1 Release 3.17.0 (?? ??? 202?)
2 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4 3.17.0 is a feature release with many improvements and the usual collection
7 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
8 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
9 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
10 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
11 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
13 * ==================== CORE CHANGES ===================
15 * ==================== TOOL CHANGES ====================
27 * ==================== OTHER CHANGES ====================
29 * ==================== FIXED BUGS ====================
31 The following bugs have been fixed or resolved. Note that "n-i-bz"
32 stands for "not in bugzilla" -- that is, a bug that was reported to us
33 but never got a bugzilla entry. We encourage you to file bugs in
34 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
35 than mailing the developers (or mailing lists) directly -- bugs that
36 are not entered into bugzilla tend to get forgotten about or ignored.
38 To see details of a given bug, visit
39 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
40 where XXXXXX is the bug number as listed below.
42 345077 linux syscall execveat support (linux 3.19)
43 n-i-bz helgrind: If hg_cli__realloc fails, return NULL.
47 Release 3.16.1 (?? June 2020)
48 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
50 3.16.1 is identical to 3.16.0, except that it fixes the following bug that
51 was discovered late in the 3.16.0 release process:
53 422715 valgrind 3.16.0 on 32-bit x86: vex: the `impossible' happened:
54 expr_is_guardable: unhandled expr
58 Release 3.16.0 (27 May 2020)
59 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
61 3.16.0 is a feature release with many improvements and the usual collection of
64 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
65 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
66 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
67 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
68 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
70 * ==================== CORE CHANGES ===================
72 * It is now possible to dynamically change the value of many command line
73 options while your program (or its children) are running under Valgrind.
75 To see the list of dynamically changeable options, run
76 "valgrind --help-dyn-options".
78 You can change the options from the shell by using vgdb to launch
79 the monitor command "v.clo <clo option>...".
80 The same monitor command can be used from a gdb connected
81 to the valgrind gdbserver.
82 Your program can also change the dynamically changeable options using
83 the client request VALGRIND_CLO_CHANGE(option).
85 * ================== PLATFORM CHANGES =================
87 * MIPS: preliminary support for nanoMIPS instruction set has been added.
89 * ==================== TOOL CHANGES ====================
93 - The implicit memcpy done by each call to realloc now counts towards the
94 read and write counts of resized heap blocks, making those counts higher
99 - cg_annotate's --auto and --show-percs options now default to 'yes', because
100 they are usually wanted.
104 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
105 because they are usually wanted.
107 - The command option --collect-systime has been enhanced to specify
108 the unit used to record the elapsed time spent during system calls.
109 The command option now accepts the values no|yes|msec|usec|nsec,
110 where yes is a synonym of msec. When giving the value nsec, the
111 system cpu time of system calls is also recorded.
115 - Several memcheck options are now dynamically changeable.
116 Use valgrind --help-dyn-options to list them.
118 - The release 3.15 introduced a backward incompatible change for
119 some suppression entries related to preadv and pwritev syscalls.
120 When reading a suppression entry using the unsupported 3.14 format,
121 valgrind will now produce a warning to say the suppression entry will not
122 work, and suggest the needed change.
124 - Significantly fewer false positive errors on optimised code generated by
125 Clang and GCC. In particular, Memcheck now deals better with the
126 situation where the compiler will transform C-level "A && B" into "B && A"
127 under certain circumstances (in which the transformation is valid).
128 Handling of integer equality/non-equality checks on partially defined
129 values is also improved on some architectures.
133 - The exprimental Stack and Global Array Checking tool has been removed.
134 It only ever worked on x86 and amd64, and even on those it had a
135 high false positive rate and was slow. An alternative for detecting
136 stack and global array overruns is using the AddressSanitizer (ASAN)
137 facility of the GCC and Clang compilers, which require you to rebuild
138 your code with -fsanitize=address.
140 * ==================== OTHER CHANGES ====================
142 * New and modified GDB server monitor features:
144 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
146 - The gdbserver monitor commands that require an address and an optional
147 length argument now accepts the alternate 'C like' syntax "address[length]".
148 For example, the memcheck command "monitor who_points_at 0x12345678 120"
149 can now also be given as "monitor who_points_at 0x12345678[120]".
151 * ==================== FIXED BUGS ====================
153 The following bugs have been fixed or resolved. Note that "n-i-bz"
154 stands for "not in bugzilla" -- that is, a bug that was reported to us
155 but never got a bugzilla entry. We encourage you to file bugs in
156 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
157 than mailing the developers (or mailing lists) directly -- bugs that
158 are not entered into bugzilla tend to get forgotten about or ignored.
160 To see details of a given bug, visit
161 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
162 where XXXXXX is the bug number as listed below.
164 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
165 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
166 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
167 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
168 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
169 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
170 400872 Add nanoMIPS support to Valgrind
171 403212 drd/tests/trylock hangs on FreeBSD
172 404406 s390x: z14 miscellaneous instructions not implemented
173 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
174 406561 mcinfcallWSRU gdbserver_test fails on ppc64
175 406824 Unsupported baseline
176 407218 Add support for the copy_file_range syscall
177 407307 Intercept stpcpy also in ld.so for arm64
178 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
180 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
181 408009 Expose rdrand and f16c even on avx if host cpu supports them
182 408091 Missing pkey syscalls
183 408414 Add support for missing for preadv2 and pwritev2 syscalls
184 409141 Valgrind hangs when SIGKILLed
185 409206 Support for Linux PPS and PTP ioctls
186 409367 exit_group() after signal to thread waiting in futex() causes hangs
187 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
188 409780 References to non-existent configure.in
189 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
190 410599 Non-deterministic behaviour of pth_self_kill_15_other test
191 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
192 411134 Allow the user to change a set of command line options during execution
193 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
194 412344 Problem setting mips flags with specific paths
195 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
196 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
197 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
198 413603 callgrind_annotate/cg_annotate truncate function names at '#'
199 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
200 415136 ARMv8.1 Compare-and-Swap instructions are not supported
201 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
202 416239 valgrind crashes when handling clock_adjtime
203 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
204 416286 DRD reports "conflicting load" error on std::mutex::lock()
205 416301 s390x: "compare and signal" not supported
206 416387 finit_module and bpf syscalls are unhandled on arm64
207 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
208 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
209 416753 new 32bit time syscalls for 2038+
210 417075 pwritev(vector[...]) suppression ignored
211 417075 is not fixed, but incompatible supp entries are detected
212 and a warning is produced for these.
213 417187 [MIPS] Conditional branch problem since 'grail' changes
214 417238 Test memcheck/tests/vbit-test fails on mips64 BE
215 417266 Make memcheck/tests/linux/sigqueue usable with musl
216 417281 s390x: /bin/true segfaults with "grail" enabled
217 417427 commit to fix vki_siginfo_t definition created numerous regression
219 417452 s390_insn_store_emit: dst->tag for HRcVec128
220 417578 Add suppressions for glibc DTV leaks
221 417906 clone with CLONE_VFORK and no CLONE_VM fails
222 418004 Grail code additions break ppc64.
223 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
224 418997 s390x: Support Iex_ITE for float and vector types
225 419503 s390x: Avoid modifying registers returned from isel functions
226 421321 gcc10 arm64 build needs __getauxval for linking with libgcc
227 421570 std_mutex fails on Arm v8.1 h/w
228 n-i-bz Fix minor one time leaks in dhat.
229 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
230 n-i-bz Add support for the Linux io_uring system calls
231 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
232 n-i-bz Fix non-glibc build of test suite with s390x_features
233 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
235 (3.16.0.RC1: 18 May 2020, git 6052ee66a0cf5234e8e2a2b49a8760226bc13b92)
236 (3.16.0.RC2: 19 May 2020, git 940ec1ca69a09f7fdae3e800b7359f85c13c4b37)
237 (3.16.0: 27 May 2020, git bf5e647edb9e96cbd5c57cc944984402eeee296d)
241 Release 3.15.0 (12 April 2019)
242 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
244 3.15.0 is a feature release with many improvements and the usual collection of
247 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
248 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
249 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
250 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
251 support for X86/macOS 10.13 and AMD64/macOS 10.13.
253 * ==================== CORE CHANGES ===================
255 * The XTree Massif output format now makes use of the information obtained
256 when specifying --read-inline-info=yes.
258 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
260 * ==================== TOOL CHANGES ====================
264 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
265 it has been promoted from an experimental tool to a regular tool. Run it
266 with --tool=dhat instead of --tool=exp-dhat.
268 - DHAT now prints only minimal data when the program ends, instead writing
269 the bulk of the profiling data to a file. As a result, the --show-top-n
270 and --sort-by options have been removed.
272 - Profile results can be viewed with the new viewer, dh_view.html. When
273 a run ends, a short message is printed, explaining how to view the result.
275 - See the documentation for more details.
279 - cg_annotate has a new option, --show-percs, which prints percentages next
284 - callgrind_annotate has a new option, --show-percs, which prints percentages
285 next to all event counts.
287 - callgrind_annotate now inserts commas in call counts, and
288 sort the caller/callee lists in the call tree.
292 - The default value for --read-inline-info is now "yes" on
293 Linux/Android/Solaris. It is still "no" on other OS.
297 - The option --xtree-leak=yes (to output leak result in xtree format)
298 automatically activates the option --show-leak-kinds=all, as xtree
299 visualisation tools such as kcachegrind can in any case select what kind
300 of leak to visualise.
302 - There has been further work to avoid false positives. In particular,
303 integer equality on partially defined inputs (C == and !=) is now handled
306 * ==================== OTHER CHANGES ====================
308 * The new option --show-error-list=no|yes displays, at the end of the run, the
309 list of detected errors and the used suppressions. Prior to this change,
310 showing this information could only be done by specifying "-v -v", but that
311 also produced a lot of other possibly-non-useful messages. The option -s is
312 equivalent to --show-error-list=yes.
314 * ==================== FIXED BUGS ====================
316 The following bugs have been fixed or resolved. Note that "n-i-bz"
317 stands for "not in bugzilla" -- that is, a bug that was reported to us
318 but never got a bugzilla entry. We encourage you to file bugs in
319 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
320 than mailing the developers (or mailing lists) directly -- bugs that
321 are not entered into bugzilla tend to get forgotten about or ignored.
323 To see details of a given bug, visit
324 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
325 where XXXXXX is the bug number as listed below.
327 385411 s390x: z13 vector floating-point instructions not implemented
328 397187 z13 vector register support for vgdb gdbserver
329 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
330 398870 Please add support for instruction vcvtps2ph
331 399287 amd64 front end: Illegal Instruction vcmptrueps
332 399301 Use inlined frames in Massif XTree output.
