help message could be more helpful
1252 359703 s390: wire up separate socketcalls system calls
1253 359724 getsockname might crash - deref_UInt should call safe_to_deref
1254 359733 amd64 implement ld.so strchr/index override like x86
1255 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1256 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1258 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1259 359871 Incorrect mask handling in ppoll
1260 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1261 360008 Contents of Power vr registers contents is not printed correctly when
1262 the --vgdb-shadow-registers=yes option is used
1263 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1264 non-zero shadow bits
1265 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1266 360425 arm64 unsupported instruction ldpsw
1268 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1269 360571 Error about the Android Runtime reading below the stack pointer on ARM
1270 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1271 360749 kludge for multiple .rodata sections on Solaris no longer needed
1272 360752 raise the number of reserved fds in m_main.c from 10 to 12
1273 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1274 361226 s390x: risbgn (EC59) not implemented
1275 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1276 361354 ppc64[le]: wire up separate socketcalls system calls
1277 361615 Inconsistent termination for multithreaded process terminated by signal
1278 361926 Unhandled Solaris syscall: sysfs(84)
1279 362009 V dumps core on unimplemented functionality before threads are created
1280 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1281 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1282 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1283 362953 Request for an update to the Valgrind Developers page
1284 363680 add renameat2() support
1285 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1286 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1287 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1288 364058 clarify in manual limitations of array overruns detections
1289 364413 pselect sycallwrapper mishandles NULL sigmask
1290 364728 Power PC, missing support for several HW registers in
1291 get_otrack_shadow_offset_wrk()
1292 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1293 365273 Invalid write to stack location reported after signal handler runs
1294 365912 ppc64BE segfault during jm-insns test (RELRO)
1295 366079 FPXX Support for MIPS32 Valgrind
1296 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1297 366344 Multiple unhandled instruction for Aarch64
1298 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1299 367995 Integration of memcheck with custom memory allocator
1300 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1301 368412 False positive result for altivec capability check
1302 368416 Add tc06_two_races_xml.exp output for ppc64
1303 368419 Perf Events ioctls not implemented
1304 368461 mmapunmap test fails on ppc64
1305 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1306 369000 AMD64 fma4 instructions unsupported.
1307 369169 ppc64 fails jm_int_isa_2_07 test
1308 369175 jm_vec_isa_2_07 test crashes on ppc64
1309 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1310 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1311 369359 msghdr_foreachfield can crash when handling bad iovec
1312 369360 Bad sigprocmask old or new sets can crash valgrind
1313 369361 vmsplice syscall wrapper crashes on bad iovec
1314 369362 Bad sigaction arguments crash valgrind
1315 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1316 369402 Bad set/get_thread_area pointer crashes valgrind
1317 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1318 369446 valgrind crashes on unknown fcntl command
1319 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1320 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1321 370265 ISA 3.0 HW cap stuff needs updating
1322 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1323 372195 Power PC, xxsel instruction is not always recognized
1325 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1326 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1327 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1328 n-i-bz Fix ppoll_alarm exclusion on OS X
1329 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1330 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1331 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1332 n-i-bz amd64: memcheck false positive with shr %edx
1333 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1334 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1335 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1336 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1338 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1339 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1340 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1344 Release 3.11.0 (22 September 2015)
1345 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1347 3.11.0 is a feature release with many improvements and the usual
1348 collection of bug fixes.
1350 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1351 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1352 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1353 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1354 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1355 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1357 * ================== PLATFORM CHANGES =================
1359 * Support for Solaris/x86 and Solaris/amd64 has been added.
1361 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1363 * Preliminary support for the Tilera TileGX architecture has been added.
1365 * s390x: It is now required for the host to have the "long displacement"
1366 facility. The oldest supported machine model is z990.
1368 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1369 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1371 * The JIT's register allocator is significantly faster, making the JIT
1372 as a whole somewhat faster, so JIT-intensive activities, for example
1373 program startup, are modestly faster, around 5%.
1375 * There have been changes to the default settings of several command
1376 line flags, as detailed below.
1378 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1379 capable hosts, the simulated CPUID will now indicate AVX2 support.
1381 * ==================== TOOL CHANGES ====================
1385 - The default value for --leak-check-heuristics has been changed from
1386 "none" to "all". This helps to reduce the number of possibly
1387 lost blocks, in particular for C++ applications.
1389 - The default value for --keep-stacktraces has been changed from
1390 "malloc-then-free" to "malloc-and-free". This has a small cost in
1391 memory (one word per malloc-ed block) but allows Memcheck to show the
1392 3 stacktraces of a dangling reference: where the block was allocated,
1393 where it was freed, and where it is acccessed after being freed.
1395 - The default value for --partial-loads-ok has been changed from "no" to
1396 "yes", so as to avoid false positive errors resulting from some kinds
1397 of vectorised loops.
1399 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1400 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1401 than get_vbits when you need to associate byte data value with
1402 their corresponding validity bits.
1404 - The 'block_list' monitor command has been enhanced:
1405 o it can print a range of loss records
1406 o it now accepts an optional argument 'limited <max_blocks>'
1407 to control the number of blocks printed.
1408 o if a block has been found using a heuristic, then
1409 'block_list' now shows the heuristic after the block size.
1410 o the loss records/blocks to print can be limited to the blocks
1411 found via specified heuristics.
1413 - The C helper functions used to instrument loads on
1414 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1415 replaced by handwritten assembly sequences. This gives speedups
1416 in the region of 0% to 7% for those targets only.
1418 - A new command line option, --expensive-definedness-checks=yes|no,
1419 has been added. This is useful for avoiding occasional invalid
1420 uninitialised-value errors in optimised code. Watch out for
1421 runtime degradation, as this can be up to 25%. As always, though,
1422 the slowdown is highly application specific. The default setting
1427 - A new monitor command 'all_snapshots <filename>' dumps all
1428 snapshots taken so far.
1432 - Significant memory reduction and moderate speedups for
1433 --history-level=full for applications accessing a lot of memory
1434 with many different stacktraces.
1436 - The default value for --conflict-cache-size=N has been doubled to
1437 2000000. Users that were not using the default value should
1438 preferably also double the value they give.
1440 The default was changed due to the changes in the "full history"
1441 implementation. Doubling the value gives on average a slightly more
1442 complete history and uses similar memory (or significantly less memory
1443 in the worst case) than the previous implementation.
1445 - The Helgrind monitor command 'info locks' now accepts an optional
1446 argument 'lock_addr', which shows information about the lock at the
1449 - When using --history-level=full, the new Helgrind monitor command
1450 'accesshistory <addr> [<len>]' will show the recorded accesses for
1451 <len> (or 1) bytes at <addr>.
1453 * ==================== OTHER CHANGES ====================
1455 * The default value for the --smc-check option has been changed from
1456 "stack" to "all-non-file" on targets that provide automatic D-I
1457 cache coherence (x86, amd64 and s390x). The result is to provide,
1458 by default, transparent support for JIT generated and self-modifying
1459 code on all targets.
1461 * Mac OS X only: the default value for the --dsymutil option has been
1462 changed from "no" to "yes", since any serious usage on Mac OS X
1463 always required it to be "yes".
1465 * The command line options --db-attach and --db-command have been removed.
1466 They were deprecated in 3.10.0.
1468 * When a process dies due to a signal, Valgrind now shows the signal
1469 and the stacktrace at default verbosity (i.e. verbosity 1).
1471 * The address description logic used by Memcheck and Helgrind now
1472 describes addresses in anonymous segments, file mmap-ed segments,
1473 shared memory segments and the brk data segment.
1475 * The new option --error-markers=<begin>,<end> can be used to mark the
1476 begin/end of errors in textual output mode, to facilitate
1477 searching/extracting errors in output files that mix valgrind errors
1478 with program output.
1480 * The new option --max-threads=<number> can be used to change the number
1481 of threads valgrind can handle. The default is 500 threads which
1482 should be more than enough for most applications.
1484 * The new option --valgrind-stacksize=<number> can be used to change the
1485 size of the private thread stacks used by Valgrind. This is useful
1486 for reducing memory use or increasing the stack size if Valgrind
1487 segfaults due to stack overflow.
1489 * The new option --avg-transtab-entry-size=<number> can be used to specify
1490 the expected instrumented block size, either to reduce memory use or
1491 to avoid excessive retranslation.
1493 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1495 * New and modified GDB server monitor features:
1497 - When a signal is reported in GDB, you can now use the GDB convenience
1498 variable $_siginfo to examine detailed signal information.
1500 - Valgrind's gdbserver now allows the user to change the signal
1501 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1502 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1503 continue without passing the signal to the process.
1505 - With GDB >= 7.10, the command 'target remote'
1506 will automatically load the executable file of the process running
1507 under Valgrind. This means you do not need to specify the executable
1508 file yourself, GDB will discover it itself. See GDB documentation about
1509 'qXfer:exec-file:read' packet for more info.
1511 * ==================== FIXED BUGS ====================
1513 The following bugs have been fixed or resolved. Note that "n-i-bz"
1514 stands for "not in bugzilla" -- that is, a bug that was reported to us
1515 but never got a bugzilla entry. We encourage you to file bugs in
1516 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1517 than mailing the developers (or mailing lists) directly -- bugs that
1518 are not entered into bugzilla tend to get forgotten about or ignored.
1520 To see details of a given bug, visit
1521 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1522 where XXXXXX is the bug number as listed below.
1524 116002 VG_(printf): Problems with justification of strings and integers
1525 155125 avoid cutting away file:lineno after long function name
1526 197259 Unsupported arch_prtctl PR_SET_GS option
1527 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
1528 201216 Fix Valgrind does not support pthread_sigmask() on OS X
1529 201435 Fix Darwin: -v does not show kernel version
1530 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
1531 211256 Fixed an outdated comment regarding the default platform.
1532 211529 Incomplete call stacks for code compiled by newer versions of MSVC
1533 211926 Avoid compilation warnings in valgrind.h with -pedantic
1534 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
1536 226609 Crediting upstream authors in man page
1537 231257 Valgrind omits path when executing script from shebang line
1538 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
1539 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
1540 269360 s390x: Fix addressing mode selection for compare-and-swap
1541 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
1543 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
1544 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
1545 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
1546 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
1547 330147 libmpiwrap PMPI_Get_count returns undefined value
1548 333051 mmap of huge pages fails due to incorrect alignment
1550 334802 valgrind does not always explain why a given option is bad
1551 335618 mov.w rN, pc/sp (ARM32)
1552 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
1557 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
1558 338602 AVX2 bit in CPUID missing
1559 338606 Strange message for scripts with invalid interpreter
1560 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
1561 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
1562 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
1564 339156 gdbsrv not called for fatal signal
1565 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
1566 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
1567 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
1568 339442 Fix testsuite build failure on OS X 10.9
1569 339542 Enable compilation with Intel's ICC compiler
1570 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
1571 339688 Mac-specific ASM does not support .version directive (cpuid,
1572 tronical and pushfpopf tests)
1573 339745 Valgrind crash when check Marmalade app (partial fix)
1574 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
1575 339778 Linux/TileGx platform support for Valgrind
1576 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
1577 339789 Fix none/tests/execve test on Mac OS X 10.9
1578 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
1579 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
1580 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
1581 340392 Allow user to select more accurate definedness checking in memcheck
1582 to avoid invalid complaints on optimised code
1583 340430 Fix some grammatical weirdness in the manual.
1584 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
1585 341419 Signal handler ucontext_t not filled out correctly on OS X
1586 341539 VG_(describe_addr) should not describe address as belonging to client
1587 segment if it is past the heap end
1588 341613 Enable building of manythreads and thread-exits tests on Mac OS X
1589 341615 Fix none/tests/darwin/access_extended test on Mac OS X
1590 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
1591 341789 aarch64: shmat fails with valgrind on ARMv8
1592 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
1593 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
1594 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
1595 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
1596 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
1597 342221 socket connect false positive uninit memory for unknown af family
1598 342353 Allow dumping full massif output while valgrind is still running
1599 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
1603 342603 Add I2C_SMBUS ioctl support
1604 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
1605 342683 Mark memory past the initial brk limit as unaddressable
1606 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
1607 342795 Internal glibc __GI_mempcpy call should be intercepted
1608 342841 s390x: Support instructions fiebr(a) and fidbr(a)
1609 343012 Unhandled syscall 319 (memfd_create)
1610 343069 Patch updating v4l2 API support
1611 343173 helgrind crash during stack unwind
1612 343219 fix GET_STARTREGS for arm
1613 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
1614 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
1615 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
1616 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
1617 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
1618 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
1619 343597 ppc64le: incorrect use of offseof macro
1620 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
1621 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
1622 343732 Unhandled syscall 144 (setgid) on aarch64
1623 343733 Unhandled syscall 187 (msgctl and related) on aarch64
1624 343802 s390x: False positive "conditional jump or move depends on [..]
1625 343902 --vgdb=yes doesn't break when --xml=yes is used
1626 343967 Don't warn about setuid/setgid/setcap executable for directories
1627 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
1628 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
1629 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
1630 344054 www - update info for Solaris/illumos
1631 344416 'make regtest' does not work cleanly on OS X
1632 344235 Remove duplicate include of pub_core_aspacemgr.h
1633 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
1634 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
1635 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
1636 344314 callgrind_annotate ... warnings about commands containing newlines
1637 344318 socketcall should wrap recvmmsg and sendmmsg
1638 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
1639 344416 Fix 'make regtest' does not work cleanly on OS X
1640 344499 Fix compilation for Linux kernel >= 4.0.0
1641 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
1642 unix:349 (__pthread_fchdir)
1643 344559 Garbage collection of unused segment names in address space manager
1644 344560 Fix stack traces missing penultimate frame on OS X
1645 344621 Fix memcheck/tests/err_disable4 test on OS X
1646 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
1647 344702 Fix missing libobjc suppressions on OS X 10.10
1649 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
1650 344939 Fix memcheck/tests/xml1 on OS X 10.10
1651 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
1652 345079 Fix build problems in VEX/useful/test_main.c
1653 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
1654 345177 arm64: prfm (reg) not implemented
1655 345215 Performance improvements for the register allocator
1656 345248 add support for Solaris OS in valgrind
1657 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
1658 345394 Fix memcheck/tests/strchr on OS X
1659 345637 Fix memcheck/tests/sendmsg on OS X
1660 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
1661 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
1662 345887 Fix an assertion in the address space manager
1663 345928 amd64: callstack only contains current function for small stacks
1664 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1665 345987 MIPS64: Implement cavium LHX instruction
1666 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1667 346185 Fix typo saving altivec register v24
1668 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1669 and LibVEX_GuestPPC64_get_CR()
1670 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1671 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1672 346307 fuse filesystem syscall deadlocks
1673 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1674 346411 MIPS: SysRes::_valEx handling is incorrect
1675 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1676 346474 PPC64 Power 8, spr TEXASRU register not supported
1677 346487 Compiler generates "note" about a future ABI change for PPC64
1678 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1679 and causing spurious "invalid read of size 8" warnings
1680 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1681 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1682 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1683 347322 Power PC regression test cleanup
1684 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1686 347389 unhandled syscall: 373 (Linux ARM syncfs)
1687 347686 Patch set to cleanup PPC64 regtests
1688 347978 Remove bash dependencies where not needed
1689 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1690 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1692 348102 Patch updating v4l2 API support
1693 348247 amd64 front end: jno jumps wrongly when overflow is not set
1694 348269 Improve mmap MAP_HUGETLB support.