333 399322 Improve callgrind_annotate output
334 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
335 400164 helgrind test encounters mips x-compiler warnings and assembler error
336 400490 s390x: VRs allocated as if separate from FPRs
337 400491 s390x: Operand of LOCH treated as unsigned integer
338 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
339 options, which specify a mips64 processor
340 401112 LLVM 5.0 generates comparison against partially initialized data
341 401277 More bugs in z13 support
342 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
343 401578 drd: crashes sometimes on fork()
344 401627 memcheck errors with glibc avx2 optimized wcsncmp
345 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
346 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
347 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
349 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
350 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
351 402123 invalid assembler opcodes for mips32r2
352 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
353 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
354 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
355 402351 mips64 libvexmultiarch_test fails on s390x
357 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
358 402480 Do not use %rsp in clobber list
359 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
360 402515 Implement new option --show-error-list=no|yes / -s
361 402519 POWER 3.0 addex instruction incorrectly implemented
362 402781 Redo the cache used to process indirect branch targets
363 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
364 403552 s390x: wrong facility bit checked for vector facility
365 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
366 404638 Add VG_(replaceIndexXA)
367 404843 s390x: backtrace sometimes ends prematurely
368 404888 autotools cleanup series
369 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
370 405182 Valgrind fails to build with Clang
371 405205 filter_libc: remove the line holding the futex syscall error entirely
372 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
373 the upper and lower 32-bits of the 64-bit result
374 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
375 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
376 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
378 405403 s390x disassembler cannot be used on x86
379 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
380 405458 MIPS mkFormVEC arguments swapped?
381 405716 drd: Fix an integer overflow in the stack margin calculation
382 405722 Support arm64 core dump
383 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
384 of the 64-bit destination field.
385 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
386 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
387 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
389 406256 PPC64, vector floating point instructions don't handle subnormal
390 according to VSCR[NJ] bit setting.
391 406352 cachegrind/callgrind fails ann tests because of missing a.c
392 406354 dhat is broken on x86 (32bit)
393 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
394 406357 gdbserver_tests fails because of gdb output change
395 406360 memcheck/tests/libstdc++.supp needs more supression variants
396 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
397 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
398 407340 PPC64, does not support the vlogefp, vexptefp instructions.
399 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
400 n-i-bz Fix callgrind_annotate non deterministic order for equal total
401 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
402 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
403 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
405 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
406 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
407 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
411 Release 3.14.0 (9 October 2018)
412 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
414 3.14.0 is a feature release with many improvements and the usual collection of
417 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
418 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
419 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
420 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
421 support for X86/macOS 10.13, AMD64/macOS 10.13.
423 * ==================== CORE CHANGES ===================
425 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
426 debug info for unloaded code. This allows saved stack traces (e.g. for
427 memory leaks) to include file/line info for code that has been dlclose'd (or
428 similar). See the user manual for more information and known limitations.
430 * Ability to specify suppressions based on source file name and line number.
432 * Majorly overhauled register allocator. No end-user changes, but the JIT
433 generates code a bit more quickly now.
435 * ================== PLATFORM CHANGES =================
437 * Preliminary support for macOS 10.13 has been added.
439 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
441 * mips: support for MIPS SIMD architecture (MSA) has been added.
443 * mips: support for MIPS N32 ABI has been added.
445 * s390: partial support for vector instructions (integer and string) has been
448 * ==================== TOOL CHANGES ====================
450 * Helgrind: Addition of a flag
451 --delta-stacktrace=no|yes [yes on linux amd64/x86]
452 which specifies how full history stack traces should be computed.
453 Setting this to =yes can speed up Helgrind by 25% when using
454 --history-level=full.
456 * Memcheck: reduced false positive rate for optimised code created by Clang 6
457 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
458 blocks more carefully to determine where it can avoid expensive definedness
459 checks without loss of precision. This is controlled by the flag
460 --expensive-definedness-checks=no|auto|yes [auto].
462 * ==================== OTHER CHANGES ====================
464 * Valgrind is now buildable with link-time optimisation (LTO). A new
465 configure option --enable-lto=yes allows building Valgrind with LTO. If the
466 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
467 Note that if you are doing Valgrind development, --enable-lto=yes massively
468 slows down the build process.
470 * ==================== FIXED BUGS ====================
472 The following bugs have been fixed or resolved. Note that "n-i-bz"
473 stands for "not in bugzilla" -- that is, a bug that was reported to us
474 but never got a bugzilla entry. We encourage you to file bugs in
475 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
476 than mailing the developers (or mailing lists) directly -- bugs that
477 are not entered into bugzilla tend to get forgotten about or ignored.
479 To see details of a given bug, visit
480 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
481 where XXXXXX is the bug number as listed below.
483 79362 Debug info is lost for .so files when they are dlclose'd
484 208052 strlcpy error when n = 0
485 255603 exp-sgcheck Assertion '!already_present' failed
486 338252 building valgrind with -flto (link time optimisation) fails
487 345763 MIPS N32 ABI support
488 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
489 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
490 372347 Replacement problem of the additional c++14/c++17 new/delete operators
491 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
492 376257 helgrind history full speed up using a cached stack
493 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
495 379748 Fix missing pselect syscall (OS X 10.11)
496 379754 Fix missing syscall ulock_wait (OS X 10.12)
497 380397 s390x: __GI_strcspn() replacemenet needed
498 381162 possible array overrun in VEX register allocator
499 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
500 381274 powerpc too chatty even with --sigill-diagnostics=no
501 381289 epoll_pwait can have a NULL sigmask
502 381553 VEX register allocator v3
503 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
504 381769 Use ucontext_t instead of struct ucontext
505 381805 arm32 needs ld.so index hardwire for new glibc security fixes
506 382256 gz compiler flag test doesn't work for gold
507 382407 vg_perf needs "--terse" command line option
508 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
509 382563 MIPS MSA ASE support
510 382998 xml-socket doesn't work
511 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
512 383723 Fix missing kevent_qos syscall (macOS 10.11)
513 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
514 384096 Mention AddrCheck at Memcheck's command line option [..]
515 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
516 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
517 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
518 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
519 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
520 384337 performance improvements to VEX register allocator v2 and v3
521 384526 reduce number of spill insns generated by VEX register allocator v3
522 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
523 384631 Sanitise client args as printed with -v
524 384633 Add a simple progress-reporting facility
525 384987 VEX regalloc: allocate caller-save registers for short lived vregs
526 385055 PPC VEX temporary storage exhausted
527 385182 PPC64 is missing support for the DSCR
528 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
529 385207 PPC64, generate_store_FPRF() generates too many Iops
530 385208 PPC64, xxperm instruction exhausts temporary memory
531 385210 PPC64, vpermr instruction could exhaust temporary memory
532 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
533 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
534 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
535 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
536 385408 s390x: z13 vector "support" instructions not implemented
537 385409 s390x: z13 vector integer instructions not implemented
538 385410 s390x: z13 vector string instructions not implemented
539 385412 s390x: new non-vector z13 instructions not implemented
540 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
541 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
542 385939 Optionally exit on the first error
543 386318 valgrind.org/info/tools.html is missing SGCheck
544 386425 running valgrind + wine on armv7l gives illegal opcode
545 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
546 387410 MIPSr6 support
547 387664 Memcheck: make expensive-definedness-checks be the default
548 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
549 387766 asm shifts cause false positive "Conditional jump or move depends
550 on uninitialised value"
551 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
552 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
553 388786 Support bpf syscall in amd64 Linux
554 388862 Add replacements for wmemchr and wcsnlen on Linux
555 389065 valgrind meets gcc flag -Wlogical-op
556 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
557 390471 suppression by specification of source-file line number
558 390723 make xtree dump files world wide readable, similar to log files
559 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
560 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
561 392118 unhandled amd64-linux syscall: 332 (statx)
562 392449 callgrind not clearing the number of calls properly
563 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
564 lxssp, stxssp and stxvl instructions.
565 393023 callgrind_control risks using the wrong vgdb
566 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
567 393099 posix_memalign() invalid write if alignment == 0
568 393146 failing assert "is_DebugInfo_active(di)"
569 395709 PPC64 is missing support for the xvnegsp instruction
570 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
572 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
573 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
574 396839 s390x: Trap instructions not implemented
575 396887 arch_prctl should return EINVAL on unknown option
576 == 397286 crash before launching binary (Unsupported arch_prctl option)
577 == 397393 valgrind: the 'impossible' happened: (Archlinux)
578 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
579 396906 compile tests failure on mips32-linux: broken inline asm in tests on
581 397012 glibc ld.so uses arch_prctl on i386
582 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
583 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
584 397424 glibc 2.27 and gdb_server tests
585 398028 Assertion `cfsi_fits` failing in simple C program
586 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
588 n-i-bz Fix missing workq_ops operations (macOS)
589 n-i-bz fix bug in strspn replacement
590 n-i-bz Add support for the Linux BLKFLSBUF ioctl
591 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
592 n-i-bz Fix possible stack trashing by semctl syscall wrapping
593 n-i-bz Add support for the Linux membarrier() system call
594 n-i-bz x86 front end: recognise and handle UD2 correctly
595 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
596 correctly aligned before entering the handler.
598 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
599 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
600 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
604 Release 3.13.0 (15 June 2017)
605 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
607 3.13.0 is a feature release with many improvements and the usual collection of
610 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
611 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
612 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
613 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
615 * ==================== CORE CHANGES ===================
617 * The translation cache size has been increased to keep up with the demands of
618 large applications. The maximum number of sectors has increased from 24 to
619 48. The default number of sectors has increased from 16 to 32 on all
620 targets except Android, where the increase is from 6 to 12.
622 * The amount of memory that Valgrind can use has been increased from 64GB to
623 128GB. In particular this means your application can allocate up to about
624 60GB when running on Memcheck.
626 * Valgrind's default load address has been changed from 0x3800'0000 to
627 0x5800'0000, so as to make it possible to load larger executables. This
628 should make it possible to load executables of size at least 1200MB.
630 * A massive spaceleak caused by reading compressed debuginfo files has been
631 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
634 * The C++ demangler has been updated.
636 * Support for demangling Rust symbols has been added.
638 * A new representation of stack traces, the "XTree", has been added. An XTree
639 is a tree of stacktraces with data associated with the stacktraces. This is
640 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
641 consumption of your program. Reporting is controlled by the new options
642 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
644 A report can also be produced on demand using the gdbserver monitor command
645 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
646 format' and 'massif format. The existing visualisers for these formats (e.g.
647 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
648 analyse these reports.
650 Memcheck can also produce XTree leak reports using the Callgrind file
651 format. For more details, see the user manual.
653 * ================== PLATFORM CHANGES =================
655 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
657 * amd64: fixes for JIT failure problems on long AVX2 code blocks
659 * amd64 and x86: support for CET prefixes has been added
661 * arm32: a few missing ARMv8 instructions have been implemented
663 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
664 Store-Conditional instructions has been added. This is to deal with
665 processor implementations that implement the LL/SC specifications strictly
666 and as a result cause Valgrind to hang in certain situations. The
667 alternative implementation is automatically enabled at startup, as required.
668 You can use the option --sim-hints=fallback-llsc to force-enable it if you
671 * Support for OSX 10.12 has been improved.
673 * On Linux, clone handling has been improved to honour CLONE_VFORK that
674 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
675 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
676 CLONE_VM exact semantics will (still) not work.
678 * The TileGX/Linux port has been removed because it appears to be both unused
681 * ==================== TOOL CHANGES ====================
685 - Memcheck should give fewer false positives when running optimised
686 Clang/LLVM generated code.
688 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
690 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
691 to produce the end of execution leak report in a xtree callgrind format
694 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
695 the leak report in an xtree file.
699 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
701 - For some workloads (typically, for big applications), Massif memory
702 consumption and CPU consumption has decreased significantly.
706 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
708 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
709 for Ada gnat compiled applications.
711 * ==================== OTHER CHANGES ====================
713 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
714 append the inner guest stacktrace to the inner host stacktrace. This helps
715 to investigate the errors reported by the outer, when they are caused by the
716 inner guest program (such as an inner regtest). See README_DEVELOPERS for
719 * To allow fast detection of callgrind files by desktop environments and file
720 managers, the format was extended to have an optional first line that
721 uniquely identifies the format ("# callgrind format"). Callgrind creates
722 this line now, as does the new xtree functionality.
724 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
725 have a new %n format letter that is replaced by a sequence number.
727 * "--version -v" now shows the SVN revision numbers from which Valgrind was
730 * ==================== FIXED BUGS ====================
732 The following bugs have been fixed or resolved. Note that "n-i-bz"
733 stands for "not in bugzilla" -- that is, a bug that was reported to us
734 but never got a bugzilla entry. We encourage you to file bugs in
735 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
736 than mailing the developers (or mailing lists) directly -- bugs that
737 are not entered into bugzilla tend to get forgotten about or ignored.
739 To see details of a given bug, visit
740 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
741 where XXXXXX is the bug number as listed below.
743 162848 --log-file output isn't split when a program forks
744 340777 Illegal instruction on mips (ar71xx)
745 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
746 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
747 to a different stack.
748 344139 x86 stack-seg overrides, needed by the Wine people
749 344524 store conditional of guest applications always fail - observed on
751 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
752 352395 Please provide SVN revision info in --version -v
753 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
754 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
755 INVALID_POSIX_THREADID' failed
756 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
757 barrier implementation
758 358697 valgrind.h: Some code remains even when defining NVALGRIND
759 359202 Add musl libc configure/compile
760 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
761 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
762 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
763 362223 assertion failed when .valgrindrc is a directory instead of a file
764 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
765 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
766 368507 can't malloc chunks larger than about 34GB
767 368529 Android arm target link error, missing atexit and pthread_atfork
768 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
769 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
770 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
771 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
772 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
773 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
774 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
775 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
776 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
777 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
778 370028 Reduce the number of compiler warnings on MIPS platforms
779 370635 arm64 missing syscall getcpu
780 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
781 371227 Clean AArch64 syscall table
782 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
783 371471 Valgrind complains about non legit memory leaks on placement new (C++)
784 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
785 371503 disInstr(arm64): unhandled instruction 0xF89F0000
786 371869 support '%' in symbol Z-encoding
787 371916 execution tree xtree concept
788 372120 c++ demangler demangles symbols which are not c++
789 372185 Support of valgrind on ARMv8 with 32 bit executable
790 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
791 372195 Power PC, xxsel instruction is not always recognized.
792 372504 Hanging on exit_group
793 372600 process loops forever when fatal signals are arriving quickly
794 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
795 373046 Stacks registered by core are never deregistered
796 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
797 373086 Implement additional Xen hypercalls
798 373192 Calling posix_spawn in glibc 2.24 completely broken
799 373488 Support for fanotify API on ARM64 architecture
800 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
801 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
802 373938 const IRExpr arguments for matchIRExpr()
803 374719 some spelling fixes
804 374963 increase valgrind's load address to prevent mmap failure
805 375514 valgrind_get_tls_addr() does not work in case of static TLS
806 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
807 for ML_(find_rx_mapping)()
808 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
809 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
810 == 377159 "vex: the `impossible' happened" still present
811 == 375150 Assertion 'tres.status == VexTransOK' failed
812 == 378068 valgrind crashes on AVX2 function in FFmpeg
813 376142 Segfaults on MIPS Cavium Octeon boards
814 376279 disInstr(arm64): unhandled instruction 0xD50320FF
815 376455 Solaris: unhandled syscall lgrpsys(180)
816 376518 Solaris: unhandled fast trap getlgrp(6)
817 376611 ppc64 and arm64 don't know about prlimit64 syscall
818 376729 PPC64, remove R2 from the clobber list
820 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
821 to be wrongly marked as addressable
822 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
823 PIE enabled by default
824 377376 memcheck/tests/linux/getregset fails with glibc2.24
825 377427 PPC64, lxv instruction failing on odd destination register
826 377478 PPC64: ISA 3.0 setup fixes
827 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
828 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
829 and 2 args for FUTEX_TRYLOCK_PI
830 377717 Fix massive space leak when reading compressed debuginfo sections
831 377891 Update Xen 4.6 domctl wrappers
832 377930 fcntl syscall wrapper is missing flock structure check
833 378524 libvexmultiarch_test regression on s390x and ppc64
834 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
835 378673 Update libiberty demangler
836 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
837 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
838 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
839 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
840 (task_register_dyld_image_infos)
841 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
842 (task_register_dyld_shared_cache_image_info)
843 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
844 379473 MIPS: add support for rdhwr cycle counter register
845 379504 remove TileGX/Linux port
846 379525 Support more x86 nop opcodes
847 379838 disAMode(x86): not an addr!
848 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
849 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
850 379895 clock_gettime does not execute POST syscall wrapper
851 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
852 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
853 380200 xtree generated callgrind files refer to files without directory name
854 380202 Assertion failure for cache line size (cls == 64) on aarch64.
855 380397 s390x: __GI_strcspn() replacement needed
856 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
858 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
859 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
860 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
864 Release 3.12.0 (20 October 2016)
865 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
867 3.12.0 is a feature release with many improvements and the usual
868 collection of bug fixes.
870 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
871 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
872 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
873 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
874 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
875 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
877 * ================== PLATFORM CHANGES =================
879 * POWER: Support for ISA 3.0 has been added
881 * mips: support for O32 FPXX ABI has been added.
882 * mips: improved recognition of different processors
883 * mips: determination of page size now done at run time
885 * amd64: Partial support for AMD FMA4 instructions.
887 * arm, arm64: Support for v8 crypto and CRC instructions.
889 * Improvements and robustification of the Solaris port.
891 * Preliminary support for MacOS 10.12 (Sierra) has been added.
893 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
894 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
895 where possible. Valgrind's support for 32-bit x86 has stagnated in
896 recent years and has fallen far behind that for 64-bit x86
897 instructions. By contrast 64-bit x86 is well supported, up to and
900 * ==================== TOOL CHANGES ====================
904 - Added meta mempool support for describing a custom allocator which:
905 - Auto-frees all chunks assuming that destroying a pool destroys all
907 - Uses itself to allocate other memory blocks
909 - New flag --ignore-range-below-sp to ignore memory accesses below
910 the stack pointer, if you really have to. The related flag
911 --workaround-gcc296-bugs=yes is now deprecated. Use
912 --ignore-range-below-sp=1024-1 as a replacement.
916 - Improved thread startup time significantly on non-Linux platforms.
920 - Added collection of the metric "tot-blocks-allocd"
922 * ==================== OTHER CHANGES ====================
924 * Replacement/wrapping of malloc/new related functions is now done not just
925 for system libraries by default, but for any globally defined malloc/new
926 related function (both in shared libraries and statically linked alternative
927 malloc implementations). The dynamic (runtime) linker is excluded, though.
928 To only intercept malloc/new related functions in
929 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
930 "nouserintercepts" can be any non-existing library name).
931 This new functionality is not implemented for MacOS X.
933 * The maximum number of callers in a suppression entry is now equal to
934 the maximum size for --num-callers (500).
935 Note that --gen-suppressions=yes|all similarly generates suppressions
936 containing up to --num-callers frames.
938 * New and modified GDB server monitor features:
940 - Valgrind's gdbserver now accepts the command 'catch syscall'.
941 Note that you must have GDB >= 7.11 to use 'catch syscall' with
944 * New option --run-cxx-freeres=<yes|no> can be used to change whether
945 __gnu_cxx::__freeres() cleanup function is called or not. Default is
948 * Valgrind is able to read compressed debuginfo sections in two formats:
949 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
950 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
952 * Modest JIT-cost improvements: the cost of instrumenting code blocks
953 for the most common use case (x86_64-linux, Memcheck) has been
956 * Improved performance for programs that do a lot of discarding of
957 instruction address ranges of 8KB or less.
959 * The C++ symbol demangler has been updated.
961 * More robustness against invalid syscall parameters on Linux.
963 * ==================== FIXED BUGS ====================
965 The following bugs have been fixed or resolved. Note that "n-i-bz"
966 stands for "not in bugzilla" -- that is, a bug that was reported to us
967 but never got a bugzilla entry. We encourage you to file bugs in
968 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
969 than mailing the developers (or mailing lists) directly -- bugs that
970 are not entered into bugzilla tend to get forgotten about or ignored.
972 To see details of a given bug, visit
973 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
974 where XXXXXX is the bug number as listed below.
976 191069 Exiting due to signal not reported in XML output
977 199468 Suppressions: stack size limited to 25
978 while --num-callers allows more frames
979 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
980 278744 cvtps2pd with redundant RexW
981 303877 valgrind doesn't support compressed debuginfo sections.
982 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
983 348345 Assertion fails for negative lineno
984 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
985 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
986 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
987 351804 Crash on generating suppressions for "printf" call on OS X 10.10
988 352197 mips: mmap2() not wrapped correctly for page size > 4096
989 353083 arm64 doesn't implement various xattr system calls
990 353084 arm64 doesn't support sigpending system call
991 353137 www: update info for Supported Platforms
992 353138 www: update "The Valgrind Developers" page
993 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
996 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
997 353398 WARNING: unhandled amd64-solaris syscall: 207
998 353660 XML in auxwhat tag not escaping reserved symbols properly
999 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
1000 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
1001 353802 ELF debug info reader confused with multiple .rodata sections
1002 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
1003 353917 unhandled amd64-solaris syscall fchdir(120)
1004 353920 unhandled amd64-solaris syscall: 170
1005 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
1006 354392 unhandled amd64-solaris syscall: 171
1007 354797 Vbit test does not include Iops for Power 8 instruction support
1008 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
1012 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
1013 355188 valgrind should intercept all malloc related global functions
1014 355454 do not intercept malloc related symbols from the runtime linker
1015 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
1016 356044 Dwarf line info reader misinterprets is_stmt register
1017 356112 mips: replace addi with addiu
1018 356393 valgrind (vex) crashes because isZeroU happened
1021 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
1022 356678 arm64-linux: unhandled syscall 232 (mincore)
1023 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
1024 356823 Unsupported ARM instruction: stlex
1025 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
1026 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
1027 357673 crash if I try to run valgrind with a binary link with libcurl
1028 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
1029 357871 pthread_spin_destroy not properly wrapped
1030 357887 Calls to VG_(fclose) do not close the file descriptor
1031 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
1032 358030 support direct socket calls on x86 32bit (new in linux 4.3)
1033 358478 drd/tests/std_thread.cpp doesn't build with GCC6
1034 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
1035 359181 Buffer Overflow during Demangling
1036 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
1037 359289 s390x: popcnt (B9E1) not implemented
1038 359472 The Power PC vsubuqm instruction doesn't always give the correct result
1039 359503 Add missing syscalls for aarch64 (arm64)
1040 359645 "You need libc6-dbg" help message could be more helpful
1041 359703 s390: wire up separate socketcalls system calls
1042 359724 getsockname might crash - deref_UInt should call safe_to_deref
1043 359733 amd64 implement ld.so strchr/index override like x86
1044 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
1045 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
1047 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
1048 359871 Incorrect mask handling in ppoll
1049 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
1050 360008 Contents of Power vr registers contents is not printed correctly when
1051 the --vgdb-shadow-registers=yes option is used
1052 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
1053 non-zero shadow bits
1054 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
1055 360425 arm64 unsupported instruction ldpsw
1057 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
1058 360571 Error about the Android Runtime reading below the stack pointer on ARM
1059 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
1060 360749 kludge for multiple .rodata sections on Solaris no longer needed
1061 360752 raise the number of reserved fds in m_main.c from 10 to 12
1062 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
1063 361226 s390x: risbgn (EC59) not implemented
1064 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
1065 361354 ppc64[le]: wire up separate socketcalls system calls
1066 361615 Inconsistent termination for multithreaded process terminated by signal
1067 361926 Unhandled Solaris syscall: sysfs(84)
1068 362009 V dumps core on unimplemented functionality before threads are created
1069 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
1070 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
1071 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
1072 362953 Request for an update to the Valgrind Developers page
1073 363680 add renameat2() support
1074 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1075 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1076 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1077 364058 clarify in manual limitations of array overruns detections
1078 364413 pselect sycallwrapper mishandles NULL sigmask
1079 364728 Power PC, missing support for several HW registers in
1080 get_otrack_shadow_offset_wrk()
1081 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1082 365273 Invalid write to stack location reported after signal handler runs
1083 365912 ppc64BE segfault during jm-insns test (RELRO)
1084 366079 FPXX Support for MIPS32 Valgrind
1085 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1086 366344 Multiple unhandled instruction for Aarch64
1087 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1088 367995 Integration of memcheck with custom memory allocator
1089 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1090 368412 False positive result for altivec capability check
1091 368416 Add tc06_two_races_xml.exp output for ppc64
1092 368419 Perf Events ioctls not implemented
1093 368461 mmapunmap test fails on ppc64
1094 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1095 369000 AMD64 fma4 instructions unsupported.