1695 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1696 348345 Assertion fails for negative lineno
1697 348377 Unsupported ARM instruction: yield
1698 348565 Fix detection of command line option availability for clang
1699 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1700 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1701 348748 Fix redundant condition
1702 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1703 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1704 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1705 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1706 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1707 349626 Implemented additional Xen hypercalls
1708 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1709 349790 Clean up of the hardware capability checking utilities.
1710 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1711 349874 Fix typos in source code
1712 349879 memcheck: add handwritten assembly for helperc_LOADV*
1713 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1714 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1715 350202 Add limited param to 'monitor block_list'
1716 350290 s390x: Support instructions fixbr(a)
1717 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1718 350809 Fix none/tests/async-sigs for Solaris
1719 350811 Remove reference to --db-attach which has been removed.
1720 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1721 350854 hard-to-understand code in VG_(load_ELF)()
1722 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1723 351386 Solaris: Cannot run ld.so.1 under Valgrind
1724 351474 Fix VG_(iseqsigset) as obvious
1725 351531 Typo in /include/vki/vki-xen-physdev.h header guard
1726 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
1727 351858 ldsoexec support on Solaris
1728 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
1729 352130 helgrind reports false races for printfs using mempcpy on FILE* state
1730 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
1731 352320 arm64 crash on none/tests/nestedfs
1732 352765 Vbit test fails on Power 6
1733 352768 The mbar instruction is missing from the Power PC support
1734 352769 Power PC program priority register (PPR) is not supported
1735 n-i-bz Provide implementations of certain compiler builtins to support
1736 compilers that may not provide those
1737 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
1738 n-i-bz Fix compilation on distros with glibc < 2.5
1739 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
1740 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
1741 n-i-bz mremap did not work properly on shared memory
1742 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
1743 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
1745 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
1746 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
1747 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
1751 Release 3.10.1 (25 November 2014)
1752 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1753 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
1754 and backports fixes for all reported missing AArch64 ARMv8 instructions
1755 and syscalls from the trunk. If you package or deliver 3.10.0 for others
1756 to use, you might want to consider upgrading to 3.10.1 instead.
1758 The following bugs have been fixed or resolved. Note that "n-i-bz"
1759 stands for "not in bugzilla" -- that is, a bug that was reported to us
1760 but never got a bugzilla entry. We encourage you to file bugs in
1761 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1762 than mailing the developers (or mailing lists) directly -- bugs that
1763 are not entered into bugzilla tend to get forgotten about or ignored.
1765 To see details of a given bug, visit
1766 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1767 where XXXXXX is the bug number as listed below.
1769 335440 arm64: ld1 (single structure) is not implemented
1770 335713 arm64: unhanded instruction: prfm (immediate)
1771 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
1772 339182 ppc64: AvSplat ought to load destination vector register with [..]
1773 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
1774 339433 ppc64 lxvw4x instruction uses four 32-byte loads
1775 339645 Use correct tag names in sys_getdents/64 wrappers
1776 339706 Fix false positive for ioctl(TIOCSIG) on linux
1777 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
1778 339853 arm64 times syscall unknown
1779 339855 arm64 unhandled getsid/setsid syscalls
1780 339858 arm64 dmb sy not implemented
1781 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
1782 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
1783 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
1785 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
1786 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
1787 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
1788 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
1789 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
1790 340509 arm64: unhandled instruction fcvtas
1791 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
1792 340632 arm64: unhandled instruction fcvtas
1793 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
1794 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
1795 340788 warning: unhandled syscall: 318 (getrandom)
1796 340807 disInstr(arm): unhandled instruction: 0xEE989B20
1797 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
1798 340922 arm64: unhandled getgroups/setgroups syscalls
1799 350251 Fix typo in VEX utility program (test_main.c).
1800 350407 arm64: unhandled instruction ucvtf (vector, integer)
1801 350809 none/tests/async-sigs breaks when run under cron on Solaris
1802 350811 update README.solaris after r15445
1803 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
1804 350854 strange code in VG_(load_ELF)()
1805 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1806 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
1807 n-i-bz Add missing ]] to terminate CDATA.
1808 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
1809 n-i-bz Enable sys_fadvise64_64 on arm32.
1810 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
1811 n-i-bz Add test cases for all known arm64 load/store instructions.
1812 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
1813 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
1814 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
1815 n-i-bz arm64: Implement "fcvtpu w, s".
1816 n-i-bz arm64: implement ADDP and various others
1817 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
1818 n-i-bz arm64: enable FCVT{A,N}S X,S.
1820 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
1824 Release 3.10.0 (10 September 2014)
1825 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1827 3.10.0 is a feature release with many improvements and the usual
1828 collection of bug fixes.
1830 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1831 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1832 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
1833 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
1834 significantly improved relative to the 3.9.0 release.
1836 * ================== PLATFORM CHANGES =================
1838 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
1839 is mostly complete, and is usable, but some SIMD instructions are as
1842 * Support for little-endian variant of the 64-bit POWER architecture.
1844 * Support for Android on MIPS32.
1846 * Support for 64bit FPU on MIPS32 platforms.
1848 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
1850 * Configuration for and running on Android targets has changed.
1851 See README.android in the source tree for details.
1853 * ================== DEPRECATED FEATURES =================
1855 * --db-attach is now deprecated and will be removed in the next
1856 valgrind feature release. The built-in GDB server capabilities are
1857 superior and should be used instead. Learn more here:
1858 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
1860 * ==================== TOOL CHANGES ====================
1864 - Client code can now selectively disable and re-enable reporting of
1865 invalid address errors in specific ranges using the new client
1866 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
1867 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
1869 - Leak checker: there is a new leak check heuristic called
1870 "length64". This is used to detect interior pointers pointing 8
1871 bytes inside a block, on the assumption that the first 8 bytes
1872 holds the value "block size - 8". This is used by
1873 sqlite3MemMalloc, for example.
1875 - Checking of system call parameters: if a syscall parameter
1876 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
1877 several fields not initialised, an error is now reported for each
1878 field. Previously, an error was reported only for the first
1879 uninitialised field.
1881 - Mismatched alloc/free checking: a new flag
1882 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
1883 such checks if necessary.
1887 - Improvements to error messages:
1889 o Race condition error message involving heap allocated blocks also
1890 show the thread number that allocated the raced-on block.
1892 o All locks referenced by an error message are now announced.
1893 Previously, some error messages only showed the lock addresses.
1895 o The message indicating where a lock was first observed now also
1896 describes the address/location of the lock.
1898 - Helgrind now understands the Ada task termination rules and
1899 creates a happens-before relationship between a terminated task
1900 and its master. This avoids some false positives and avoids a big
1901 memory leak when a lot of Ada tasks are created and terminated.
1902 The interceptions are only activated with forthcoming releases of
1903 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
1905 - A new GDB server monitor command "info locks" giving the list of
1906 locks, their location, and their status.
1910 - callgrind_control now supports the --vgdb-prefix argument,
1911 which is needed if valgrind was started with this same argument.
1913 * ==================== OTHER CHANGES ====================
1915 * Unwinding through inlined function calls. Stack unwinding can now
1916 make use of Dwarf3 inlined-unwind information if it is available.
1917 The practical effect is that inlined calls become visible in stack
1918 traces. The suppression matching machinery has been adjusted
1919 accordingly. This is controlled by the new option
1920 --read-inline-info=yes|no. Currently this is enabled by default
1921 only on Linux and Android targets and only for the tools Memcheck,
1924 * Valgrind can now read EXIDX unwind information on 32-bit ARM
1925 targets. If an object contains both CFI and EXIDX unwind
1926 information, Valgrind will prefer the CFI over the EXIDX. This
1927 facilitates unwinding through system libraries on arm-android
1930 * Address description logic has been improved and is now common
1931 between Memcheck and Helgrind, resulting in better address
1932 descriptions for some kinds of error messages.
1934 * Error messages about dubious arguments (eg, to malloc or calloc) are
1935 output like other errors. This means that they can be suppressed
1936 and they have a stack trace.
1938 * The C++ demangler has been updated for better C++11 support.
1940 * New and modified GDB server monitor features:
1942 - Thread local variables/storage (__thread) can now be displayed.
1944 - The GDB server monitor command "v.info location <address>"
1945 displays information about an address. The information produced
1946 depends on the tool and on the options given to valgrind.
1947 Possibly, the following are described: global variables, local
1948 (stack) variables, allocated or freed blocks, ...
1950 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
1951 ask the GDB server to stop at the start of program execution, at
1952 the end of the program execution and on Valgrind internal errors.
1954 - A new monitor command "v.info stats" shows various Valgrind core
1955 and tool statistics.
1957 - A new monitor command "v.set hostvisibility" allows the GDB server
1958 to provide access to Valgrind internal host status/memory.
1960 * A new option "--aspace-minaddr=<address>" can in some situations
1961 allow the use of more memory by decreasing the address above which
1962 Valgrind maps memory. It can also be used to solve address
1963 conflicts with system libraries by increasing the default value.
1964 See user manual for details.
1966 * The amount of memory used by Valgrind to store debug info (unwind
1967 info, line number information and symbol data) has been
1968 significantly reduced, even though Valgrind now reads more
1969 information in order to support unwinding of inlined function calls.
1971 * Dwarf3 handling with --read-var-info=yes has been improved:
1973 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
1975 - Code compiled with
1976 -ffunction-sections -fdata-sections -Wl,--gc-sections
1977 no longer causes assertion failures.
1979 * Improved checking for the --sim-hints= and --kernel-variant=
1980 options. Unknown strings are now detected and reported to the user
1983 * The semantics of stack start/end boundaries in the valgrind.h
1984 VALGRIND_STACK_REGISTER client request has been clarified and
1985 documented. The convention is that start and end are respectively
1986 the lowest and highest addressable bytes of the stack.
1988 * ==================== FIXED BUGS ====================
1990 The following bugs have been fixed or resolved. Note that "n-i-bz"
1991 stands for "not in bugzilla" -- that is, a bug that was reported to us
1992 but never got a bugzilla entry. We encourage you to file bugs in
1993 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1994 than mailing the developers (or mailing lists) directly -- bugs that
1995 are not entered into bugzilla tend to get forgotten about or ignored.
1997 To see details of a given bug, visit
1998 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1999 where XXXXXX is the bug number as listed below.
2001 175819 Support for ipv6 socket reporting with --track-fds
2002 232510 make distcheck fails
2003 249435 Analyzing wine programs with callgrind triggers a crash
2004 278972 support for inlined function calls in stacktraces and suppression
2006 291310 FXSAVE instruction marks memory as undefined on amd64
2007 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
2008 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
2009 315199 vgcore file for threaded app does not show which thread crashed
2010 315952 tun/tap ioctls are not supported
2011 323178 Unhandled instruction: PLDW register (ARM)
2012 323179 Unhandled instruction: PLDW immediate (ARM)
2013 324050 Helgrind: SEGV because of unaligned stack when using movdqa
2014 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
2015 325124 [MIPSEL] Compilation error
2016 325477 Phase 4 support for IBM Power ISA 2.07
2017 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
2018 325628 Phase 5 support for IBM Power ISA 2.07
2019 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
2020 325751 Missing the two privileged Power PC Transactional Memory Instructions
2021 325816 Phase 6 support for IBM Power ISA 2.07
2022 325856 Make SGCheck fail gracefully on unsupported platforms
2023 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
2024 326436 DRD: False positive in libstdc++ std::list::push_back
2025 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
2026 326462 Refactor vgdb to isolate invoker stuff into separate module
2027 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
2028 326623 DRD: false positive conflict report in a field assignment
2029 326724 Valgrind does not compile on OSX 1.9 Mavericks
2030 326816 Intercept for __strncpy_sse2_unaligned missing?
2031 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
2032 326983 Clear direction flag after tests on amd64.
2033 327212 Do not prepend the current directory to absolute path names.