1096 369169 ppc64 fails jm_int_isa_2_07 test
1097 369175 jm_vec_isa_2_07 test crashes on ppc64
1098 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1099 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1100 369359 msghdr_foreachfield can crash when handling bad iovec
1101 369360 Bad sigprocmask old or new sets can crash valgrind
1102 369361 vmsplice syscall wrapper crashes on bad iovec
1103 369362 Bad sigaction arguments crash valgrind
1104 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1105 369402 Bad set/get_thread_area pointer crashes valgrind
1106 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1107 369446 valgrind crashes on unknown fcntl command
1108 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1109 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1110 370265 ISA 3.0 HW cap stuff needs updating
1111 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1112 372195 Power PC, xxsel instruction is not always recognized
1114 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1115 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1116 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1117 n-i-bz Fix ppoll_alarm exclusion on OS X
1118 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1119 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1120 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1121 n-i-bz amd64: memcheck false positive with shr %edx
1122 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1123 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1124 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1125 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1127 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1128 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1129 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1133 Release 3.11.0 (22 September 2015)
1134 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1136 3.11.0 is a feature release with many improvements and the usual
1137 collection of bug fixes.
1139 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1140 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1141 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1142 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1143 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1144 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1146 * ================== PLATFORM CHANGES =================
1148 * Support for Solaris/x86 and Solaris/amd64 has been added.
1150 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1152 * Preliminary support for the Tilera TileGX architecture has been added.
1154 * s390x: It is now required for the host to have the "long displacement"
1155 facility. The oldest supported machine model is z990.
1157 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1158 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1160 * The JIT's register allocator is significantly faster, making the JIT
1161 as a whole somewhat faster, so JIT-intensive activities, for example
1162 program startup, are modestly faster, around 5%.
1164 * There have been changes to the default settings of several command
1165 line flags, as detailed below.
1167 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1168 capable hosts, the simulated CPUID will now indicate AVX2 support.
1170 * ==================== TOOL CHANGES ====================
1174 - The default value for --leak-check-heuristics has been changed from
1175 "none" to "all". This helps to reduce the number of possibly
1176 lost blocks, in particular for C++ applications.
1178 - The default value for --keep-stacktraces has been changed from
1179 "malloc-then-free" to "malloc-and-free". This has a small cost in
1180 memory (one word per malloc-ed block) but allows Memcheck to show the
1181 3 stacktraces of a dangling reference: where the block was allocated,
1182 where it was freed, and where it is acccessed after being freed.
1184 - The default value for --partial-loads-ok has been changed from "no" to
1185 "yes", so as to avoid false positive errors resulting from some kinds
1186 of vectorised loops.
1188 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1189 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1190 than get_vbits when you need to associate byte data value with
1191 their corresponding validity bits.
1193 - The 'block_list' monitor command has been enhanced:
1194 o it can print a range of loss records
1195 o it now accepts an optional argument 'limited <max_blocks>'
1196 to control the number of blocks printed.
1197 o if a block has been found using a heuristic, then
1198 'block_list' now shows the heuristic after the block size.
1199 o the loss records/blocks to print can be limited to the blocks
1200 found via specified heuristics.
1202 - The C helper functions used to instrument loads on
1203 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1204 replaced by handwritten assembly sequences. This gives speedups
1205 in the region of 0% to 7% for those targets only.
1207 - A new command line option, --expensive-definedness-checks=yes|no,
1208 has been added. This is useful for avoiding occasional invalid
1209 uninitialised-value errors in optimised code. Watch out for
1210 runtime degradation, as this can be up to 25%. As always, though,
1211 the slowdown is highly application specific. The default setting
1216 - A new monitor command 'all_snapshots <filename>' dumps all
1217 snapshots taken so far.
1221 - Significant memory reduction and moderate speedups for
1222 --history-level=full for applications accessing a lot of memory
1223 with many different stacktraces.
1225 - The default value for --conflict-cache-size=N has been doubled to
1226 2000000. Users that were not using the default value should
1227 preferably also double the value they give.
1229 The default was changed due to the changes in the "full history"
1230 implementation. Doubling the value gives on average a slightly more
1231 complete history and uses similar memory (or significantly less memory
1232 in the worst case) than the previous implementation.
1234 - The Helgrind monitor command 'info locks' now accepts an optional
1235 argument 'lock_addr', which shows information about the lock at the
1238 - When using --history-level=full, the new Helgrind monitor command
1239 'accesshistory <addr> [<len>]' will show the recorded accesses for
1240 <len> (or 1) bytes at <addr>.
1242 * ==================== OTHER CHANGES ====================
1244 * The default value for the --smc-check option has been changed from
1245 "stack" to "all-non-file" on targets that provide automatic D-I
1246 cache coherence (x86, amd64 and s390x). The result is to provide,
1247 by default, transparent support for JIT generated and self-modifying
1248 code on all targets.
1250 * Mac OS X only: the default value for the --dsymutil option has been
1251 changed from "no" to "yes", since any serious usage on Mac OS X
1252 always required it to be "yes".
1254 * The command line options --db-attach and --db-command have been removed.
1255 They were deprecated in 3.10.0.
1257 * When a process dies due to a signal, Valgrind now shows the signal
1258 and the stacktrace at default verbosity (i.e. verbosity 1).
1260 * The address description logic used by Memcheck and Helgrind now
1261 describes addresses in anonymous segments, file mmap-ed segments,
1262 shared memory segments and the brk data segment.
1264 * The new option --error-markers=<begin>,<end> can be used to mark the
1265 begin/end of errors in textual output mode, to facilitate
1266 searching/extracting errors in output files that mix valgrind errors
1267 with program output.
1269 * The new option --max-threads=<number> can be used to change the number
1270 of threads valgrind can handle. The default is 500 threads which
1271 should be more than enough for most applications.
1273 * The new option --valgrind-stacksize=<number> can be used to change the
1274 size of the private thread stacks used by Valgrind. This is useful
1275 for reducing memory use or increasing the stack size if Valgrind
1276 segfaults due to stack overflow.
1278 * The new option --avg-transtab-entry-size=<number> can be used to specify
1279 the expected instrumented block size, either to reduce memory use or
1280 to avoid excessive retranslation.
1282 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1284 * New and modified GDB server monitor features:
1286 - When a signal is reported in GDB, you can now use the GDB convenience
1287 variable $_siginfo to examine detailed signal information.
1289 - Valgrind's gdbserver now allows the user to change the signal
1290 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1291 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1292 continue without passing the signal to the process.
1294 - With GDB >= 7.10, the command 'target remote'
1295 will automatically load the executable file of the process running
1296 under Valgrind. This means you do not need to specify the executable
1297 file yourself, GDB will discover it itself. See GDB documentation about
1298 'qXfer:exec-file:read' packet for more info.
1300 * ==================== FIXED BUGS ====================
1302 The following bugs have been fixed or resolved. Note that "n-i-bz"
1303 stands for "not in bugzilla" -- that is, a bug that was reported to us
1304 but never got a bugzilla entry. We encourage you to file bugs in
1305 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1306 than mailing the developers (or mailing lists) directly -- bugs that
1307 are not entered into bugzilla tend to get forgotten about or ignored.
1309 To see details of a given bug, visit
1310 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1311 where XXXXXX is the bug number as listed below.
1313 116002 VG_(printf): Problems with justification of strings and integers
1314 155125 avoid cutting away file:lineno after long function name
1315 197259 Unsupported arch_prtctl PR_SET_GS option
1316 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
1317 201216 Fix Valgrind does not support pthread_sigmask() on OS X
1318 201435 Fix Darwin: -v does not show kernel version
1319 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
1320 211256 Fixed an outdated comment regarding the default platform.
1321 211529 Incomplete call stacks for code compiled by newer versions of MSVC
1322 211926 Avoid compilation warnings in valgrind.h with -pedantic
1323 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
1325 226609 Crediting upstream authors in man page
1326 231257 Valgrind omits path when executing script from shebang line
1327 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
1328 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
1329 269360 s390x: Fix addressing mode selection for compare-and-swap
1330 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
1332 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
1333 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
1334 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
1335 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
1336 330147 libmpiwrap PMPI_Get_count returns undefined value
1337 333051 mmap of huge pages fails due to incorrect alignment
1339 334802 valgrind does not always explain why a given option is bad
1340 335618 mov.w rN, pc/sp (ARM32)
1341 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
1346 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
1347 338602 AVX2 bit in CPUID missing
1348 338606 Strange message for scripts with invalid interpreter
1349 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
1350 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
1351 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
1353 339156 gdbsrv not called for fatal signal
1354 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
1355 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
1356 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
1357 339442 Fix testsuite build failure on OS X 10.9
1358 339542 Enable compilation with Intel's ICC compiler
1359 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
1360 339688 Mac-specific ASM does not support .version directive (cpuid,
1361 tronical and pushfpopf tests)
1362 339745 Valgrind crash when check Marmalade app (partial fix)
1363 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
1364 339778 Linux/TileGx platform support for Valgrind
1365 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
1366 339789 Fix none/tests/execve test on Mac OS X 10.9
1367 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
1368 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
1369 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
1370 340392 Allow user to select more accurate definedness checking in memcheck
1371 to avoid invalid complaints on optimised code
1372 340430 Fix some grammatical weirdness in the manual.
1373 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
1374 341419 Signal handler ucontext_t not filled out correctly on OS X
1375 341539 VG_(describe_addr) should not describe address as belonging to client
1376 segment if it is past the heap end
1377 341613 Enable building of manythreads and thread-exits tests on Mac OS X
1378 341615 Fix none/tests/darwin/access_extended test on Mac OS X
1379 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
1380 341789 aarch64: shmat fails with valgrind on ARMv8
1381 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
1382 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
1383 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
1384 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
1385 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
1386 342221 socket connect false positive uninit memory for unknown af family
1387 342353 Allow dumping full massif output while valgrind is still running
1388 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
1392 342603 Add I2C_SMBUS ioctl support
1393 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
1394 342683 Mark memory past the initial brk limit as unaddressable
1395 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
1396 342795 Internal glibc __GI_mempcpy call should be intercepted
1397 342841 s390x: Support instructions fiebr(a) and fidbr(a)
1398 343012 Unhandled syscall 319 (memfd_create)
1399 343069 Patch updating v4l2 API support
1400 343173 helgrind crash during stack unwind
1401 343219 fix GET_STARTREGS for arm
1402 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
1403 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
1404 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
1405 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
1406 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
1407 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
1408 343597 ppc64le: incorrect use of offseof macro
1409 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
1410 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
1411 343732 Unhandled syscall 144 (setgid) on aarch64
1412 343733 Unhandled syscall 187 (msgctl and related) on aarch64
1413 343802 s390x: False positive "conditional jump or move depends on [..]
1414 343902 --vgdb=yes doesn't break when --xml=yes is used
1415 343967 Don't warn about setuid/setgid/setcap executable for directories
1416 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
1417 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
1418 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
1419 344054 www - update info for Solaris/illumos
1420 344416 'make regtest' does not work cleanly on OS X
1421 344235 Remove duplicate include of pub_core_aspacemgr.h
1422 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
1423 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
1424 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
1425 344314 callgrind_annotate ... warnings about commands containing newlines
1426 344318 socketcall should wrap recvmmsg and sendmmsg
1427 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
1428 344416 Fix 'make regtest' does not work cleanly on OS X
1429 344499 Fix compilation for Linux kernel >= 4.0.0
1430 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
1431 unix:349 (__pthread_fchdir)
1432 344559 Garbage collection of unused segment names in address space manager
1433 344560 Fix stack traces missing penultimate frame on OS X
1434 344621 Fix memcheck/tests/err_disable4 test on OS X
1435 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
1436 344702 Fix missing libobjc suppressions on OS X 10.10
1438 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
1439 344939 Fix memcheck/tests/xml1 on OS X 10.10
1440 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
1441 345079 Fix build problems in VEX/useful/test_main.c
1442 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
1443 345177 arm64: prfm (reg) not implemented
1444 345215 Performance improvements for the register allocator
1445 345248 add support for Solaris OS in valgrind
1446 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
1447 345394 Fix memcheck/tests/strchr on OS X
1448 345637 Fix memcheck/tests/sendmsg on OS X
1449 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
1450 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
1451 345887 Fix an assertion in the address space manager
1452 345928 amd64: callstack only contains current function for small stacks
1453 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1454 345987 MIPS64: Implement cavium LHX instruction
1455 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1456 346185 Fix typo saving altivec register v24
1457 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1458 and LibVEX_GuestPPC64_get_CR()
1459 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1460 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1461 346307 fuse filesystem syscall deadlocks
1462 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1463 346411 MIPS: SysRes::_valEx handling is incorrect
1464 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1465 346474 PPC64 Power 8, spr TEXASRU register not supported
1466 346487 Compiler generates "note" about a future ABI change for PPC64
1467 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1468 and causing spurious "invalid read of size 8" warnings
1469 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1470 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1471 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1472 347322 Power PC regression test cleanup
1473 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1475 347389 unhandled syscall: 373 (Linux ARM syncfs)
1476 347686 Patch set to cleanup PPC64 regtests
1477 347978 Remove bash dependencies where not needed
1478 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1479 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1481 348102 Patch updating v4l2 API support
1482 348247 amd64 front end: jno jumps wrongly when overflow is not set
1483 348269 Improve mmap MAP_HUGETLB support.
1484 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1485 348345 Assertion fails for negative lineno
1486 348377 Unsupported ARM instruction: yield
1487 348565 Fix detection of command line option availability for clang
1488 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1489 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1490 348748 Fix redundant condition
1491 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1492 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1493 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1494 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1495 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1496 349626 Implemented additional Xen hypercalls
1497 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1498 349790 Clean up of the hardware capability checking utilities.
1499 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1500 349874 Fix typos in source code
1501 349879 memcheck: add handwritten assembly for helperc_LOADV*
1502 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1503 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1504 350202 Add limited param to 'monitor block_list'
1505 350290 s390x: Support instructions fixbr(a)
1506 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1507 350809 Fix none/tests/async-sigs for Solaris
1508 350811 Remove reference to --db-attach which has been removed.
1509 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1510 350854 hard-to-understand code in VG_(load_ELF)()
1511 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1512 351386 Solaris: Cannot run ld.so.1 under Valgrind
1513 351474 Fix VG_(iseqsigset) as obvious
1514 351531 Typo in /include/vki/vki-xen-physdev.h header guard
1515 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
1516 351858 ldsoexec support on Solaris
1517 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
1518 352130 helgrind reports false races for printfs using mempcpy on FILE* state
1519 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
1520 352320 arm64 crash on none/tests/nestedfs
1521 352765 Vbit test fails on Power 6
1522 352768 The mbar instruction is missing from the Power PC support
1523 352769 Power PC program priority register (PPR) is not supported
1524 n-i-bz Provide implementations of certain compiler builtins to support
1525 compilers that may not provide those
1526 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
1527 n-i-bz Fix compilation on distros with glibc < 2.5
1528 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
1529 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
1530 n-i-bz mremap did not work properly on shared memory
1531 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
1532 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
1534 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
1535 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
1536 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
1540 Release 3.10.1 (25 November 2014)
1541 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1542 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
1543 and backports fixes for all reported missing AArch64 ARMv8 instructions
1544 and syscalls from the trunk. If you package or deliver 3.10.0 for others
1545 to use, you might want to consider upgrading to 3.10.1 instead.
1547 The following bugs have been fixed or resolved. Note that "n-i-bz"
1548 stands for "not in bugzilla" -- that is, a bug that was reported to us
1549 but never got a bugzilla entry. We encourage you to file bugs in
1550 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1551 than mailing the developers (or mailing lists) directly -- bugs that
1552 are not entered into bugzilla tend to get forgotten about or ignored.
1554 To see details of a given bug, visit
1555 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1556 where XXXXXX is the bug number as listed below.
1558 335440 arm64: ld1 (single structure) is not implemented
1559 335713 arm64: unhanded instruction: prfm (immediate)
1560 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
1561 339182 ppc64: AvSplat ought to load destination vector register with [..]
1562 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
1563 339433 ppc64 lxvw4x instruction uses four 32-byte loads
1564 339645 Use correct tag names in sys_getdents/64 wrappers
1565 339706 Fix false positive for ioctl(TIOCSIG) on linux
1566 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
1567 339853 arm64 times syscall unknown
1568 339855 arm64 unhandled getsid/setsid syscalls
1569 339858 arm64 dmb sy not implemented
1570 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
1571 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
1572 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
1574 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
1575 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
1576 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
1577 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
1578 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
1579 340509 arm64: unhandled instruction fcvtas
1580 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
1581 340632 arm64: unhandled instruction fcvtas
1582 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
1583 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
1584 340788 warning: unhandled syscall: 318 (getrandom)
1585 340807 disInstr(arm): unhandled instruction: 0xEE989B20
1586 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
1587 340922 arm64: unhandled getgroups/setgroups syscalls
1588 350251 Fix typo in VEX utility program (test_main.c).
1589 350407 arm64: unhandled instruction ucvtf (vector, integer)
1590 350809 none/tests/async-sigs breaks when run under cron on Solaris
1591 350811 update README.solaris after r15445
1592 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
1593 350854 strange code in VG_(load_ELF)()
1594 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1595 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
1596 n-i-bz Add missing ]] to terminate CDATA.
1597 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
1598 n-i-bz Enable sys_fadvise64_64 on arm32.
1599 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
1600 n-i-bz Add test cases for all known arm64 load/store instructions.
1601 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
1602 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
1603 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
1604 n-i-bz arm64: Implement "fcvtpu w, s".
1605 n-i-bz arm64: implement ADDP and various others
1606 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
1607 n-i-bz arm64: enable FCVT{A,N}S X,S.
1609 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
1613 Release 3.10.0 (10 September 2014)
1614 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1616 3.10.0 is a feature release with many improvements and the usual
1617 collection of bug fixes.
1619 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1620 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1621 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
1622 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
1623 significantly improved relative to the 3.9.0 release.
1625 * ================== PLATFORM CHANGES =================
1627 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
1628 is mostly complete, and is usable, but some SIMD instructions are as
1631 * Support for little-endian variant of the 64-bit POWER architecture.
1633 * Support for Android on MIPS32.
1635 * Support for 64bit FPU on MIPS32 platforms.
1637 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
1639 * Configuration for and running on Android targets has changed.
1640 See README.android in the source tree for details.
1642 * ================== DEPRECATED FEATURES =================
1644 * --db-attach is now deprecated and will be removed in the next
1645 valgrind feature release. The built-in GDB server capabilities are
1646 superior and should be used instead. Learn more here:
1647 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
1649 * ==================== TOOL CHANGES ====================
1653 - Client code can now selectively disable and re-enable reporting of
1654 invalid address errors in specific ranges using the new client
1655 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
1656 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
1658 - Leak checker: there is a new leak check heuristic called
1659 "length64". This is used to detect interior pointers pointing 8
1660 bytes inside a block, on the assumption that the first 8 bytes
1661 holds the value "block size - 8". This is used by
1662 sqlite3MemMalloc, for example.
1664 - Checking of system call parameters: if a syscall parameter
1665 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
1666 several fields not initialised, an error is now reported for each
1667 field. Previously, an error was reported only for the first
1668 uninitialised field.
1670 - Mismatched alloc/free checking: a new flag
1671 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
1672 such checks if necessary.
1676 - Improvements to error messages:
1678 o Race condition error message involving heap allocated blocks also
1679 show the thread number that allocated the raced-on block.
1681 o All locks referenced by an error message are now announced.
1682 Previously, some error messages only showed the lock addresses.
1684 o The message indicating where a lock was first observed now also
1685 describes the address/location of the lock.
1687 - Helgrind now understands the Ada task termination rules and
1688 creates a happens-before relationship between a terminated task
1689 and its master. This avoids some false positives and avoids a big
1690 memory leak when a lot of Ada tasks are created and terminated.
1691 The interceptions are only activated with forthcoming releases of
1692 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
1694 - A new GDB server monitor command "info locks" giving the list of
1695 locks, their location, and their status.
1699 - callgrind_control now supports the --vgdb-prefix argument,
1700 which is needed if valgrind was started with this same argument.
1702 * ==================== OTHER CHANGES ====================
1704 * Unwinding through inlined function calls. Stack unwinding can now
1705 make use of Dwarf3 inlined-unwind information if it is available.
1706 The practical effect is that inlined calls become visible in stack
1707 traces. The suppression matching machinery has been adjusted
1708 accordingly. This is controlled by the new option
1709 --read-inline-info=yes|no. Currently this is enabled by default
1710 only on Linux and Android targets and only for the tools Memcheck,
1713 * Valgrind can now read EXIDX unwind information on 32-bit ARM
1714 targets. If an object contains both CFI and EXIDX unwind
1715 information, Valgrind will prefer the CFI over the EXIDX. This
1716 facilitates unwinding through system libraries on arm-android
1719 * Address description logic has been improved and is now common
1720 between Memcheck and Helgrind, resulting in better address
1721 descriptions for some kinds of error messages.