2034 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
2035 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
2036 327284 s390x: Fix translation of the risbg instruction
2037 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
2038 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
2039 327916 DW_TAG_typedef may have no name
2040 327943 s390x: add a redirection for the 'index' function
2041 328100 XABORT not implemented
2042 328205 Implement additional Xen hypercalls
2043 328454 add support Backtraces with ARM unwind tables (EXIDX)
2044 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
2045 328711 valgrind.1 manpage "memcheck options" section is badly generated
2046 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
2047 329612 Incorrect handling of AT_BASE for image execution
2048 329694 clang warns about using uninitialized variable
2049 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
2050 330228 mmap must align to VKI_SHMLBA on mips32
2051 330257 LLVM does not support `-mno-dynamic-no-pic` option
2052 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
2053 330459 --track-fds=yes doesn't track eventfds
2054 330469 Add clock_adjtime syscall support
2055 330594 Missing sysalls on PowerPC / uClibc
2056 330622 Add test to regression suite for POWER instruction: dcbzl
2057 330939 Support for AMD's syscall instruction on x86
2059 330941 Typo in PRE(poll) syscall wrapper
2060 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
2061 331254 Fix expected output for memcheck/tests/dw4
2062 331255 Fix race condition in test none/tests/coolo_sigaction
2063 331257 Fix type of jump buffer in test none/tests/faultstatus
2064 331305 configure uses bash specific syntax
2065 331337 s390x WARNING: unhandled syscall: 326 (dup3)
2066 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
2067 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
2068 331829 Unexpected ioctl opcode sign extension
2069 331830 ppc64: WARNING: unhandled syscall: 96/97
2070 331839 drd/tests/sem_open specifies invalid semaphore name
2071 331847 outcome of drd/tests/thread_name is nondeterministic
2072 332037 Valgrind cannot handle Thumb "add pc, reg"
2073 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
2074 consistency checks enabled
2075 332263 intercepts for pthread_rwlock_timedrdlock and
2076 pthread_rwlock_timedwrlock are incorrect
2077 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
2079 332276 Implement additional Xen hypercalls
2080 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
2081 332765 Fix ms_print to create temporary files in a proper directory
2082 333072 drd: Add semaphore annotations
2083 333145 Tests for missaligned PC+#imm access for arm
2084 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
2085 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
2086 333248 WARNING: unhandled syscall: unix:443
2087 333428 ldr.w pc [rD, #imm] instruction leads to assertion
2088 333501 cachegrind: assertion: Cache set count is not a power of two.
2091 333666 Recognize MPX instructions and bnd prefix.
2092 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
2093 333817 Valgrind reports the memory areas written to by the SG_IO
2095 334049 lzcnt fails silently (x86_32)
2096 334384 Valgrind does not have support Little Endian support for
2098 334585 recvmmsg unhandled (+patch) (arm)
2099 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
2100 334727 Build fails with -Werror=format-security
2101 334788 clarify doc about --log-file initial program directory
2102 334834 PPC64 Little Endian support, patch 2
2103 334836 PPC64 Little Endian support, patch 3 testcase fixes
2104 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
2105 335034 Unhandled ioctl: HCIGETDEVLIST
2106 335155 vgdb, fix error print statement.
2107 335262 arm64: movi 8bit version is not supported
2108 335263 arm64: dmb instruction is not implemented
2109 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
2110 335496 arm64: sbc/abc instructions are not implemented
2111 335554 arm64: unhandled instruction: abs
2112 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
2113 335735 arm64: unhandled instruction: cnt
2114 335736 arm64: unhandled instruction: uaddlv
2115 335848 arm64: unhandled instruction: {s,u}cvtf
2116 335902 arm64: unhandled instruction: sli
2117 335903 arm64: unhandled instruction: umull (vector)
2118 336055 arm64: unhandled instruction: mov (element)
2119 336062 arm64: unhandled instruction: shrn{,2}
2120 336139 mip64: [...] valgrind hangs and spins on a single core [...]
2121 336189 arm64: unhandled Instruction: mvn
2122 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
2123 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
2124 336772 Make moans about unknown ioctls more informative
2125 336957 Add a section about the Solaris/illumos port on the webpage
2126 337094 ifunc wrapper is broken on ppc64
2127 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
2128 337528 leak check heuristic for block prefixed by length as 64bit number
2129 337740 Implement additional Xen hypercalls
2130 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
2131 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
2132 337871 deprecate --db-attach
2133 338023 Add support for all V4L2/media ioctls
2134 338024 inlined functions are not shown if DW_AT_ranges is used
2135 338106 Add support for 'kcmp' syscall
2136 338115 DRD: computed conflict set differs from actual after fork
2137 338160 implement display of thread local storage in gdbsrv
2138 338205 configure.ac and check for -Wno-tautological-compare
2139 338300 coredumps are missing one byte of every segment
2140 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
2141 338499 --sim-hints parsing broken due to wrong order in tokens
2142 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
2143 338681 Unable to unwind through clone thread created on i386-linux
2144 338698 race condition between gdbsrv and vgdb on startup
2145 338703 helgrind on arm-linux gets false positives in dynamic loader
2146 338791 alt dwz files can be relative of debug/main file
2147 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
2148 338932 build V-trunk with gcc-trunk
2149 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
2150 345079 Fix build problems in VEX/useful/test_main.c
2151 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
2152 n-i-bz s390x: Fix memory corruption for multithreaded applications
2153 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
2154 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
2155 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
2156 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
2157 n-i-bz Add command-line option --default-suppressions
2158 n-i-bz Add support for BLKDISCARDZEROES ioctl
2159 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
2160 n-i-bz Add support for sys_pivot_root and sys_unshare
2162 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
2163 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
2164 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
2168 Release 3.9.0 (31 October 2013)
2169 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2170 3.9.0 is a feature release with many improvements and the usual
2171 collection of bug fixes.
2173 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2174 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
2175 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
2176 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
2178 * ================== PLATFORM CHANGES =================
2180 * Support for MIPS64 LE and BE running Linux. Valgrind has been
2181 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
2183 * Support for MIPS DSP ASE on MIPS32 platforms.
2185 * Support for s390x Decimal Floating Point instructions on hosts that
2186 have the DFP facility installed.
2188 * Support for POWER8 (Power ISA 2.07) instructions
2190 * Support for Intel AVX2 instructions. This is available only on 64
2193 * Initial support for Intel Transactional Synchronization Extensions,
2196 * Initial support for Hardware Transactional Memory on POWER.
2198 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
2199 run large GUI apps tolerably well.
2201 * ==================== TOOL CHANGES ====================
2205 - Improvements in handling of vectorised code, leading to
2206 significantly fewer false error reports. You need to use the flag
2207 --partial-loads-ok=yes to get the benefits of these changes.
2209 - Better control over the leak checker. It is now possible to
2210 specify which leak kinds (definite/indirect/possible/reachable)
2211 should be displayed, which should be regarded as errors, and which
2212 should be suppressed by a given leak suppression. This is done
2213 using the options --show-leak-kinds=kind1,kind2,..,
2214 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2215 "match-leak-kinds:" line in suppression entries, respectively.
2217 Note that generated leak suppressions contain this new line and
2218 are therefore more specific than in previous releases. To get the
2219 same behaviour as previous releases, remove the "match-leak-kinds:"
2220 line from generated suppressions before using them.
2222 - Reduced "possible leak" reports from the leak checker by the use
2223 of better heuristics. The available heuristics provide detection
2224 of valid interior pointers to std::stdstring, to new[] allocated
2225 arrays with elements having destructors and to interior pointers
2226 pointing to an inner part of a C++ object using multiple
2227 inheritance. They can be selected individually using the
2228 option --leak-check-heuristics=heur1,heur2,...
2230 - Better control of stacktrace acquisition for heap-allocated
2231 blocks. Using the --keep-stacktraces option, it is possible to
2232 control independently whether a stack trace is acquired for each
2233 allocation and deallocation. This can be used to create better
2234 "use after free" errors or to decrease Valgrind's resource
2235 consumption by recording less information.
2237 - Better reporting of leak suppression usage. The list of used
2238 suppressions (shown when the -v option is given) now shows, for
2239 each leak suppressions, how many blocks and bytes it suppressed
2240 during the last leak search.
2244 - False errors resulting from the use of statically initialised
2245 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2248 - False errors resulting from the use of pthread_cond_waits that
2249 timeout, have been removed.
2251 * ==================== OTHER CHANGES ====================
2253 * Some attempt to tune Valgrind's space requirements to the expected
2254 capabilities of the target:
2256 - The default size of the translation cache has been reduced from 8
2257 sectors to 6 on Android platforms, since each sector occupies
2258 about 40MB when using Memcheck.
2260 - The default size of the translation cache has been increased to 16
2261 sectors on all other platforms, reflecting the fact that large
2262 applications require instrumentation and storage of huge amounts
2263 of code. For similar reasons, the number of memory mapped
2264 segments that can be tracked has been increased by a factor of 6.
2266 - In all cases, the maximum number of sectors in the translation
2267 cache can be controlled by the new flag --num-transtab-sectors.
2269 * Changes in how debug info (line numbers, etc) is read:
2271 - Valgrind no longer temporarily mmaps the entire object to read
2272 from it. Instead, reading is done through a small fixed sized
2273 buffer. This avoids virtual memory usage spikes when Valgrind
2274 reads debuginfo from large shared objects.
2276 - A new experimental remote debug info server. Valgrind can read
2277 debug info from a different machine (typically, a build host)
2278 where debuginfo objects are stored. This can save a lot of time
2279 and hassle when running Valgrind on resource-constrained targets
2280 (phones, tablets) when the full debuginfo objects are stored
2281 somewhere else. This is enabled by the --debuginfo-server=
2284 - Consistency checking between main and debug objects can be
2285 disabled using the --allow-mismatched-debuginfo option.
2287 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2288 scanning can recover stack traces in some cases when the normal
2289 unwind mechanisms fail. Stack scanning is best described as "a
2290 nasty, dangerous and misleading hack" and so is disabled by default.
2291 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2294 * Detection and merging of recursive stack frame cycles. When your
2295 program has recursive algorithms, this limits the memory used by
2296 Valgrind for recorded stack traces and avoids recording
2297 uninteresting repeated calls. This is controlled by the command
2298 line option --merge-recursive-frame and by the monitor command
2299 "v.set merge-recursive-frames".
2301 * File name and line numbers for used suppressions. The list of used
2302 suppressions (shown when the -v option is given) now shows, for each
2303 used suppression, the file name and line number where the suppression
2306 * New and modified GDB server monitor features:
2308 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2309 that can be used to execute gdbserver monitor commands from the
2312 - A new monitor command, "v.info open_fds", that gives the list of
2313 open file descriptors and additional details.
2315 - An optional message in the "v.info n_errs_found" monitor command,
2316 for example "v.info n_errs_found test 1234 finished", allowing a
2317 comment string to be added to the process output, perhaps for the
2318 purpose of separating errors of different tests or test phases.
2320 - A new monitor command "v.info execontext" that shows information
2321 about the stack traces recorded by Valgrind.
2323 - A new monitor command "v.do expensive_sanity_check_general" to run
2324 some internal consistency checks.
2326 * New flag --sigill-diagnostics to control whether a diagnostic
2327 message is printed when the JIT encounters an instruction it can't
2328 translate. The actual behavior -- delivery of SIGILL to the
2329 application -- is unchanged.
2331 * The maximum amount of memory that Valgrind can use on 64 bit targets
2332 has been increased from 32GB to 64GB. This should make it possible
2333 to run applications on Memcheck that natively require up to about 35GB.
2335 * ==================== FIXED BUGS ====================
2337 The following bugs have been fixed or resolved. Note that "n-i-bz"
2338 stands for "not in bugzilla" -- that is, a bug that was reported to us
2339 but never got a bugzilla entry. We encourage you to file bugs in
2340 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2341 than mailing the developers (or mailing lists) directly -- bugs that
2342 are not entered into bugzilla tend to get forgotten about or ignored.
2344 To see details of a given bug, visit
2345 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2346 where XXXXXX is the bug number as listed below.
2348 123837 system call: 4th argument is optional, depending on cmd
2349 135425 memcheck should tell you where Freed blocks were Mallocd
2350 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2351 207815 Adds some of the drm ioctls to syswrap-linux.c
2352 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2353 252955 Impossible to compile with ccache
2354 253519 Memcheck reports auxv pointer accesses as invalid reads.
2355 263034 Crash when loading some PPC64 binaries
2356 269599 Increase deepest backtrace
2357 274695 s390x: Support "compare to/from logical" instructions (z196)
2358 275800 s390x: Autodetect cache info (part 2)
2359 280271 Valgrind reports possible memory leaks on still-reachable std::string
2360 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2361 289578 Backtraces with ARM unwind tables (stack scan flags)
2362 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2363 304832 ppc32: build failure
2364 305431 Use find_buildid shdr fallback for separate .debug files
2365 305728 Add support for AVX2 instructions
2366 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2367 306035 s390x: Fix IR generation for LAAG and friends
2368 306054 s390x: Condition code computation for convert-to-int/logical
2369 306098 s390x: alternate opcode form for convert to/from fixed
2370 306587 Fix cache line detection from auxiliary vector for PPC.
2371 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2372 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2373 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2374 307101 sys_capget second argument can be NULL
2375 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2376 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2377 307113 s390x: DFP support
2378 307141 valgrind does't work in mips-linux system
2379 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2380 307285 x86_amd64 feature test for avx in test suite is wrong
2381 307290 memcheck overlap testcase needs memcpy version filter
2382 307463 Please add "&limit=0" to the "all open bugs" link
2383 307465 --show-possibly-lost=no should reduce the error count / exit code
2384 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2385 307729 pkgconfig support broken valgrind.pc
2386 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2387 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2388 308089 Unhandled syscall on ppc64: prctl
2389 308135 PPC32 MPC8xx has 16 bytes cache size
2390 308321 testsuite memcheck filter interferes with gdb_filter
2392 308341 vgdb should report process exit (or fatal signal)
2393 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2394 308495 Remove build dependency on installed Xen headers
2395 308573 Internal error on 64-bit instruction executed in 32-bit mode
2397 308627 pmovmskb validity bit propagation is imprecise
2398 308644 vgdb command for having the info for the track-fds option
2399 308711 give more info about aspacemgr and arenas in out_of_memory
2400 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2401 308718 ARM implement SMLALBB family of instructions
2402 308886 Missing support for PTRACE_SET/GETREGSET
2403 308930 syscall name_to_handle_at (303 on amd64) not handled
2404 309229 V-bit tester does not report number of tests generated
2405 309323 print unrecognized instuction on MIPS
2406 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2407 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2408 309430 Self hosting ppc64 encounters a vassert error on operand type
2409 309600 valgrind is a bit confused about 0-sized sections
2410 309823 Generate errors for still reachable blocks
2411 309921 PCMPISTRI validity bit propagation is imprecise
2412 309922 none/tests/ppc64/test_dfp5 sometimes fails
2413 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2414 310424 --read-var-info does not properly describe static variables
2415 310792 search additional path for debug symbols
2416 310931 s390x: Message-security assist (MSA) instruction extension [..]