1723 * Error messages about dubious arguments (eg, to malloc or calloc) are
1724 output like other errors. This means that they can be suppressed
1725 and they have a stack trace.
1727 * The C++ demangler has been updated for better C++11 support.
1729 * New and modified GDB server monitor features:
1731 - Thread local variables/storage (__thread) can now be displayed.
1733 - The GDB server monitor command "v.info location <address>"
1734 displays information about an address. The information produced
1735 depends on the tool and on the options given to valgrind.
1736 Possibly, the following are described: global variables, local
1737 (stack) variables, allocated or freed blocks, ...
1739 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
1740 ask the GDB server to stop at the start of program execution, at
1741 the end of the program execution and on Valgrind internal errors.
1743 - A new monitor command "v.info stats" shows various Valgrind core
1744 and tool statistics.
1746 - A new monitor command "v.set hostvisibility" allows the GDB server
1747 to provide access to Valgrind internal host status/memory.
1749 * A new option "--aspace-minaddr=<address>" can in some situations
1750 allow the use of more memory by decreasing the address above which
1751 Valgrind maps memory. It can also be used to solve address
1752 conflicts with system libraries by increasing the default value.
1753 See user manual for details.
1755 * The amount of memory used by Valgrind to store debug info (unwind
1756 info, line number information and symbol data) has been
1757 significantly reduced, even though Valgrind now reads more
1758 information in order to support unwinding of inlined function calls.
1760 * Dwarf3 handling with --read-var-info=yes has been improved:
1762 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
1764 - Code compiled with
1765 -ffunction-sections -fdata-sections -Wl,--gc-sections
1766 no longer causes assertion failures.
1768 * Improved checking for the --sim-hints= and --kernel-variant=
1769 options. Unknown strings are now detected and reported to the user
1772 * The semantics of stack start/end boundaries in the valgrind.h
1773 VALGRIND_STACK_REGISTER client request has been clarified and
1774 documented. The convention is that start and end are respectively
1775 the lowest and highest addressable bytes of the stack.
1777 * ==================== FIXED BUGS ====================
1779 The following bugs have been fixed or resolved. Note that "n-i-bz"
1780 stands for "not in bugzilla" -- that is, a bug that was reported to us
1781 but never got a bugzilla entry. We encourage you to file bugs in
1782 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1783 than mailing the developers (or mailing lists) directly -- bugs that
1784 are not entered into bugzilla tend to get forgotten about or ignored.
1786 To see details of a given bug, visit
1787 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1788 where XXXXXX is the bug number as listed below.
1790 175819 Support for ipv6 socket reporting with --track-fds
1791 232510 make distcheck fails
1792 249435 Analyzing wine programs with callgrind triggers a crash
1793 278972 support for inlined function calls in stacktraces and suppression
1795 291310 FXSAVE instruction marks memory as undefined on amd64
1796 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
1797 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
1798 315199 vgcore file for threaded app does not show which thread crashed
1799 315952 tun/tap ioctls are not supported
1800 323178 Unhandled instruction: PLDW register (ARM)
1801 323179 Unhandled instruction: PLDW immediate (ARM)
1802 324050 Helgrind: SEGV because of unaligned stack when using movdqa
1803 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
1804 325124 [MIPSEL] Compilation error
1805 325477 Phase 4 support for IBM Power ISA 2.07
1806 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
1807 325628 Phase 5 support for IBM Power ISA 2.07
1808 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
1809 325751 Missing the two privileged Power PC Transactional Memory Instructions
1810 325816 Phase 6 support for IBM Power ISA 2.07
1811 325856 Make SGCheck fail gracefully on unsupported platforms
1812 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
1813 326436 DRD: False positive in libstdc++ std::list::push_back
1814 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
1815 326462 Refactor vgdb to isolate invoker stuff into separate module
1816 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
1817 326623 DRD: false positive conflict report in a field assignment
1818 326724 Valgrind does not compile on OSX 1.9 Mavericks
1819 326816 Intercept for __strncpy_sse2_unaligned missing?
1820 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
1821 326983 Clear direction flag after tests on amd64.
1822 327212 Do not prepend the current directory to absolute path names.
1823 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
1824 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
1825 327284 s390x: Fix translation of the risbg instruction
1826 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
1827 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
1828 327916 DW_TAG_typedef may have no name
1829 327943 s390x: add a redirection for the 'index' function
1830 328100 XABORT not implemented
1831 328205 Implement additional Xen hypercalls
1832 328454 add support Backtraces with ARM unwind tables (EXIDX)
1833 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
1834 328711 valgrind.1 manpage "memcheck options" section is badly generated
1835 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
1836 329612 Incorrect handling of AT_BASE for image execution
1837 329694 clang warns about using uninitialized variable
1838 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
1839 330228 mmap must align to VKI_SHMLBA on mips32
1840 330257 LLVM does not support `-mno-dynamic-no-pic` option
1841 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
1842 330459 --track-fds=yes doesn't track eventfds
1843 330469 Add clock_adjtime syscall support
1844 330594 Missing sysalls on PowerPC / uClibc
1845 330622 Add test to regression suite for POWER instruction: dcbzl
1846 330939 Support for AMD's syscall instruction on x86
1848 330941 Typo in PRE(poll) syscall wrapper
1849 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
1850 331254 Fix expected output for memcheck/tests/dw4
1851 331255 Fix race condition in test none/tests/coolo_sigaction
1852 331257 Fix type of jump buffer in test none/tests/faultstatus
1853 331305 configure uses bash specific syntax
1854 331337 s390x WARNING: unhandled syscall: 326 (dup3)
1855 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
1856 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
1857 331829 Unexpected ioctl opcode sign extension
1858 331830 ppc64: WARNING: unhandled syscall: 96/97
1859 331839 drd/tests/sem_open specifies invalid semaphore name
1860 331847 outcome of drd/tests/thread_name is nondeterministic
1861 332037 Valgrind cannot handle Thumb "add pc, reg"
1862 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
1863 consistency checks enabled
1864 332263 intercepts for pthread_rwlock_timedrdlock and
1865 pthread_rwlock_timedwrlock are incorrect
1866 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
1868 332276 Implement additional Xen hypercalls
1869 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
1870 332765 Fix ms_print to create temporary files in a proper directory
1871 333072 drd: Add semaphore annotations
1872 333145 Tests for missaligned PC+#imm access for arm
1873 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
1874 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
1875 333248 WARNING: unhandled syscall: unix:443
1876 333428 ldr.w pc [rD, #imm] instruction leads to assertion
1877 333501 cachegrind: assertion: Cache set count is not a power of two.
1880 333666 Recognize MPX instructions and bnd prefix.
1881 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
1882 333817 Valgrind reports the memory areas written to by the SG_IO
1884 334049 lzcnt fails silently (x86_32)
1885 334384 Valgrind does not have support Little Endian support for
1887 334585 recvmmsg unhandled (+patch) (arm)
1888 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
1889 334727 Build fails with -Werror=format-security
1890 334788 clarify doc about --log-file initial program directory
1891 334834 PPC64 Little Endian support, patch 2
1892 334836 PPC64 Little Endian support, patch 3 testcase fixes
1893 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
1894 335034 Unhandled ioctl: HCIGETDEVLIST
1895 335155 vgdb, fix error print statement.
1896 335262 arm64: movi 8bit version is not supported
1897 335263 arm64: dmb instruction is not implemented
1898 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
1899 335496 arm64: sbc/abc instructions are not implemented
1900 335554 arm64: unhandled instruction: abs
1901 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
1902 335735 arm64: unhandled instruction: cnt
1903 335736 arm64: unhandled instruction: uaddlv
1904 335848 arm64: unhandled instruction: {s,u}cvtf
1905 335902 arm64: unhandled instruction: sli
1906 335903 arm64: unhandled instruction: umull (vector)
1907 336055 arm64: unhandled instruction: mov (element)
1908 336062 arm64: unhandled instruction: shrn{,2}
1909 336139 mip64: [...] valgrind hangs and spins on a single core [...]
1910 336189 arm64: unhandled Instruction: mvn
1911 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
1912 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
1913 336772 Make moans about unknown ioctls more informative
1914 336957 Add a section about the Solaris/illumos port on the webpage
1915 337094 ifunc wrapper is broken on ppc64
1916 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
1917 337528 leak check heuristic for block prefixed by length as 64bit number
1918 337740 Implement additional Xen hypercalls
1919 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
1920 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
1921 337871 deprecate --db-attach
1922 338023 Add support for all V4L2/media ioctls
1923 338024 inlined functions are not shown if DW_AT_ranges is used
1924 338106 Add support for 'kcmp' syscall
1925 338115 DRD: computed conflict set differs from actual after fork
1926 338160 implement display of thread local storage in gdbsrv
1927 338205 configure.ac and check for -Wno-tautological-compare
1928 338300 coredumps are missing one byte of every segment
1929 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
1930 338499 --sim-hints parsing broken due to wrong order in tokens
1931 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
1932 338681 Unable to unwind through clone thread created on i386-linux
1933 338698 race condition between gdbsrv and vgdb on startup
1934 338703 helgrind on arm-linux gets false positives in dynamic loader
1935 338791 alt dwz files can be relative of debug/main file
1936 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
1937 338932 build V-trunk with gcc-trunk
1938 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
1939 345079 Fix build problems in VEX/useful/test_main.c
1940 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
1941 n-i-bz s390x: Fix memory corruption for multithreaded applications
1942 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
1943 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
1944 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
1945 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
1946 n-i-bz Add command-line option --default-suppressions
1947 n-i-bz Add support for BLKDISCARDZEROES ioctl
1948 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
1949 n-i-bz Add support for sys_pivot_root and sys_unshare
1951 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
1952 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
1953 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
1957 Release 3.9.0 (31 October 2013)
1958 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1959 3.9.0 is a feature release with many improvements and the usual
1960 collection of bug fixes.
1962 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
1963 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
1964 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
1965 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
1967 * ================== PLATFORM CHANGES =================
1969 * Support for MIPS64 LE and BE running Linux. Valgrind has been
1970 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
1972 * Support for MIPS DSP ASE on MIPS32 platforms.
1974 * Support for s390x Decimal Floating Point instructions on hosts that
1975 have the DFP facility installed.
1977 * Support for POWER8 (Power ISA 2.07) instructions
1979 * Support for Intel AVX2 instructions. This is available only on 64
1982 * Initial support for Intel Transactional Synchronization Extensions,
1985 * Initial support for Hardware Transactional Memory on POWER.
1987 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
1988 run large GUI apps tolerably well.
1990 * ==================== TOOL CHANGES ====================
1994 - Improvements in handling of vectorised code, leading to
1995 significantly fewer false error reports. You need to use the flag
1996 --partial-loads-ok=yes to get the benefits of these changes.
1998 - Better control over the leak checker. It is now possible to
1999 specify which leak kinds (definite/indirect/possible/reachable)
2000 should be displayed, which should be regarded as errors, and which
2001 should be suppressed by a given leak suppression. This is done
2002 using the options --show-leak-kinds=kind1,kind2,..,
2003 --errors-for-leak-kinds=kind1,kind2,.. and an optional
2004 "match-leak-kinds:" line in suppression entries, respectively.
2006 Note that generated leak suppressions contain this new line and
2007 are therefore more specific than in previous releases. To get the
2008 same behaviour as previous releases, remove the "match-leak-kinds:"
2009 line from generated suppressions before using them.