2417 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2418 311318 ARM: "128-bit constant is not implemented" error message
2419 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2420 311690 V crashes because it redirects branches inside of a redirected function
2421 311880 x86_64: make regtest hangs at shell_valid1
2422 311922 WARNING: unhandled syscall: 170
2424 312171 ppc: insn selection for DFP
2425 312571 Rounding mode call wrong for the DFP Iops [..]
2426 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2427 312913 Dangling pointers error should also report the alloc stack trace
2428 312980 Building on Mountain Lion generates some compiler warnings
2429 313267 Adding MIPS64/Linux port to Valgrind
2432 313811 Buffer overflow in assert_fail
2433 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2434 314269 ppc: dead code in insn selection
2435 314718 ARM: implement integer divide instruction (sdiv and udiv)
2436 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2437 315441 sendmsg syscall should ignore unset msghdr msg_flags
2438 315534 msgrcv inside a thread causes valgrind to hang (block)
2439 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2440 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2441 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2442 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2443 316144 valgrind.1 manpage contains unknown ??? strings [..]
2444 316145 callgrind command line options in manpage reference (unknown) [..]
2445 316145 callgrind command line options in manpage reference [..]
2446 316181 drd: Fixed a 4x slowdown for certain applications
2447 316503 Valgrind does not support SSE4 "movntdqa" instruction
2448 316535 Use of |signed int| instead of |size_t| in valgrind messages
2449 316696 fluidanimate program of parsec 2.1 stuck
2450 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2451 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2452 317186 "Impossible happens" when occurs VCVT instruction on ARM
2453 317318 Support for Threading Building Blocks "scalable_malloc"
2454 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2455 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2456 317463 bmi testcase IR SANITY CHECK FAILURE
2457 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2458 318050 libmpiwrap fails to compile with out-of-source build
2459 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2460 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2461 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2462 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2463 318932 Add missing PPC64 and PPC32 system call support
2464 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2465 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2466 319494 VEX Makefile-gcc standalone build update after r2702
2467 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2468 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2469 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2470 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2471 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2472 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2473 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2474 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2475 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2476 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2477 320895 add fanotify support (patch included)
2478 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2479 321065 Valgrind updates for Xen 4.3
2480 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2481 321363 Unhandled instruction: SSAX (ARM + Thumb)
2482 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2483 321466 Unhandled instruction: SHASX (ARM + Thumb)
2484 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2485 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2486 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2487 321620 Unhandled instruction: UASX (ARM + Thumb)
2488 321621 Unhandled instruction: USAX (ARM + Thumb)
2489 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2490 321693 Unhandled instruction: LDRSBT (Thumb)
2491 321694 Unhandled instruction: UQASX (ARM + Thumb)
2492 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2493 321697 Unhandled instruction: UHASX (ARM + Thumb)
2494 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2495 321704 Unhandled instruction: REVSH (ARM + Thumb)
2496 321730 Add cg_diff and cg_merge man pages
2497 321738 Add vgdb and valgrind-listener man pages
2499 321891 Unhandled instruction: LDRHT (Thumb)
2500 321960 pthread_create() then alloca() causing invalid stack write errors
2501 321969 ppc32 and ppc64 don't support [lf]setxattr
2502 322254 Show threadname together with tid if set by application
2503 322294 Add initial support for IBM Power ISA 2.07
2504 322368 Assertion failure in wqthread_hijack under OS X 10.8
2505 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2506 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2507 322851 0bXXX binary literal syntax is not standard
2508 323035 Unhandled instruction: LDRSHT(Thumb)
2509 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2510 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2511 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2512 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2513 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2514 323437 Phase 2 support for IBM Power ISA 2.07
2515 323713 Support mmxext (integer sse) subset on i386 (athlon)
2516 323803 Transactional memory instructions are not supported for Power
2517 323893 SSE3 not available on amd cpus in valgrind
2518 323905 Probable false positive from Valgrind/drd on close()
2519 323912 valgrind.h header isn't compatible for mingw64
2520 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
2521 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
2522 324181 mmap does not handle MAP_32BIT
2523 324227 memcheck false positive leak when a thread calls exit+block [..]
2524 324421 Support for fanotify API on ARM architecture
2525 324514 gdbserver monitor cmd output behaviour consistency [..]
2526 324518 ppc64: Emulation of dcbt instructions does not handle [..]
2527 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
2528 324582 When access is made to freed memory, report both allocation [..]
2529 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
2530 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
2531 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
2532 324834 Unhandled instructions in Microsoft C run-time for x86_64
2533 324894 Phase 3 support for IBM Power ISA 2.07
2534 326091 drd: Avoid false race reports from optimized strlen() impls
2535 326113 valgrind libvex hwcaps error on AMD64
2536 n-i-bz Some wrong command line options could be ignored
2537 n-i-bz patch to allow fair-sched on android
2538 n-i-bz report error for vgdb snapshot requested before execution
2539 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
2541 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
2545 Release 3.8.1 (19 September 2012)
2546 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2547 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
2548 that occur moderately frequently in real use cases, adds support for
2549 some missing instructions on ARM, and fixes a deadlock condition on
2550 MacOSX. If you package or deliver 3.8.0 for others to use, you might
2551 want to consider upgrading to 3.8.1 instead.
2553 The following bugs have been fixed or resolved. Note that "n-i-bz"
2554 stands for "not in bugzilla" -- that is, a bug that was reported to us
2555 but never got a bugzilla entry. We encourage you to file bugs in
2556 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2557 than mailing the developers (or mailing lists) directly -- bugs that
2558 are not entered into bugzilla tend to get forgotten about or ignored.
2560 To see details of a given bug, visit
2561 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2562 where XXXXXX is the bug number as listed below.
2565 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
2566 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
2567 298281 wcslen causes false(?) uninitialised value warnings
2568 301281 valgrind hangs on OS X when the process calls system()
2569 304035 disInstr(arm): unhandled instruction 0xE1023053
2570 304867 implement MOVBE instruction in x86 mode
2571 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
2572 305042 amd64: implement 0F 7F encoding of movq between two registers
2573 305199 ARM: implement QDADD and QDSUB
2574 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
2575 305513 killed by fatal signal: SIGSEGV
2576 305690 DRD reporting invalid semaphore when sem_trywait fails
2577 305926 Invalid alignment checks for some AVX instructions
2578 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
2579 306310 3.8.0 release tarball missing some files
2580 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
2581 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
2582 n-i-bz shmat of a segment > 4Gb does not work
2583 n-i-bz simulate_control_c script wrong USR1 signal number on mips
2584 n-i-bz vgdb ptrace calls wrong on mips [...]
2585 n-i-bz Fixes for more MPI false positives
2586 n-i-bz exp-sgcheck's memcpy causes programs to segfault
2587 n-i-bz OSX build w/ clang: asserts at startup
2588 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
2589 n-i-bz fix a couple of union tag-vs-field mixups
2590 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
2592 The following bugs were fixed in 3.8.0 but not listed in this NEWS
2595 254088 Valgrind should know about UD2 instruction
2598 304754 NEWS blows TeX's little mind
2600 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
2604 Release 3.8.0 (10 August 2012)
2605 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2606 3.8.0 is a feature release with many improvements and the usual
2607 collection of bug fixes.
2609 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2610 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
2611 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
2612 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
2613 There is initial support for MacOSX 10.8, but it is not usable for
2614 serious work at present.
2616 * ================== PLATFORM CHANGES =================
2618 * Support for MIPS32 platforms running Linux. Valgrind has been
2619 tested on MIPS32 and MIPS32r2 platforms running different Debian
2620 Squeeze and MeeGo distributions. Both little-endian and big-endian
2621 cores are supported. The tools Memcheck, Massif and Lackey have
2622 been tested and are known to work. See README.mips for more details.
2624 * Preliminary support for Android running on x86.
2626 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
2628 * Support for Intel AVX instructions and for AES instructions. This
2629 support is available only for 64 bit code.
2631 * Support for POWER Decimal Floating Point instructions.
2633 * ==================== TOOL CHANGES ====================
2635 * Non-libc malloc implementations are now supported. This is useful
2636 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
2637 Using the new option --soname-synonyms, such tools can be informed
2638 that the malloc implementation is either linked statically into the
2639 executable, or is present in some other shared library different
2640 from libc.so. This makes it possible to process statically linked
2641 programs, and programs using other malloc libraries, for example
2642 TCMalloc or JEMalloc.
2644 * For tools that provide their own replacement for malloc et al, the
2645 option --redzone-size=<number> allows users to specify the size of
2646 the padding blocks (redzones) added before and after each client
2647 allocated block. Smaller redzones decrease the memory needed by
2648 Valgrind. Bigger redzones increase the chance to detect blocks
2649 overrun or underrun. Prior to this change, the redzone size was
2650 hardwired to 16 bytes in Memcheck.
2654 - The leak_check GDB server monitor command now can
2655 control the maximum nr of loss records to output.
2657 - Reduction of memory use for applications allocating
2658 many blocks and/or having many partially defined bytes.
2660 - Addition of GDB server monitor command 'block_list' that lists
2661 the addresses/sizes of the blocks of a leak search loss record.
2663 - Addition of GDB server monitor command 'who_points_at' that lists
2664 the locations pointing at a block.
2666 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2667 detect an invalid access of these redzones, by marking them
2668 noaccess. Similarly, if a redzone size is given for a memory
2669 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2670 This still allows to find some bugs if the user has forgotten to
2671 mark the pool superblock noaccess.
2673 - Performance of memory leak check has been improved, especially in
2674 cases where there are many leaked blocks and/or many suppression
2675 rules used to suppress leak reports.
2677 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2678 more precise analysis, which is important for LLVM/Clang
2679 generated code. This is at the cost of somewhat reduced
2680 performance. Note there is no change to analysis precision or
2681 costs on Linux targets.
2685 - Added even more facilities that can help finding the cause of a data
2686 race, namely the command-line option --ptrace-addr and the macro
2687 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2689 - Fixed a subtle bug that could cause false positive data race reports.
2691 * ==================== OTHER CHANGES ====================
2693 * The C++ demangler has been updated so as to work well with C++
2694 compiled by up to at least g++ 4.6.
2696 * Tool developers can make replacement/wrapping more flexible thanks
2697 to the new option --soname-synonyms. This was reported above, but
2698 in fact is very general and applies to all function
2699 replacement/wrapping, not just to malloc-family functions.
2701 * Round-robin scheduling of threads can be selected, using the new
2702 option --fair-sched= yes. Prior to this change, the pipe-based
2703 thread serialisation mechanism (which is still the default) could
2704 give very unfair scheduling. --fair-sched=yes improves
2705 responsiveness of interactive multithreaded applications, and
2706 improves repeatability of results from the thread checkers Helgrind
2709 * For tool developers: support to run Valgrind on Valgrind has been
2710 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2712 * gdbserver now shows the float shadow registers as integer
2713 rather than float values, as the shadow values are mostly
2714 used as bit patterns.
2716 * Increased limit for the --num-callers command line flag to 500.
2718 * Performance improvements for error matching when there are many
2719 suppression records in use.
2721 * Improved support for DWARF4 debugging information (bug 284184).
2723 * Initial support for DWZ compressed Dwarf debug info.
2725 * Improved control over the IR optimiser's handling of the tradeoff
2726 between performance and precision of exceptions. Specifically,
2727 --vex-iropt-precise-memory-exns has been removed and replaced by
2728 --vex-iropt-register-updates, with extended functionality. This
2729 allows the Valgrind gdbserver to always show up to date register
2732 * Modest performance gains through the use of translation chaining for
2735 * ==================== FIXED BUGS ====================
2737 The following bugs have been fixed or resolved. Note that "n-i-bz"
2738 stands for "not in bugzilla" -- that is, a bug that was reported to us
2739 but never got a bugzilla entry. We encourage you to file bugs in
2740 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2741 than mailing the developers (or mailing lists) directly -- bugs that
2742 are not entered into bugzilla tend to get forgotten about or ignored.
2744 To see details of a given bug, visit
2745 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2746 where XXXXXX is the bug number as listed below.
2748 197914 Building valgrind from svn now requires automake-1.10
2749 203877 increase to 16Mb maximum allowed alignment for memalign et al
2750 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
2751 247386 make perf does not run all performance tests
2752 270006 Valgrind scheduler unfair
2753 270777 Adding MIPS/Linux port to Valgrind
2754 270796 s390x: Removed broken support for the TS insn
2755 271438 Fix configure for proper SSE4.2 detection
2756 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
2757 273475 Add support for AVX instructions
2758 274078 improved configure logic for mpicc
2759 276993 fix mremap 'no thrash checks'
2760 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
2761 281482 memcheck incorrect byte allocation count in realloc() for silly argument
2762 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
2763 283413 Fix wrong sanity check
2764 283671 Robustize alignment computation in LibVEX_Alloc
2765 283961 Adding support for some HCI IOCTLs
2766 284124 parse_type_DIE: confused by: DWARF 4
2767 284864 == 273475 (Add support for AVX instructions)
2768 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
2769 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
2770 285725 == 273475 (Add support for AVX instructions)
2771 286261 add wrapper for linux I2C_RDWR ioctl
2772 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
2773 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
2774 286384 configure fails "checking for a supported version of gcc"
2775 286497 == 273475 (Add support for AVX instructions)
2776 286596 == 273475 (Add support for AVX instructions)
2777 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
2778 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
2779 287260 Incorrect conditional jump or move depends on uninitialised value(s)
2780 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
2781 287307 == 273475 (Add support for AVX instructions)
2782 287858 VG_(strerror): unknown error
2783 288298 (MacOSX) unhandled syscall shm_unlink
2784 288995 == 273475 (Add support for AVX instructions)
2785 289470 Loading of large Mach-O thin binaries fails.