2011 - Reduced "possible leak" reports from the leak checker by the use
2012 of better heuristics. The available heuristics provide detection
2013 of valid interior pointers to std::stdstring, to new[] allocated
2014 arrays with elements having destructors and to interior pointers
2015 pointing to an inner part of a C++ object using multiple
2016 inheritance. They can be selected individually using the
2017 option --leak-check-heuristics=heur1,heur2,...
2019 - Better control of stacktrace acquisition for heap-allocated
2020 blocks. Using the --keep-stacktraces option, it is possible to
2021 control independently whether a stack trace is acquired for each
2022 allocation and deallocation. This can be used to create better
2023 "use after free" errors or to decrease Valgrind's resource
2024 consumption by recording less information.
2026 - Better reporting of leak suppression usage. The list of used
2027 suppressions (shown when the -v option is given) now shows, for
2028 each leak suppressions, how many blocks and bytes it suppressed
2029 during the last leak search.
2033 - False errors resulting from the use of statically initialised
2034 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
2037 - False errors resulting from the use of pthread_cond_waits that
2038 timeout, have been removed.
2040 * ==================== OTHER CHANGES ====================
2042 * Some attempt to tune Valgrind's space requirements to the expected
2043 capabilities of the target:
2045 - The default size of the translation cache has been reduced from 8
2046 sectors to 6 on Android platforms, since each sector occupies
2047 about 40MB when using Memcheck.
2049 - The default size of the translation cache has been increased to 16
2050 sectors on all other platforms, reflecting the fact that large
2051 applications require instrumentation and storage of huge amounts
2052 of code. For similar reasons, the number of memory mapped
2053 segments that can be tracked has been increased by a factor of 6.
2055 - In all cases, the maximum number of sectors in the translation
2056 cache can be controlled by the new flag --num-transtab-sectors.
2058 * Changes in how debug info (line numbers, etc) is read:
2060 - Valgrind no longer temporarily mmaps the entire object to read
2061 from it. Instead, reading is done through a small fixed sized
2062 buffer. This avoids virtual memory usage spikes when Valgrind
2063 reads debuginfo from large shared objects.
2065 - A new experimental remote debug info server. Valgrind can read
2066 debug info from a different machine (typically, a build host)
2067 where debuginfo objects are stored. This can save a lot of time
2068 and hassle when running Valgrind on resource-constrained targets
2069 (phones, tablets) when the full debuginfo objects are stored
2070 somewhere else. This is enabled by the --debuginfo-server=
2073 - Consistency checking between main and debug objects can be
2074 disabled using the --allow-mismatched-debuginfo option.
2076 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2077 scanning can recover stack traces in some cases when the normal
2078 unwind mechanisms fail. Stack scanning is best described as "a
2079 nasty, dangerous and misleading hack" and so is disabled by default.
2080 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2083 * Detection and merging of recursive stack frame cycles. When your
2084 program has recursive algorithms, this limits the memory used by
2085 Valgrind for recorded stack traces and avoids recording
2086 uninteresting repeated calls. This is controlled by the command
2087 line option --merge-recursive-frame and by the monitor command
2088 "v.set merge-recursive-frames".
2090 * File name and line numbers for used suppressions. The list of used
2091 suppressions (shown when the -v option is given) now shows, for each
2092 used suppression, the file name and line number where the suppression
2095 * New and modified GDB server monitor features:
2097 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2098 that can be used to execute gdbserver monitor commands from the
2101 - A new monitor command, "v.info open_fds", that gives the list of
2102 open file descriptors and additional details.
2104 - An optional message in the "v.info n_errs_found" monitor command,
2105 for example "v.info n_errs_found test 1234 finished", allowing a
2106 comment string to be added to the process output, perhaps for the
2107 purpose of separating errors of different tests or test phases.
2109 - A new monitor command "v.info execontext" that shows information
2110 about the stack traces recorded by Valgrind.
2112 - A new monitor command "v.do expensive_sanity_check_general" to run
2113 some internal consistency checks.
2115 * New flag --sigill-diagnostics to control whether a diagnostic
2116 message is printed when the JIT encounters an instruction it can't
2117 translate. The actual behavior -- delivery of SIGILL to the
2118 application -- is unchanged.
2120 * The maximum amount of memory that Valgrind can use on 64 bit targets
2121 has been increased from 32GB to 64GB. This should make it possible
2122 to run applications on Memcheck that natively require up to about 35GB.
2124 * ==================== FIXED BUGS ====================
2126 The following bugs have been fixed or resolved. Note that "n-i-bz"
2127 stands for "not in bugzilla" -- that is, a bug that was reported to us
2128 but never got a bugzilla entry. We encourage you to file bugs in
2129 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2130 than mailing the developers (or mailing lists) directly -- bugs that
2131 are not entered into bugzilla tend to get forgotten about or ignored.
2133 To see details of a given bug, visit
2134 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2135 where XXXXXX is the bug number as listed below.
2137 123837 system call: 4th argument is optional, depending on cmd
2138 135425 memcheck should tell you where Freed blocks were Mallocd
2139 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2140 207815 Adds some of the drm ioctls to syswrap-linux.c
2141 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2142 252955 Impossible to compile with ccache
2143 253519 Memcheck reports auxv pointer accesses as invalid reads.
2144 263034 Crash when loading some PPC64 binaries
2145 269599 Increase deepest backtrace
2146 274695 s390x: Support "compare to/from logical" instructions (z196)
2147 275800 s390x: Autodetect cache info (part 2)
2148 280271 Valgrind reports possible memory leaks on still-reachable std::string
2149 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2150 289578 Backtraces with ARM unwind tables (stack scan flags)
2151 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2152 304832 ppc32: build failure
2153 305431 Use find_buildid shdr fallback for separate .debug files
2154 305728 Add support for AVX2 instructions
2155 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2156 306035 s390x: Fix IR generation for LAAG and friends
2157 306054 s390x: Condition code computation for convert-to-int/logical
2158 306098 s390x: alternate opcode form for convert to/from fixed
2159 306587 Fix cache line detection from auxiliary vector for PPC.
2160 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2161 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2162 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2163 307101 sys_capget second argument can be NULL
2164 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2165 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2166 307113 s390x: DFP support
2167 307141 valgrind does't work in mips-linux system
2168 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2169 307285 x86_amd64 feature test for avx in test suite is wrong
2170 307290 memcheck overlap testcase needs memcpy version filter
2171 307463 Please add "&limit=0" to the "all open bugs" link
2172 307465 --show-possibly-lost=no should reduce the error count / exit code
2173 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2174 307729 pkgconfig support broken valgrind.pc
2175 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2176 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2177 308089 Unhandled syscall on ppc64: prctl
2178 308135 PPC32 MPC8xx has 16 bytes cache size
2179 308321 testsuite memcheck filter interferes with gdb_filter
2181 308341 vgdb should report process exit (or fatal signal)
2182 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2183 308495 Remove build dependency on installed Xen headers
2184 308573 Internal error on 64-bit instruction executed in 32-bit mode
2186 308627 pmovmskb validity bit propagation is imprecise
2187 308644 vgdb command for having the info for the track-fds option
2188 308711 give more info about aspacemgr and arenas in out_of_memory
2189 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2190 308718 ARM implement SMLALBB family of instructions
2191 308886 Missing support for PTRACE_SET/GETREGSET
2192 308930 syscall name_to_handle_at (303 on amd64) not handled
2193 309229 V-bit tester does not report number of tests generated
2194 309323 print unrecognized instuction on MIPS
2195 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2196 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2197 309430 Self hosting ppc64 encounters a vassert error on operand type
2198 309600 valgrind is a bit confused about 0-sized sections
2199 309823 Generate errors for still reachable blocks
2200 309921 PCMPISTRI validity bit propagation is imprecise
2201 309922 none/tests/ppc64/test_dfp5 sometimes fails
2202 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2203 310424 --read-var-info does not properly describe static variables
2204 310792 search additional path for debug symbols
2205 310931 s390x: Message-security assist (MSA) instruction extension [..]
2206 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2207 311318 ARM: "128-bit constant is not implemented" error message
2208 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2209 311690 V crashes because it redirects branches inside of a redirected function
2210 311880 x86_64: make regtest hangs at shell_valid1
2211 311922 WARNING: unhandled syscall: 170
2213 312171 ppc: insn selection for DFP
2214 312571 Rounding mode call wrong for the DFP Iops [..]
2215 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2216 312913 Dangling pointers error should also report the alloc stack trace
2217 312980 Building on Mountain Lion generates some compiler warnings
2218 313267 Adding MIPS64/Linux port to Valgrind
2221 313811 Buffer overflow in assert_fail
2222 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2223 314269 ppc: dead code in insn selection
2224 314718 ARM: implement integer divide instruction (sdiv and udiv)
2225 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2226 315441 sendmsg syscall should ignore unset msghdr msg_flags
2227 315534 msgrcv inside a thread causes valgrind to hang (block)
2228 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2229 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2230 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2231 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2232 316144 valgrind.1 manpage contains unknown ??? strings [..]
2233 316145 callgrind command line options in manpage reference (unknown) [..]
2234 316145 callgrind command line options in manpage reference [..]
2235 316181 drd: Fixed a 4x slowdown for certain applications
2236 316503 Valgrind does not support SSE4 "movntdqa" instruction
2237 316535 Use of |signed int| instead of |size_t| in valgrind messages
2238 316696 fluidanimate program of parsec 2.1 stuck
2239 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2240 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2241 317186 "Impossible happens" when occurs VCVT instruction on ARM
2242 317318 Support for Threading Building Blocks "scalable_malloc"
2243 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2244 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2245 317463 bmi testcase IR SANITY CHECK FAILURE
2246 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2247 318050 libmpiwrap fails to compile with out-of-source build
2248 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2249 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2250 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2251 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2252 318932 Add missing PPC64 and PPC32 system call support
2253 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2254 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2255 319494 VEX Makefile-gcc standalone build update after r2702
2256 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2257 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2258 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2259 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2260 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2261 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2262 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2263 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2264 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2265 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2266 320895 add fanotify support (patch included)
2267 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2268 321065 Valgrind updates for Xen 4.3
2269 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2270 321363 Unhandled instruction: SSAX (ARM + Thumb)
2271 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2272 321466 Unhandled instruction: SHASX (ARM + Thumb)
2273 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2274 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2275 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2276 321620 Unhandled instruction: UASX (ARM + Thumb)
2277 321621 Unhandled instruction: USAX (ARM + Thumb)
2278 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2279 321693 Unhandled instruction: LDRSBT (Thumb)
2280 321694 Unhandled instruction: UQASX (ARM + Thumb)
2281 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2282 321697 Unhandled instruction: UHASX (ARM + Thumb)
2283 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2284 321704 Unhandled instruction: REVSH (ARM + Thumb)
2285 321730 Add cg_diff and cg_merge man pages
2286 321738 Add vgdb and valgrind-listener man pages
2288 321891 Unhandled instruction: LDRHT (Thumb)
2289 321960 pthread_create() then alloca() causing invalid stack write errors
2290 321969 ppc32 and ppc64 don't support [lf]setxattr
2291 322254 Show threadname together with tid if set by application
2292 322294 Add initial support for IBM Power ISA 2.07
2293 322368 Assertion failure in wqthread_hijack under OS X 10.8
2294 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2295 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2296 322851 0bXXX binary literal syntax is not standard
2297 323035 Unhandled instruction: LDRSHT(Thumb)
2298 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2299 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2300 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2301 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2302 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2303 323437 Phase 2 support for IBM Power ISA 2.07
2304 323713 Support mmxext (integer sse) subset on i386 (athlon)
2305 323803 Transactional memory instructions are not supported for Power
2306 323893 SSE3 not available on amd cpus in valgrind
2307 323905 Probable false positive from Valgrind/drd on close()
2308 323912 valgrind.h header isn't compatible for mingw64
2309 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
2310 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
2311 324181 mmap does not handle MAP_32BIT
2312 324227 memcheck false positive leak when a thread calls exit+block [..]