2786 289656 == 273475 (Add support for AVX instructions)
2787 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
2788 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
2789 289839 s390x: Provide support for unicode conversion instructions
2790 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
2791 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
2792 290655 Add support for AESKEYGENASSIST instruction
2793 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
2794 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
2795 291253 ES register not initialised in valgrind simulation
2796 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
2797 291865 s390x: Support the "Compare Double and Swap" family of instructions
2798 292300 == 273475 (Add support for AVX instructions)
2799 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
2800 292493 == 273475 (Add support for AVX instructions)
2801 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
2802 292627 Missing support for some SCSI ioctls
2803 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
2804 292841 == 273475 (Add support for AVX instructions)
2805 292993 implement the getcpu syscall on amd64-linux
2806 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
2807 293088 Add some VEX sanity checks for ppc64 unhandled instructions
2808 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
2809 293754 PCMPxSTRx not implemented for 16-bit characters
2810 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
2811 293808 CLFLUSH not supported by latest VEX for amd64
2812 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
2813 294048 MPSADBW instruction not implemented
2814 294055 regtest none/tests/shell fails when locale is not set to C
2815 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
2816 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
2817 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
2818 294260 disInstr_AMD64: disInstr miscalculated next %rip
2819 294523 --partial-loads-ok=yes causes false negatives
2820 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
2821 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
2822 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
2823 295089 can not annotate source for both helgrind and drd
2824 295221 POWER Processor decimal floating point instruction support missing
2825 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
2826 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
2827 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
2828 295617 ARM - Add some missing syscalls
2829 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
2830 296229 Linux user input device ioctls missing wrappers
2831 296318 ELF Debug info improvements (more than one rx/rw mapping)
2832 296422 Add translation chaining support
2833 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
2834 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
2835 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
2836 297078 gdbserver signal handling problems [..]
2837 297147 drd false positives on newly allocated memory
2838 297329 disallow decoding of IBM Power DFP insns on some machines
2839 297497 POWER Processor decimal floating point instruction support missing
2840 297701 Another alias for strncasecmp_l in libc-2.13.so
2841 297911 'invalid write' not reported when using APIs for custom mem allocators.
2842 297976 s390x: revisit EX implementation
2843 297991 Valgrind interferes with mmap()+ftell()
2844 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
2845 297993 Fix compilation of valgrind with gcc -g3.
2846 298080 POWER Processor DFP support missing, part 3
2847 298227 == 273475 (Add support for AVX instructions)
2848 298335 == 273475 (Add support for AVX instructions)
2849 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
2850 298394 s390x: Don't bail out on an unknown machine model. [..]
2851 298421 accept4() syscall (366) support is missing for ARM
2852 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
2853 298732 valgrind installation problem in ubuntu with kernel version 3.x
2854 298862 POWER Processor DFP instruction support missing, part 4
2855 298864 DWARF reader mis-parses DW_FORM_ref_addr
2856 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
2857 299053 Support DWARF4 DW_AT_high_pc constant form
2858 299104 == 273475 (Add support for AVX instructions)
2859 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
2860 299629 dup3() syscall (358) support is missing for ARM
2861 299694 POWER Processor DFP instruction support missing, part 5
2862 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
2863 299803 == 273475 (Add support for AVX instructions)
2864 299804 == 273475 (Add support for AVX instructions)
2865 299805 == 273475 (Add support for AVX instructions)
2866 300140 ARM - Missing (T1) SMMUL
2867 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
2868 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
2869 300414 FCOM and FCOMP unimplemented for amd64 guest
2870 301204 infinite loop in canonicaliseSymtab with ifunc symbol
2871 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
2872 301265 add x86 support to Android build
2873 301984 configure script doesn't detect certain versions of clang
2874 302205 Fix compiler warnings for POWER VEX code and POWER test cases
2875 302287 Unhandled movbe instruction on Atom processors
2876 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
2877 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
2878 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
2879 302656 == 273475 (Add support for AVX instructions)
2880 302709 valgrind for ARM needs extra tls support for android emulator [..]
2881 302827 add wrapper for CDROM_GET_CAPABILITY
2882 302901 Valgrind crashes with dwz optimized debuginfo
2883 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
2884 303116 Add support for the POWER instruction popcntb
2885 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
2886 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
2887 303466 == 273475 (Add support for AVX instructions)
2888 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
2889 303963 strstr() function produces wrong results under valgrind callgrind
2890 304054 CALL_FN_xx macros need to enforce stack alignment
2891 304561 tee system call not supported
2892 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
2893 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
2894 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
2895 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
2896 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
2897 n-i-bz s390x: Shadow registers can now be examined using vgdb
2899 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
2900 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
2904 Release 3.7.0 (5 November 2011)
2905 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2906 3.7.0 is a feature release with many significant improvements and the
2907 usual collection of bug fixes.
2909 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2910 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
2911 Support for recent distros and toolchain components (glibc 2.14, gcc
2912 4.6, MacOSX 10.7) has been added.
2914 * ================== PLATFORM CHANGES =================
2916 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
2917 analyse 64-bit programs running on z/Architecture. Most user space
2918 instructions up to and including z10 are supported. Valgrind has
2919 been tested extensively on z9, z10, and z196 machines running SLES
2920 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
2921 known to work well. Callgrind, Helgrind, and DRD work reasonably
2922 well on z9 and later models. See README.s390 for more details.
2924 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
2925 64-bit processes are supported. Some complex threaded applications
2926 (Firefox) are observed to hang when run as 32 bit applications,
2927 whereas 64-bit versions run OK. The cause is unknown. Memcheck
2928 will likely report some false errors. In general, expect some rough
2929 spots. This release also supports MacOSX 10.6, but drops support
2932 * Preliminary support for Android (on ARM). Valgrind can now run
2933 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
2934 README.android for more details, plus instructions on how to get
2937 * Support for the IBM Power ISA 2.06 (Power7 instructions)
2939 * General correctness and performance improvements for ARM/Linux, and,
2940 by extension, ARM/Android.
2942 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
2943 instruction set support is under development but is not available in
2946 * Support for AIX5 has been removed.
2948 * ==================== TOOL CHANGES ====================
2950 * Memcheck: some incremental changes:
2952 - reduction of memory use in some circumstances
2954 - improved handling of freed memory, which in some circumstances
2955 can cause detection of use-after-free that would previously have
2958 - fix of a longstanding bug that could cause false negatives (missed
2959 errors) in programs doing vector saturated narrowing instructions.
2961 * Helgrind: performance improvements and major memory use reductions,
2962 particularly for large, long running applications which perform many
2963 synchronisation (lock, unlock, etc) events. Plus many smaller
2966 - display of locksets for both threads involved in a race
2968 - general improvements in formatting/clarity of error messages
2970 - addition of facilities and documentation regarding annotation
2971 of thread safe reference counted C++ classes
2973 - new flag --check-stack-refs=no|yes [yes], to disable race checking
2974 on thread stacks (a performance hack)
2976 - new flag --free-is-write=no|yes [no], to enable detection of races
2977 where one thread accesses heap memory but another one frees it,
2978 without any coordinating synchronisation event
2980 * DRD: enabled XML output; added support for delayed thread deletion
2981 in order to detect races that occur close to the end of a thread
2982 (--join-list-vol); fixed a memory leak triggered by repeated client
2983 memory allocatation and deallocation; improved Darwin support.
2985 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
2987 * exp-sgcheck: this tool has been reduced in scope so as to improve
2988 performance and remove checking that Memcheck does better.
2989 Specifically, the ability to check for overruns for stack and global
2990 arrays is unchanged, but the ability to check for overruns of heap
2991 blocks has been removed. The tool has accordingly been renamed to
2992 exp-sgcheck ("Stack and Global Array Checking").
2994 * ==================== OTHER CHANGES ====================
2996 * GDB server: Valgrind now has an embedded GDB server. That means it
2997 is possible to control a Valgrind run from GDB, doing all the usual
2998 things that GDB can do (single stepping, breakpoints, examining
2999 data, etc). Tool-specific functionality is also available. For
3000 example, it is possible to query the definedness state of variables
3001 or memory from within GDB when running Memcheck; arbitrarily large
3002 memory watchpoints are supported, etc. To use the GDB server, start
3003 Valgrind with the flag --vgdb-error=0 and follow the on-screen
3006 * Improved support for unfriendly self-modifying code: a new option
3007 --smc-check=all-non-file is available. This adds the relevant
3008 consistency checks only to code that originates in non-file-backed
3009 mappings. In effect this confines the consistency checking only to
3010 code that is or might be JIT generated, and avoids checks on code
3011 that must have been compiled ahead of time. This significantly
3012 improves performance on applications that generate code at run time.
3014 * It is now possible to build a working Valgrind using Clang-2.9 on
3017 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
3018 These enable and disable error reporting on a per-thread, and
3019 nestable, basis. This is useful for hiding errors in particularly
3020 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
3021 now uses this facility.
3023 * Added the --mod-funcname option to cg_diff.
3025 * ==================== FIXED BUGS ====================
3027 The following bugs have been fixed or resolved. Note that "n-i-bz"
3028 stands for "not in bugzilla" -- that is, a bug that was reported to us
3029 but never got a bugzilla entry. We encourage you to file bugs in
3030 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3031 mailing the developers (or mailing lists) directly -- bugs that are
3032 not entered into bugzilla tend to get forgotten about or ignored.
3034 To see details of a given bug, visit
3035 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3036 where XXXXXX is the bug number as listed below.
3038 79311 malloc silly arg warning does not give stack trace
3039 210935 port valgrind.h (not valgrind) to win32 to support client requests
3040 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
3041 243404 Port to zSeries
3042 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
3043 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
3044 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
3045 253206 Some fixes for the faultstatus testcase
3046 255223 capget testcase fails when running as root
3047 256703 xlc_dbl_u32.c testcase broken
3048 256726 Helgrind tests have broken inline asm
3049 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
3050 264800 testcase compile failure on zseries
3051 265762 make public VEX headers compilable by G++ 3.x
3052 265771 assertion in jumps.c (r11523) fails with glibc-2.3
3053 266753 configure script does not give the user the option to not use QtCore
3054 266931 gen_insn_test.pl is broken
3055 266961 ld-linux.so.2 i?86-linux strlen issues
3056 266990 setns instruction causes false positive
3057 267020 Make directory for temporary files configurable at run-time.
3058 267342 == 267997 (segmentation fault on Mac OS 10.6)
3059 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
3060 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
3061 267488 regtest: darwin support for 64-bit build
3062 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
3063 267630 Add support for IBM Power ISA 2.06 -- stage 1
3064 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
3065 267819 Add client request for informing the core about reallocation
3066 267925 laog data structure quadratic for a single sequence of lock
3067 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
3068 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
3069 268513 missed optimizations in fold_Expr
3070 268619 s390x: fpr - gpr transfer facility
3071 268620 s390x: reconsider "long displacement" requirement
3072 268621 s390x: improve IR generation for XC
3073 268715 s390x: FLOGR is not universally available
3074 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
3075 268930 s390x: MHY is not universally available
3076 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
3077 269079 Support ptrace system call on ARM
3078 269144 missing "Bad option" error message
3079 269209 conditional load and store facility (z196)
3080 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
3081 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
3082 269736 s390x: minor code generation tweaks
3083 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
3084 269863 s390x: remove unused function parameters
3085 269864 s390x: tweak s390_emit_load_cc
3086 269884 == 250101 (overhead for huge blocks exhausts space too soon)
3087 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
3088 270115 s390x: rewrite some testcases
3089 270309 == 267997 (valgrind crash on startup)
3090 270320 add support for Linux FIOQSIZE ioctl() call
3091 270326 segfault while trying to sanitize the environment passed to execle
3092 270794 IBM POWER7 support patch causes regression in none/tests
3093 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
3094 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
3095 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
3096 270959 s390x: invalid use of R0 as base register
3097 271042 VSX configure check fails when it should not
3098 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
3099 271259 s390x: fix code confusion
3100 271337 == 267997 (Valgrind segfaults on MacOS X)
3101 271385 s390x: Implement Ist_MBE
3102 271501 s390x: misc cleanups
3103 271504 s390x: promote likely and unlikely
3104 271579 ppc: using wrong enum type
3105 271615 unhandled instruction "popcnt" (arch=amd10h)
3106 271730 Fix bug when checking ioctls: duplicate check
3107 271776 s390x: provide STFLE instruction support
3108 271779 s390x: provide clock instructions like STCK
3109 271799 Darwin: ioctls without an arg report a memory error
3110 271820 arm: fix type confusion
3111 271917 pthread_cond_timedwait failure leads to not-locked false positive
3112 272067 s390x: fix DISP20 macro
3113 272615 A typo in debug output in mc_leakcheck.c
3114 272661 callgrind_annotate chokes when run from paths containing regex chars
3115 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
3116 272955 Unhandled syscall error for pwrite64 on ppc64 arch
3117 272967 make documentation build-system more robust
3118 272986 Fix gcc-4.6 warnings with valgrind.h
3119 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
3120 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
3121 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
3122 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
3123 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
3124 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
3125 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
3126 273778 exp-ptrcheck: unhandled sysno == 259
3127 274089 exp-ptrcheck: unhandled sysno == 208
3128 274378 s390x: Various dispatcher tweaks
3129 274447 WARNING: unhandled syscall: 340
3130 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
3131 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
3132 274926 valgrind does not build against linux-3
3133 275148 configure FAIL with glibc-2.14
3134 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
3135 275168 Make Valgrind work for MacOSX 10.7 Lion
3136 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
3137 275278 valgrind does not build on Linux kernel 3.0.* due to silly
3138 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
3139 275308 Fix implementation for ppc64 fres instruc
3140 275339 s390x: fix testcase compile warnings
3141 275517 s390x: Provide support for CKSM instruction
3142 275710 s390x: get rid of redundant address mode calculation
3143 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
3144 275852 == 250101 (valgrind uses all swap space and is killed)
3145 276784 Add support for IBM Power ISA 2.06 -- stage 3
3146 276987 gdbsrv: fix tests following recent commits
3147 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
3148 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
3149 277471 Unhandled syscall: 340
3150 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
3151 277653 ARM: support Thumb2 PLD instruction
3152 277663 ARM: NEON float VMUL by scalar incorrect
3153 277689 ARM: tests for VSTn with register post-index are broken
3154 277694 ARM: BLX LR instruction broken in ARM mode
3155 277780 ARM: VMOV.F32 (immediate) instruction is broken
3156 278057 fuse filesystem syscall deadlocks
3157 278078 Unimplemented syscall 280 on ppc32
3158 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
3159 278454 VALGRIND_STACK_DEREGISTER has wrong output type
3160 278502 == 275284 (Valgrind confuses memcpy() and memmove())
3161 278892 gdbsrv: factorize gdb version handling, fix doc and typos
3162 279027 Support for MVCL and CLCL instruction
3163 279027 s390x: Provide support for CLCL and MVCL instructions
3164 279062 Remove a redundant check in the insn selector for ppc.
3165 279071 JDK creates PTEST with redundant REX.W prefix
3166 279212 gdbsrv: add monitor cmd v.info scheduler.
3167 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
3168 279698 memcheck discards valid-bits for packuswb
3169 279795 memcheck reports uninitialised values for mincore on amd64
3170 279994 Add support for IBM Power ISA 2.06 -- stage 3
3171 280083 mempolicy syscall check errors
3172 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
3173 280710 s390x: config files for nightly builds
3174 280757 /tmp dir still used by valgrind even if TMPDIR is specified
3175 280965 Valgrind breaks fcntl locks when program does mmap
3176 281138 WARNING: unhandled syscall: 340
3177 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
3178 281304 == 275168 (Darwin: dyld "cannot load inserted library")
3179 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
3180 281468 s390x: handle do_clone and gcc clones in call traces
3181 281488 ARM: VFP register corruption
3182 281828 == 275284 (false memmove warning: "Source and destination overlap")
3183 281883 s390x: Fix system call wrapper for "clone".
3184 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
3185 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
3186 282238 SLES10: make check fails
3187 282979 strcasestr needs replacement with recent(>=2.12) glibc
3188 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
3189 283243 Regression in ppc64 memcheck tests
3190 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
3191 283427 re-connect epoll_pwait syscall on ARM linux
3192 283600 gdbsrv: android: port vgdb.c
3193 283709 none/tests/faultstatus needs to account for page size
3194 284305 filter_gdb needs enhancement to work on ppc64
3195 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
3196 284472 Thumb2 ROR.W encoding T2 not implemented
3197 284621 XML-escape process command line in XML output
3198 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
3199 that have non-power-of-2 sizes (also AMDs)
3200 n-i-bz don't be spooked by libraries mashed by elfhack
3201 n-i-bz don't be spooked by libxul.so linked with gold
3202 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
3204 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
3205 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
3206 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
3210 Release 3.6.1 (16 February 2011)
3211 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3212 3.6.1 is a bug fix release. It adds support for some SSE4
3213 instructions that were omitted in 3.6.0 due to lack of time. Initial
3214 support for glibc-2.13 has been added. A number of bugs causing
3215 crashing or assertion failures have been fixed.
3217 The following bugs have been fixed or resolved. Note that "n-i-bz"
3218 stands for "not in bugzilla" -- that is, a bug that was reported to us
3219 but never got a bugzilla entry. We encourage you to file bugs in
3220 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3221 mailing the developers (or mailing lists) directly -- bugs that are
3222 not entered into bugzilla tend to get forgotten about or ignored.
3224 To see details of a given bug, visit
3225 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3226 where XXXXXX is the bug number as listed below.
3228 188572 Valgrind on Mac should suppress setenv() mem leak
3229 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3230 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3231 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3232 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3233 254420 memory pool tracking broken
3234 254957 Test code failing to compile due to changes in memcheck.h
3235 255009 helgrind/drd: crash on chmod with invalid parameter
3236 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3237 255355 helgrind/drd: crash on threaded programs doing fork
3239 255418 (SSE4.x) rint call compiled with ICC
3240 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3241 255888 closing valgrindoutput tag outputted to log-stream on error
3242 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3243 255966 Slowness when using mempool annotations
3244 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3245 256600 super-optimized strcasecmp() false positive
3246 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3247 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3248 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3249 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3250 257276 Missing case in memcheck --track-origins=yes
3251 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3252 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3253 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3254 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3255 263099 callgrind_annotate counts Ir improperly [...]
3256 263877 undefined coprocessor instruction on ARMv7
3257 265964 configure FAIL with glibc-2.13
3258 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3259 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3260 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3261 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3262 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3264 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3268 Release 3.6.0 (21 October 2010)
3269 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3270 3.6.0 is a feature release with many significant improvements and the
3271 usual collection of bug fixes.
3273 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3274 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3275 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3277 -------------------------
3279 Here are some highlights. Details are shown further down:
3281 * Support for ARM/Linux.
3283 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3285 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3287 * Support for the SSE4.2 instruction set.
3289 * Enhancements to the Callgrind profiler, including the ability to
3290 handle CPUs with three levels of cache.
3292 * A new experimental heap profiler, DHAT.
3294 * A huge number of bug fixes and small enhancements.
3296 -------------------------
3298 Here are details of the above changes, together with descriptions of
3299 many other changes, and a list of fixed bugs.
3301 * ================== PLATFORM CHANGES =================
3303 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3304 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3305 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3307 This requires a CPU capable of running the ARMv7-A instruction set
3308 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3309 of the user space instruction set, including ARM and Thumb integer
3310 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3311 Cachegrind and Massif tools work properly; other tools work to
3314 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3315 with support for recent releases of the underlying toolchain
3316 components, notably gcc-4.5 and glibc-2.12.
3318 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3319 support also works much better on OS X 10.5, and is as solid as
3322 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3323 64-bit mode. In 32-bit mode, support is only available up to and
3324 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3325 supported in 64-bit mode, and 32-bit mode does in fact support the
3326 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3327 10.6 on 32-bit targets.
3329 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3330 and including version 2.05 is supported.
3332 * ==================== TOOL CHANGES ====================
3334 * Cachegrind has a new processing script, cg_diff, which finds the
3335 difference between two profiles. It's very useful for evaluating
3336 the performance effects of a change in a program.
3338 Related to this change, the meaning of cg_annotate's (rarely-used)
3339 --threshold option has changed; this is unlikely to affect many
3340 people, if you do use it please see the user manual for details.
3342 * Callgrind now can do branch prediction simulation, similar to
3343 Cachegrind. In addition, it optionally can count the number of
3344 executed global bus events. Both can be used for a better
3345 approximation of a "Cycle Estimation" as derived event (you need to
3346 update the event formula in KCachegrind yourself).
3348 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3349 rather than the L2 cache. This is to accommodate machines with
3350 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3351 cache configuration of such a machine it will run the simulation as
3352 if the L2 cache isn't present. This means the results are less
3353 likely to match the true result for the machine, but
3354 Cachegrind/Callgrind's results are already only approximate, and
3355 should not be considered authoritative. The results are still
3356 useful for giving a general idea about a program's locality.
3358 * Massif has a new option, --pages-as-heap, which is disabled by
3359 default. When enabled, instead of tracking allocations at the level
3360 of heap blocks (as allocated with malloc/new/new[]), it instead
3361 tracks memory allocations at the level of memory pages (as mapped by
3362 mmap, brk, etc). Each mapped page is treated as its own block.
3363 Interpreting the page-level output is harder than the heap-level
3364 output, but this option is useful if you want to account for every
3365 byte of memory used by a program.
3367 * DRD has two new command-line options: --free-is-write and
3368 --trace-alloc. The former allows to detect reading from already freed
3369 memory, and the latter allows tracing of all memory allocations and
3372 * DRD has several new annotations. Custom barrier implementations can
3373 now be annotated, as well as benign races on static variables.
3375 * DRD's happens before / happens after annotations have been made more
3376 powerful, so that they can now also be used to annotate e.g. a smart
3377 pointer implementation.
3379 * Helgrind's annotation set has also been drastically improved, so as
3380 to provide to users a general set of annotations to describe locks,
3381 semaphores, barriers and condition variables. Annotations to
3382 describe thread-safe reference counted heap objects have also been
3385 * Memcheck has a new command-line option, --show-possibly-lost, which
3386 is enabled by default. When disabled, the leak detector will not
3387 show possibly-lost blocks.
3389 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3390 has been added. DHAT keeps track of allocated heap blocks, and also
3391 inspects every memory reference to see which block (if any) is being
3392 accessed. This gives a lot of insight into block lifetimes,
3393 utilisation, turnover, liveness, and the location of hot and cold
3394 fields. You can use DHAT to do hot-field profiling.
3396 * ==================== OTHER CHANGES ====================
3398 * Improved support for unfriendly self-modifying code: the extra
3399 overhead incurred by --smc-check=all has been reduced by
3400 approximately a factor of 5 as compared with 3.5.0.
3402 * Ability to show directory names for source files in error messages.
3403 This is combined with a flexible mechanism for specifying which
3404 parts of the paths should be shown. This is enabled by the new flag
3407 * A new flag, --require-text-symbol, which will stop the run if a
3408 specified symbol is not found it a given shared object when it is
3409 loaded into the process. This makes advanced working with function
3410 intercepting and wrapping safer and more reliable.
3412 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3413 and control of Valgrind is now available for the tools Memcheck and
3414 Helgrind. XML output from Valgrind is available for Memcheck,
3415 Helgrind and exp-Ptrcheck.
3417 * More reliable stack unwinding on amd64-linux, particularly in the
3418 presence of function wrappers, and with gcc-4.5 compiled code.
3420 * Modest scalability (performance improvements) for massive
3421 long-running applications, particularly for those with huge amounts
3424 * Support for analyzing programs running under Wine with has been
3425 improved. The header files <valgrind/valgrind.h>,
3426 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3427 Windows-programs compiled with MinGW or one of the Microsoft Visual
3430 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3431 The 32-bit simulator was not affected. This did not occur often,
3432 but when it did would usually crash the program under test.
3435 * A large number of bugs were fixed. These are shown below.
3437 * A number of bugs were investigated, and were candidates for fixing,
3438 but are not fixed in 3.6.0, due to lack of developer time. They may
3439 get fixed in later releases. They are:
3441 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3442 212419 false positive "lock order violated" (A+B vs A)
3443 213685 Undefined value propagates past dependency breaking instruction
3444 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3445 237920 valgrind segfault on fork failure
3446 242137 support for code compiled by LLVM-2.8
3447 242423 Another unknown Intel cache config value
3448 243232 Inconsistent Lock Orderings report with trylock
3449 243483 ppc: callgrind triggers VEX assertion failure
3450 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3451 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3453 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3454 249435 Analyzing wine programs with callgrind triggers a crash
3455 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3456 250065 Handling large allocations
3457 250101 huge "free" memory usage due to m_mallocfree.c
3458 "superblocks fragmentation"
3459 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3460 252091 Callgrind on ARM does not detect function returns correctly
3461 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3462 254420 memory pool tracking broken
3463 n-i-bz support for adding symbols for JIT generated code
3466 The following bugs have been fixed or resolved. Note that "n-i-bz"
3467 stands for "not in bugzilla" -- that is, a bug that was reported to us
3468 but never got a bugzilla entry. We encourage you to file bugs in
3469 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3470 mailing the developers (or mailing lists) directly -- bugs that are
3471 not entered into bugzilla tend to get forgotten about or ignored.
3473 To see details of a given bug, visit
3474 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3475 where XXXXXX is the bug number as listed below.
3477 135264 dcbzl instruction missing
3479 153699 Valgrind should report unaligned reads with movdqa
3481 190429 Valgrind reports lost of errors in ld.so
3482 with x86_64 2.9.90 glibc
3483 197266 valgrind appears to choke on the xmms instruction
3485 197988 Crash when demangling very large symbol names
3486 202315 unhandled syscall: 332 (inotify_init1)
3487 203256 Add page-level profiling to Massif
3488 205093 dsymutil=yes needs quotes, locking (partial fix)
3489 205241 Snow Leopard 10.6 support (partial fix)
3490 206600 Leak checker fails to upgrade indirect blocks when their
3491 parent becomes reachable
3492 210935 port valgrind.h (not valgrind) to win32 so apps run under
3493 wine can make client requests
3494 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3495 within Linux ip-stack checksum functions
3496 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3498 213685 Undefined value propagates past dependency breaking instruction
3500 215914 Valgrind inserts bogus empty environment variable
3502 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3503 222545 shmat fails under valgind on some arm targets
3504 222560 ARM NEON support
3507 232509 Docs build fails with formatting inside <title></title> elements
3509 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3510 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3511 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3512 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3513 237485 symlink (syscall 57) is not supported on Mac OS
3514 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3516 238208 is_just_below_ESP doesn't take into account red-zone
3517 238345 valgrind passes wrong $0 when executing a shell script
3518 238679 mq_timedreceive syscall doesn't flag the reception buffer
3520 238696 fcntl command F_DUPFD_CLOEXEC not supported
3521 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3522 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3523 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
3525 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
3531 242606 unhandled syscall: setegid (in Ptrcheck)
3532 242814 Helgrind "Impossible has happened" during
3533 QApplication::initInstance();
3534 243064 Valgrind attempting to read debug information from iso
3535 243270 Make stack unwinding in Valgrind wrappers more reliable
3536 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
3537 sysno = 277 (mq_open)
3538 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
3539 244493 ARM VFP d16-d31 registers support
3540 244670 add support for audit_session_self syscall on Mac OS 10.6
3541 244921 The xml report of helgrind tool is not well format
3542 244923 In the xml report file, the <preamble> not escape the
3543 xml char, eg '<','&','>'
3544 245535 print full path names in plain text reports
3545 245925 x86-64 red zone handling problem
3546 246258 Valgrind not catching integer underruns + new [] s
3547 246311 reg/reg cmpxchg doesn't work on amd64
3548 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
3549 246888 Improve Makefile.vex.am
3550 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
3551 to [f]chmod_extended
3552 247526 IBM POWER6 (ISA 2.05) support is incomplete
3553 247561 Some leak testcases fails due to reachable addresses in
3555 247875 sizeofIRType to handle Ity_I128
3556 247894 [PATCH] unhandled syscall sys_readahead
3557 247980 Doesn't honor CFLAGS passed to configure
3558 248373 darwin10.supp is empty in the trunk
3559 248822 Linux FIBMAP ioctl has int parameter instead of long
3560 248893 [PATCH] make readdwarf.c big endianess safe to enable
3561 unwinding on big endian systems
3562 249224 Syscall 336 not supported (SYS_proc_info)
3564 249775 Incorrect scheme for detecting NEON capabilities of host CPU
3565 249943 jni JVM init fails when using valgrind
3566 249991 Valgrind incorrectly declares AESKEYGENASSIST support
3568 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
3569 250799 frexp$fenv_access_off function generates SIGILL
3570 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
3571 251251 support pclmulqdq insn
3572 251362 valgrind: ARM: attach to debugger either fails or provokes
3574 251674 Unhandled syscall 294
3577 254257 Add support for debugfiles found by build-id
3578 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
3579 254646 Wrapped functions cause stack misalignment on OS X
3580 (and possibly Linux)
3581 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
3583 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
3587 Release 3.5.0 (19 August 2009)
3588 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3589 3.5.0 is a feature release with many significant improvements and the
3590 usual collection of bug fixes. The main improvement is that Valgrind
3591 now works on Mac OS X.
3593 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
3594 and X86/Darwin. Support for recent distros and toolchain components
3595 (glibc 2.10, gcc 4.5) has been added.
3597 -------------------------
3599 Here is a short summary of the changes. Details are shown further
3602 * Support for Mac OS X (10.5.x).
3604 * Improvements and simplifications to Memcheck's leak checker.
3606 * Clarification and simplifications in various aspects of Valgrind's
3609 * XML output for Helgrind and Ptrcheck.
3611 * Performance and stability improvements for Helgrind and DRD.
3613 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
3615 * A new experimental tool, BBV, useful for computer architecture
3618 * Improved Wine support, including ability to read Windows PDB
3621 -------------------------
3623 Here are details of the above changes, followed by descriptions of
3624 many other minor changes, and a list of fixed bugs.
3627 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
3628 called "Darwin" because that is the name of the OS core, which is the
3629 level that Valgrind works at.)
3633 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
3634 because it would require work and 10.4 is only becoming less common.
3636 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
3637 fairly well. For 10.5.x, 32-bit programs are the default even on
3638 64-bit machines, so it handles most current programs.
3640 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
3641 officially supported, but simple programs at least will probably work.
3642 However, start-up is slow.
3644 - PowerPC machines are not supported.
3646 Things that don't work:
3648 - The Ptrcheck tool.
3650 - Objective-C garbage collection.
3654 - If you have Rogue Amoeba's "Instant Hijack" program installed,
3655 Valgrind will fail with a SIGTRAP at start-up. See
3656 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
3661 - You will likely find --dsymutil=yes a useful option, as error
3662 messages may be imprecise without it.
3664 - Mac OS X support is new and therefore will be less robust than the
3665 Linux support. Please report any bugs you find.
3667 - Threaded programs may run more slowly than on Linux.
3669 Many thanks to Greg Parker for developing this port over several years.
3672 * Memcheck's leak checker has been improved.
3674 - The results for --leak-check=summary now match the summary results
3675 for --leak-check=full. Previously they could differ because
3676 --leak-check=summary counted "indirectly lost" blocks and
3677 "suppressed" blocks as "definitely lost".
3679 - Blocks that are only reachable via at least one interior-pointer,
3680 but are directly pointed to by a start-pointer, were previously
3681 marked as "still reachable". They are now correctly marked as
3684 - The default value for the --leak-resolution option has been
3685 changed from "low" to "high". In general, this means that more
3686 leak reports will be produced, but each leak report will describe
3687 fewer leaked blocks.
3689 - With --leak-check=full, "definitely lost" and "possibly lost"
3690 leaks are now considered as proper errors, ie. they are counted
3691 for the "ERROR SUMMARY" and affect the behaviour of
3692 --error-exitcode. These leaks are not counted as errors if
3693 --leak-check=summary is specified, however.
3695 - Documentation for the leak checker has been improved.
3698 * Various aspects of Valgrind's text output have changed.
3700 - Valgrind's start-up message has changed. It is shorter but also
3701 includes the command being run, which makes it easier to use
3702 --trace-children=yes. An example:
3704 - Valgrind's shut-down messages have also changed. This is most
3705 noticeable with Memcheck, where the leak summary now occurs before
3706 the error summary. This change was necessary to allow leaks to be
3707 counted as proper errors (see the description of the leak checker
3708 changes above for more details). This was also necessary to fix a
3709 longstanding bug in which uses of suppressions against leaks were
3710 not "counted", leading to difficulties in maintaining suppression
3711 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3713 - Behavior of -v has changed. In previous versions, -v printed out
3714 a mixture of marginally-user-useful information, and tool/core
3715 statistics. The statistics printing has now been moved to its own
3716 flag, --stats=yes. This means -v is less verbose and more likely
3717 to convey useful end-user information.
3719 - The format of some (non-XML) stack trace entries has changed a
3720 little. Previously there were six possible forms:
3722 0x80483BF: really (a.c:20)
3723 0x80483BF: really (in /foo/a.out)
3725 0x80483BF: (within /foo/a.out)
3726 0x80483BF: ??? (a.c:20)
3729 The third and fourth of these forms have been made more consistent
3730 with the others. The six possible forms are now:
3732 0x80483BF: really (a.c:20)
3733 0x80483BF: really (in /foo/a.out)
3734 0x80483BF: really (in ???)
3735 0x80483BF: ??? (in /foo/a.out)
3736 0x80483BF: ??? (a.c:20)
3739 Stack traces produced when --xml=yes is specified are different
3743 * Helgrind and Ptrcheck now support XML output, so they can be used
3744 from GUI tools. Also, the XML output mechanism has been
3747 - The XML format has been overhauled and generalised, so it is more
3748 suitable for error reporting tools in general. The Memcheck
3749 specific aspects of it have been removed. The new format, which
3750 is an evolution of the old format, is described in
3751 docs/internals/xml-output-protocol4.txt.
3753 - Memcheck has been updated to use the new format.
3755 - Helgrind and Ptrcheck are now able to emit output in this format.
3757 - The XML output mechanism has been overhauled. XML is now output
3758 to its own file descriptor, which means that:
3760 * Valgrind can output text and XML independently.
3762 * The longstanding problem of XML output being corrupted by
3763 unexpected un-tagged text messages is solved.
3765 As before, the destination for text output is specified using
3766 --log-file=, --log-fd= or --log-socket=.
3768 As before, XML output for a tool is enabled using --xml=yes.
3770 Because there's a new XML output channel, the XML output
3771 destination is now specified by --xml-file=, --xml-fd= or
3774 Initial feedback has shown this causes some confusion. To
3775 clarify, the two envisaged usage scenarios are:
3777 (1) Normal text output. In this case, do not specify --xml=yes
3778 nor any of --xml-file=, --xml-fd= or --xml-socket=.
3780 (2) XML output. In this case, specify --xml=yes, and one of
3781 --xml-file=, --xml-fd= or --xml-socket= to select the XML
3782 destination, one of --log-file=, --log-fd= or --log-socket=
3783 to select the destination for any remaining text messages,
3784 and, importantly, -q.
3786 -q makes Valgrind completely silent on the text channel,
3787 except in the case of critical failures, such as Valgrind
3788 itself segfaulting, or failing to read debugging information.
3789 Hence, in this scenario, it suffices to check whether or not
3790 any output appeared on the text channel. If yes, then it is
3791 likely to be a critical error which should be brought to the
3792 attention of the user. If no (the text channel produced no
3793 output) then it can be assumed that the run was successful.
3795 This allows GUIs to make the critical distinction they need to
3796 make (did the run fail or not?) without having to search or
3797 filter the text output channel in any way.
3799 It is also recommended to use --child-silent-after-fork=yes in
3803 * Improvements and changes in Helgrind:
3805 - XML output, as described above
3807 - Checks for consistent association between pthread condition
3808 variables and their associated mutexes are now performed.
3810 - pthread_spinlock functions are supported.
3812 - Modest performance improvements.
3814 - Initial (skeletal) support for describing the behaviour of
3815 non-POSIX synchronisation objects through ThreadSanitizer
3816 compatible ANNOTATE_* macros.
3818 - More controllable tradeoffs between performance and the level of
3819 detail of "previous" accesses in a race. There are now three
3822 * --history-level=full. This is the default, and was also the
3823 default in 3.4.x. It shows both stacks involved in a race, but
3824 requires a lot of memory and can be very slow in programs that
3825 do many inter-thread synchronisation events.
3827 * --history-level=none. This only shows the later stack involved
3828 in a race. This can be much faster than --history-level=full,
3829 but makes it much more difficult to find the other access
3830 involved in the race.
3832 The new intermediate setting is
3834 * --history-level=approx
3836 For the earlier (other) access, two stacks are presented. The
3837 earlier access is guaranteed to be somewhere in between the two
3838 program points denoted by those stacks. This is not as useful
3839 as showing the exact stack for the previous access (as per
3840 --history-level=full), but it is better than nothing, and it's
3841 almost as fast as --history-level=none.
3844 * New features and improvements in DRD:
3846 - The error messages printed by DRD are now easier to interpret.
3847 Instead of using two different numbers to identify each thread
3848 (Valgrind thread ID and DRD thread ID), DRD does now identify
3849 threads via a single number (the DRD thread ID). Furthermore
3850 "first observed at" information is now printed for all error
3851 messages related to synchronization objects.
3853 - Added support for named semaphores (sem_open() and sem_close()).
3855 - Race conditions between pthread_barrier_wait() and
3856 pthread_barrier_destroy() calls are now reported.
3858 - Added support for custom allocators through the macros
3859 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
3860 in <valgrind/valgrind.h>). An alternative for these two macros is
3861 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
3864 - Added support for annotating non-POSIX synchronization objects
3865 through several new ANNOTATE_*() macros.
3867 - OpenMP: added support for the OpenMP runtime (libgomp) included
3868 with gcc versions 4.3.0 and 4.4.0.
3872 - Added two new command-line options (--first-race-only and
3873 --segment-merging-interval).
3876 * Genuinely atomic support for x86/amd64/ppc atomic instructions
3878 Valgrind will now preserve (memory-access) atomicity of LOCK-
3879 prefixed x86/amd64 instructions, and any others implying a global
3880 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
3882 This means that Valgrinded processes will "play nicely" in
3883 situations where communication with other processes, or the kernel,
3884 is done through shared memory and coordinated with such atomic
3885 instructions. Prior to this change, such arrangements usually
3886 resulted in hangs, races or other synchronisation failures, because
3887 Valgrind did not honour atomicity of such instructions.
3890 * A new experimental tool, BBV, has been added. BBV generates basic
3891 block vectors for use with the SimPoint analysis tool, which allows
3892 a program's overall behaviour to be approximated by running only a
3893 fraction of it. This is useful for computer architecture
3894 researchers. You can run BBV by specifying --tool=exp-bbv (the
3895 "exp-" prefix is short for "experimental"). BBV was written by
3899 * Modestly improved support for running Windows applications under
3900 Wine. In particular, initial support for reading Windows .PDB debug
3901 information has been added.
3904 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
3905 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
3909 * The Valgrind client requests VALGRIND_PRINTF and
3910 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
3911 the string was always printed immediately on its own line. Now, the
3912 string will be added to a buffer but not printed until a newline is
3913 encountered, or other Valgrind output is printed (note that for
3914 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
3915 "other Valgrind output"). This allows you to use multiple
3916 VALGRIND_PRINTF calls to build up a single output line, and also to
3917 print multiple output lines with a single request (by embedding
3918 multiple newlines in the string).
3921 * The graphs drawn by Massif's ms_print program have changed slightly:
3923 - The half-height chars '.' and ',' are no longer drawn, because
3924 they are confusing. The --y option can be used if the default
3925 y-resolution is not high enough.
3927 - Horizontal lines are now drawn after the top of a snapshot if
3928 there is a gap until the next snapshot. This makes it clear that
3929 the memory usage has not dropped to zero between snapshots.
3932 * Something that happened in 3.4.0, but wasn't clearly announced: the
3933 option --read-var-info=yes can be used by some tools (Memcheck,
3934 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
3935 variable type and location information. This makes those tools
3936 start up more slowly and increases memory consumption, but
3937 descriptions of data addresses in error messages become more
3941 * exp-Omega, an experimental instantaneous leak-detecting tool, was
3942 disabled in 3.4.0 due to a lack of interest and maintenance,
3943 although the source code was still in the distribution. The source
3944 code has now been removed from the distribution. For anyone
3945 interested, the removal occurred in SVN revision r10247.
3948 * Some changes have been made to the build system.
3950 - VEX/ is now integrated properly into the build system. This means
3951 that dependency tracking within VEX/ now works properly, "make
3952 install" will work without requiring "make" before it, and
3953 parallel builds (ie. 'make -j') now work (previously a
3954 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
3955 was effectively ignored).
3957 - The --with-vex configure option has been removed. It was of
3958 little use and removing it simplified the build system.
3960 - The location of some install files has changed. This should not
3961 affect most users. Those who might be affected:
3963 * For people who use Valgrind with MPI programs, the installed
3964 libmpiwrap.so library has moved from
3965 $(INSTALL)/<platform>/libmpiwrap.so to
3966 $(INSTALL)/libmpiwrap-<platform>.so.
3968 * For people who distribute standalone Valgrind tools, the
3969 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
3970 have moved to $(INSTALL)/libcoregrind-<platform>.a.
3972 These changes simplify the build system.
3974 - Previously, all the distributed suppression (*.supp) files were
3975 installed. Now, only default.supp is installed. This should not
3976 affect users as the other installed suppression files were not
3977 read; the fact that they were installed was a mistake.
3980 * KNOWN LIMITATIONS:
3982 - Memcheck is unusable with the Intel compiler suite version 11.1,
3983 when it generates code for SSE2-and-above capable targets. This
3984 is because of icc's use of highly optimised inlined strlen
3985 implementations. It causes Memcheck to report huge numbers of
3986 false errors even in simple programs. Helgrind and DRD may also
3989 Versions 11.0 and earlier may be OK, but this has not been
3993 The following bugs have been fixed or resolved. Note that "n-i-bz"
3994 stands for "not in bugzilla" -- that is, a bug that was reported to us
3995 but never got a bugzilla entry. We encourage you to file bugs in
3996 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3997 mailing the developers (or mailing lists) directly -- bugs that are
3998 not entered into bugzilla tend to get forgotten about or ignored.
4000 To see details of a given bug, visit
4001 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
4002 where XXXXXX is the bug number as listed below.
4004 84303 How about a LockCheck tool?
4005 91633 dereference of null ptr in vgPlain_st_basetype
4006 97452 Valgrind doesn't report any pthreads problems
4007 100628 leak-check gets assertion failure when using
4008 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
4009 108528 NPTL pthread cleanup handlers not called
4010 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
4011 110128 mallinfo is not implemented...
4012 110770 VEX: Generated files not always updated when making valgrind
4013 111102 Memcheck: problems with large (memory footprint) applications
4014 115673 Vex's decoder should never assert
4015 117564 False positive: Syscall param clone(child_tidptr) contains
4016 uninitialised byte(s)
4017 119404 executing ssh from inside valgrind fails
4018 133679 Callgrind does not write path names to sources with dwarf debug
4020 135847 configure.in problem with non gnu compilers (and possible fix)
4021 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
4022 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
4023 136230 memcheck reports "possibly lost", should be "still reachable"
4024 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
4025 137904 Valgrind reports a memory leak when using POSIX threads,
4027 139076 valgrind VT_GETSTATE error
4028 142228 complaint of elf_dynamic_do_rela in trivial usage
4029 145347 spurious warning with USBDEVFS_REAPURB
4030 148441 (wine) can't find memory leak in Wine, win32 binary
4032 148742 Leak-check fails assert on exit
4033 149878 add (proper) check for calloc integer overflow
4034 150606 Call graph is broken when using callgrind control
4035 152393 leak errors produce an exit code of 0. I need some way to
4036 cause leak errors to result in a nonzero exit code.
4037 157154 documentation (leak-resolution doc speaks about num-callers
4038 def=4) + what is a loss record
4039 159501 incorrect handling of ALSA ioctls
4040 162020 Valgrinding an empty/zero-byte file crashes valgrind
4041 162482 ppc: Valgrind crashes while reading stabs information
4042 162718 x86: avoid segment selector 0 in sys_set_thread_area()
4043 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
4044 163560 VEX/test_main.c is missing from valgrind-3.3.1
4045 164353 malloc_usable_size() doesn't return a usable size
4046 165468 Inconsistent formatting in memcheck manual -- please fix
4047 169505 main.c:286 (endOfInstr):
4048 Assertion 'ii->cost_offset == *cost_offset' failed
4049 177206 Generate default.supp during compile instead of configure
4050 177209 Configure valt_load_address based on arch+os
4051 177305 eventfd / syscall 323 patch lost
4052 179731 Tests fail to build because of inlining of non-local asm labels
4053 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
4054 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
4055 181594 Bogus warning for empty text segment
4056 181707 dwarf doesn't require enumerations to have name
4057 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4058 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4059 Assertion '!already_present' failed.
4060 185359 exp-ptrcheck: unhandled syscall getresuid()
4061 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
4062 185816 Valgrind is unable to handle debug info for files with split
4063 debug info that are prelinked afterwards
4064 185980 [darwin] unhandled syscall: sem_open
4065 186238 bbToIR_AMD64: disInstr miscalculated next %rip
4066 186507 exp-ptrcheck unhandled syscalls prctl, etc.
4067 186790 Suppression pattern used for leaks are not reported
4068 186796 Symbols with length>200 in suppression files are ignored
4069 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
4070 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
4071 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
4072 188046 bashisms in the configure script
4073 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
4074 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
4075 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
4076 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
4077 assertion fail "!lock->heldBy"
4078 188427 Add support for epoll_create1 (with patch)
4079 188530 Support for SIOCGSTAMPNS
4080 188560 Include valgrind.spec in the tarball
4081 188572 Valgrind on Mac should suppress setenv() mem leak
4082 189054 Valgrind fails to build because of duplicate non-local asm labels
4083 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
4084 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
4085 189763 drd assertion failure: s_threadinfo[tid].is_recording
4086 190219 unhandled syscall: 328 (x86-linux)
4087 190391 dup of 181394; see above
4088 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
4089 190820 No debug information on powerpc-linux
4090 191095 PATCH: Improve usbdevfs ioctl handling
4091 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
4093 191189 --xml=yes should obey --gen-suppressions=all
4094 191192 syslog() needs a suppression on macosx
4095 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
4096 191761 getrlimit on MacOSX
4097 191992 multiple --fn-skip only works sometimes; dependent on order
4098 192634 V. reports "aspacem sync_check_mapping_callback:
4099 segment mismatch" on Darwin
4100 192954 __extension__ missing on 2 client requests
4101 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
4102 194474 "INSTALL" file has different build instructions than "README"
4103 194671 Unhandled syscall (sem_wait?) from mac valgrind
4104 195069 memcheck: reports leak (memory still reachable) for
4106 195169 drd: (vgDrd_barrier_post_wait):
4107 Assertion 'r->sg[p->post_iteration]' failed.
4108 195268 valgrind --log-file doesn't accept ~/...
4109 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
4110 195860 WARNING: unhandled syscall: unix:223
4111 196528 need a error suppression for pthread_rwlock_init under os x?
4112 197227 Support aio_* syscalls on Darwin
4113 197456 valgrind should reject --suppressions=(directory)
4114 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
4115 197591 unhandled syscall 27 (mincore)
4116 197793 Merge DCAS branch to the trunk == 85756, 142103
4117 197794 Avoid duplicate filenames in Vex
4118 197898 make check fails on current SVN
4119 197901 make check fails also under exp-ptrcheck in current SVN
4120 197929 Make --leak-resolution=high the default
4121 197930 Reduce spacing between leak reports
4122 197933 Print command line of client at start-up, and shorten preamble
4123 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
4124 198395 add BBV to the distribution as an experimental tool
4125 198624 Missing syscalls on Darwin: 82, 167, 281, 347
4126 198649 callgrind_annotate doesn't cumulate counters
4127 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
4128 199977 Valgrind complains about an unrecognized instruction in the
4129 atomic_incs test program
4130 200029 valgrind isn't able to read Fedora 12 debuginfo
4131 200760 darwin unhandled syscall: unix:284
4132 200827 DRD doesn't work on Mac OS X
4133 200990 VG_(read_millisecond_timer)() does not work correctly
4134 201016 Valgrind does not support pthread_kill() on Mac OS
4135 201169 Document --read-var-info
4136 201323 Pre-3.5.0 performance sanity checking
4137 201384 Review user manual for the 3.5.0 release
4138 201585 mfpvr not implemented on ppc
4139 201708 tests failing because x86 direction flag is left set
4140 201757 Valgrind doesn't handle any recent sys_futex additions
4141 204377 64-bit valgrind can not start a shell script
4142 (with #!/path/to/shell) if the shell is a 32-bit executable
4143 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
4144 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
4145 about memory allocation events with memory access tracing enabled
4146 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
4148 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
4152 Release 3.4.1 (28 February 2009)
4153 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4154 3.4.1 is a bug-fix release that fixes some regressions and assertion
4155 failures in debug info reading in 3.4.0, most notably incorrect stack
4156 traces on amd64-linux on older (glibc-2.3 based) systems. Various
4157 other debug info problems are also fixed. A number of bugs in the
4158 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
4160 In view of the fact that 3.4.0 contains user-visible regressions
4161 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
4162 encouraged to ship 3.4.1 in preference to 3.4.0.
4164 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
4165 bugzilla" -- that is, a bug that was reported to us but never got a
4166 bugzilla entry. We encourage you to file bugs in bugzilla
4167 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
4168 developers (or mailing lists) directly -- bugs that are not entered
4169 into bugzilla tend to get forgotten about or ignored.
4171 n-i-bz Fix various bugs reading icc-11 generated debug info
4172 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
4173 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
4174 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
4175 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
4176 179618 exp-ptrcheck crashed / exit prematurely
4177 179624 helgrind: false positive races with pthread_create and
4178 recv/open/close/read
4179 134207 pkg-config output contains @VG_PLATFORM@
4180 176926 floating point exception at valgrind startup with PPC 440EPX
4181 181594 Bogus warning for empty text segment
4182 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
4183 181707 Dwarf3 doesn't require enumerations to have name
4184 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
4185 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
4186 Assertion '!already_present' failed.
4187 185359 exp-ptrcheck unhandled syscall getresuid()
4189 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
4190 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
4194 Release 3.4.0 (2 January 2009)
4195 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4196 3.4.0 is a feature release with many significant improvements and the
4197 usual collection of bug fixes. This release supports X86/Linux,
4198 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
4199 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
4201 3.4.0 brings some significant tool improvements. Memcheck can now
4202 report the origin of uninitialised values, the thread checkers
4203 Helgrind and DRD are much improved, and we have a new experimental
4204 tool, exp-Ptrcheck, which is able to detect overruns of stack and
4205 global arrays. In detail:
4207 * Memcheck is now able to track the origin of uninitialised values.
4208 When it reports an uninitialised value error, it will try to show
4209 the origin of the value, as either a heap or stack allocation.
4210 Origin tracking is expensive and so is not enabled by default. To
4211 use it, specify --track-origins=yes. Memcheck's speed will be
4212 essentially halved, and memory usage will be significantly
4213 increased. Nevertheless it can drastically reduce the effort
4214 required to identify the root cause of uninitialised value errors,
4215 and so is often a programmer productivity win, despite running more
4218 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4219 3.4.0, will be released shortly.
4221 * Helgrind's race detection algorithm has been completely redesigned
4222 and reimplemented, to address usability and scalability concerns:
4224 - The new algorithm has a lower false-error rate: it is much less
4225 likely to report races that do not really exist.
4227 - Helgrind will display full call stacks for both accesses involved
4228 in a race. This makes it easier to identify the root causes of
4231 - Limitations on the size of program that can run have been removed.
4233 - Performance has been modestly improved, although that is very
4236 - Direct support for Qt4 threading has been added.
4238 - pthread_barriers are now directly supported.
4240 - Helgrind works well on all supported Linux targets.
4242 * The DRD thread debugging tool has seen major improvements:
4244 - Greatly improved performance and significantly reduced memory
4247 - Support for several major threading libraries (Boost.Thread, Qt4,
4248 glib, OpenMP) has been added.
4250 - Support for atomic instructions, POSIX semaphores, barriers and
4251 reader-writer locks has been added.
4253 - Works now on PowerPC CPUs too.
4255 - Added support for printing thread stack usage at thread exit time.
4257 - Added support for debugging lock contention.
4259 - Added a manual for Drd.
4261 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4262 checks for misuses of pointers. In that sense it is a bit like
4263 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4264 detect overruns of stack and global arrays, it can detect
4265 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4266 detect accesses heap blocks that have been freed a very long time
4267 ago (millions of blocks in the past).
4269 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4270 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4271 of the main Valgrind documentation. As this is an experimental
4272 tool, we would be particularly interested in hearing about your
4273 experiences with it.
4275 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4276 longer built by default, although the code remains in the repository
4277 and the tarball. This is due to three factors: a perceived lack of
4278 users, a lack of maintenance, and concerns that it may not be
4279 possible to achieve reliable operation using the existing design.
4281 * As usual, support for the latest Linux distros and toolchain
4282 components has been added. It should work well on Fedora Core 10,
4283 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4284 state) is supported, as is glibc-2.9. The C++ demangler has been
4285 updated so as to work well with C++ compiled by even the most recent
4288 * You can now use frame-level wildcards in suppressions. This was a
4289 frequently-requested enhancement. A line "..." in a suppression now
4290 matches zero or more frames. This makes it easier to write
4291 suppressions which are precise yet insensitive to changes in
4294 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4296 * Very basic support for IBM Power6 has been added (64-bit processes only).
4298 * Valgrind is now cross-compilable. For example, it is possible to
4299 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4300 on a ppc32/64-linux target.
4302 * You can set the main thread's stack size at startup using the
4303 new --main-stacksize= flag (subject of course to ulimit settings).
4304 This is useful for running apps that need a lot of stack space.
4306 * The limitation that you can't use --trace-children=yes together
4307 with --db-attach=yes has been removed.
4309 * The following bugs have been fixed. Note that "n-i-bz" stands for
4310 "not in bugzilla" -- that is, a bug that was reported to us but
4311 never got a bugzilla entry. We encourage you to file bugs in
4312 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4313 mailing the developers (or mailing lists) directly.
4315 n-i-bz Make return types for some client requests 64-bit clean
4316 n-i-bz glibc 2.9 support
4317 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4318 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4319 n-i-bz Building in an env without gdb gives bogus gdb attach
4320 92456 Tracing the origin of uninitialised memory
4321 106497 Valgrind does not demangle some C++ template symbols
4323 151612 Suppression with "..." (frame-level wildcards in .supp files)
4324 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4325 159285 unhandled syscall:25 (stime, on x86-linux)
4326 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4327 160954 ppc build of valgrind crashes with illegal instruction (isel)
4328 160956 mallinfo implementation, w/ patch
4329 162092 Valgrind fails to start gnome-system-monitor
4330 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4331 163794 assertion failure with "--track-origins=yes"
4332 163933 sigcontext.err and .trapno must be set together
4333 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4334 164476 Missing kernel module loading system calls
4335 164669 SVN regression: mmap() drops posix file locks
4336 166581 Callgrind output corruption when program forks
4337 167288 Patch file for missing system calls on Cell BE
4338 168943 unsupported scas instruction pentium
4339 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4340 172417 x86->IR: 0x82 ...
4341 172563 amd64->IR: 0xD9 0xF5 - fprem1
4342 173099 .lds linker script generation error
4343 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4344 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4346 174908 --log-file value not expanded correctly for core file
4347 175044 Add lookup_dcookie for amd64
4348 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4350 Developer-visible changes:
4352 * Valgrind's debug-info reading machinery has been majorly overhauled.
4353 It can now correctly establish the addresses for ELF data symbols,
4354 which is something that has never worked properly before now.
4356 Also, Valgrind can now read DWARF3 type and location information for
4357 stack and global variables. This makes it possible to use the
4358 framework to build tools that rely on knowing the type and locations
4359 of stack and global variables, for example exp-Ptrcheck.
4361 Reading of such information is disabled by default, because most
4362 tools don't need it, and because it is expensive in space and time.
4363 However, you can force Valgrind to read it, using the
4364 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4365 make use of such information, if present, to provide source-level
4366 descriptions of data addresses in the error messages they create.
4368 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4369 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).