2313 324421 Support for fanotify API on ARM architecture
2314 324514 gdbserver monitor cmd output behaviour consistency [..]
2315 324518 ppc64: Emulation of dcbt instructions does not handle [..]
2316 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
2317 324582 When access is made to freed memory, report both allocation [..]
2318 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
2319 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
2320 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
2321 324834 Unhandled instructions in Microsoft C run-time for x86_64
2322 324894 Phase 3 support for IBM Power ISA 2.07
2323 326091 drd: Avoid false race reports from optimized strlen() impls
2324 326113 valgrind libvex hwcaps error on AMD64
2325 n-i-bz Some wrong command line options could be ignored
2326 n-i-bz patch to allow fair-sched on android
2327 n-i-bz report error for vgdb snapshot requested before execution
2328 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
2330 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
2334 Release 3.8.1 (19 September 2012)
2335 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2336 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
2337 that occur moderately frequently in real use cases, adds support for
2338 some missing instructions on ARM, and fixes a deadlock condition on
2339 MacOSX. If you package or deliver 3.8.0 for others to use, you might
2340 want to consider upgrading to 3.8.1 instead.
2342 The following bugs have been fixed or resolved. Note that "n-i-bz"
2343 stands for "not in bugzilla" -- that is, a bug that was reported to us
2344 but never got a bugzilla entry. We encourage you to file bugs in
2345 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2346 than mailing the developers (or mailing lists) directly -- bugs that
2347 are not entered into bugzilla tend to get forgotten about or ignored.
2349 To see details of a given bug, visit
2350 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2351 where XXXXXX is the bug number as listed below.
2354 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
2355 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
2356 298281 wcslen causes false(?) uninitialised value warnings
2357 301281 valgrind hangs on OS X when the process calls system()
2358 304035 disInstr(arm): unhandled instruction 0xE1023053
2359 304867 implement MOVBE instruction in x86 mode
2360 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
2361 305042 amd64: implement 0F 7F encoding of movq between two registers
2362 305199 ARM: implement QDADD and QDSUB
2363 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
2364 305513 killed by fatal signal: SIGSEGV
2365 305690 DRD reporting invalid semaphore when sem_trywait fails
2366 305926 Invalid alignment checks for some AVX instructions
2367 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
2368 306310 3.8.0 release tarball missing some files
2369 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
2370 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
2371 n-i-bz shmat of a segment > 4Gb does not work
2372 n-i-bz simulate_control_c script wrong USR1 signal number on mips
2373 n-i-bz vgdb ptrace calls wrong on mips [...]
2374 n-i-bz Fixes for more MPI false positives
2375 n-i-bz exp-sgcheck's memcpy causes programs to segfault
2376 n-i-bz OSX build w/ clang: asserts at startup
2377 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
2378 n-i-bz fix a couple of union tag-vs-field mixups
2379 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
2381 The following bugs were fixed in 3.8.0 but not listed in this NEWS
2384 254088 Valgrind should know about UD2 instruction
2387 304754 NEWS blows TeX's little mind
2389 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
2393 Release 3.8.0 (10 August 2012)
2394 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2395 3.8.0 is a feature release with many improvements and the usual
2396 collection of bug fixes.
2398 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2399 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
2400 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
2401 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
2402 There is initial support for MacOSX 10.8, but it is not usable for
2403 serious work at present.
2405 * ================== PLATFORM CHANGES =================
2407 * Support for MIPS32 platforms running Linux. Valgrind has been
2408 tested on MIPS32 and MIPS32r2 platforms running different Debian
2409 Squeeze and MeeGo distributions. Both little-endian and big-endian
2410 cores are supported. The tools Memcheck, Massif and Lackey have
2411 been tested and are known to work. See README.mips for more details.
2413 * Preliminary support for Android running on x86.
2415 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
2417 * Support for Intel AVX instructions and for AES instructions. This
2418 support is available only for 64 bit code.
2420 * Support for POWER Decimal Floating Point instructions.
2422 * ==================== TOOL CHANGES ====================
2424 * Non-libc malloc implementations are now supported. This is useful
2425 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
2426 Using the new option --soname-synonyms, such tools can be informed
2427 that the malloc implementation is either linked statically into the
2428 executable, or is present in some other shared library different
2429 from libc.so. This makes it possible to process statically linked
2430 programs, and programs using other malloc libraries, for example
2431 TCMalloc or JEMalloc.
2433 * For tools that provide their own replacement for malloc et al, the
2434 option --redzone-size=<number> allows users to specify the size of
2435 the padding blocks (redzones) added before and after each client
2436 allocated block. Smaller redzones decrease the memory needed by
2437 Valgrind. Bigger redzones increase the chance to detect blocks
2438 overrun or underrun. Prior to this change, the redzone size was
2439 hardwired to 16 bytes in Memcheck.
2443 - The leak_check GDB server monitor command now can
2444 control the maximum nr of loss records to output.
2446 - Reduction of memory use for applications allocating
2447 many blocks and/or having many partially defined bytes.
2449 - Addition of GDB server monitor command 'block_list' that lists
2450 the addresses/sizes of the blocks of a leak search loss record.
2452 - Addition of GDB server monitor command 'who_points_at' that lists
2453 the locations pointing at a block.
2455 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2456 detect an invalid access of these redzones, by marking them
2457 noaccess. Similarly, if a redzone size is given for a memory
2458 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2459 This still allows to find some bugs if the user has forgotten to
2460 mark the pool superblock noaccess.
2462 - Performance of memory leak check has been improved, especially in
2463 cases where there are many leaked blocks and/or many suppression
2464 rules used to suppress leak reports.
2466 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2467 more precise analysis, which is important for LLVM/Clang
2468 generated code. This is at the cost of somewhat reduced
2469 performance. Note there is no change to analysis precision or
2470 costs on Linux targets.
2474 - Added even more facilities that can help finding the cause of a data
2475 race, namely the command-line option --ptrace-addr and the macro
2476 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2478 - Fixed a subtle bug that could cause false positive data race reports.
2480 * ==================== OTHER CHANGES ====================
2482 * The C++ demangler has been updated so as to work well with C++
2483 compiled by up to at least g++ 4.6.
2485 * Tool developers can make replacement/wrapping more flexible thanks
2486 to the new option --soname-synonyms. This was reported above, but
2487 in fact is very general and applies to all function
2488 replacement/wrapping, not just to malloc-family functions.
2490 * Round-robin scheduling of threads can be selected, using the new
2491 option --fair-sched= yes. Prior to this change, the pipe-based
2492 thread serialisation mechanism (which is still the default) could
2493 give very unfair scheduling. --fair-sched=yes improves
2494 responsiveness of interactive multithreaded applications, and
2495 improves repeatability of results from the thread checkers Helgrind
2498 * For tool developers: support to run Valgrind on Valgrind has been
2499 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2501 * gdbserver now shows the float shadow registers as integer
2502 rather than float values, as the shadow values are mostly
2503 used as bit patterns.
2505 * Increased limit for the --num-callers command line flag to 500.
2507 * Performance improvements for error matching when there are many
2508 suppression records in use.
2510 * Improved support for DWARF4 debugging information (bug 284184).
2512 * Initial support for DWZ compressed Dwarf debug info.
2514 * Improved control over the IR optimiser's handling of the tradeoff
2515 between performance and precision of exceptions. Specifically,
2516 --vex-iropt-precise-memory-exns has been removed and replaced by
2517 --vex-iropt-register-updates, with extended functionality. This
2518 allows the Valgrind gdbserver to always show up to date register
2521 * Modest performance gains through the use of translation chaining for
2524 * ==================== FIXED BUGS ====================
2526 The following bugs have been fixed or resolved. Note that "n-i-bz"
2527 stands for "not in bugzilla" -- that is, a bug that was reported to us
2528 but never got a bugzilla entry. We encourage you to file bugs in
2529 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2530 than mailing the developers (or mailing lists) directly -- bugs that
2531 are not entered into bugzilla tend to get forgotten about or ignored.
2533 To see details of a given bug, visit
2534 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2535 where XXXXXX is the bug number as listed below.
2537 197914 Building valgrind from svn now requires automake-1.10
2538 203877 increase to 16Mb maximum allowed alignment for memalign et al
2539 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
2540 247386 make perf does not run all performance tests
2541 270006 Valgrind scheduler unfair
2542 270777 Adding MIPS/Linux port to Valgrind
2543 270796 s390x: Removed broken support for the TS insn
2544 271438 Fix configure for proper SSE4.2 detection
2545 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
2546 273475 Add support for AVX instructions
2547 274078 improved configure logic for mpicc
2548 276993 fix mremap 'no thrash checks'
2549 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
2550 281482 memcheck incorrect byte allocation count in realloc() for silly argument
2551 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
2552 283413 Fix wrong sanity check
2553 283671 Robustize alignment computation in LibVEX_Alloc
2554 283961 Adding support for some HCI IOCTLs
2555 284124 parse_type_DIE: confused by: DWARF 4
2556 284864 == 273475 (Add support for AVX instructions)
2557 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
2558 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
2559 285725 == 273475 (Add support for AVX instructions)
2560 286261 add wrapper for linux I2C_RDWR ioctl
2561 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
2562 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
2563 286384 configure fails "checking for a supported version of gcc"
2564 286497 == 273475 (Add support for AVX instructions)
2565 286596 == 273475 (Add support for AVX instructions)
2566 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
2567 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
2568 287260 Incorrect conditional jump or move depends on uninitialised value(s)
2569 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
2570 287307 == 273475 (Add support for AVX instructions)
2571 287858 VG_(strerror): unknown error
2572 288298 (MacOSX) unhandled syscall shm_unlink
2573 288995 == 273475 (Add support for AVX instructions)
2574 289470 Loading of large Mach-O thin binaries fails.
2575 289656 == 273475 (Add support for AVX instructions)
2576 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
2577 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
2578 289839 s390x: Provide support for unicode conversion instructions
2579 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
2580 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
2581 290655 Add support for AESKEYGENASSIST instruction
2582 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
2583 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
2584 291253 ES register not initialised in valgrind simulation
2585 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
2586 291865 s390x: Support the "Compare Double and Swap" family of instructions
2587 292300 == 273475 (Add support for AVX instructions)
2588 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
2589 292493 == 273475 (Add support for AVX instructions)
2590 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
2591 292627 Missing support for some SCSI ioctls
2592 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
2593 292841 == 273475 (Add support for AVX instructions)
2594 292993 implement the getcpu syscall on amd64-linux
2595 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
2596 293088 Add some VEX sanity checks for ppc64 unhandled instructions
2597 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
2598 293754 PCMPxSTRx not implemented for 16-bit characters
2599 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
2600 293808 CLFLUSH not supported by latest VEX for amd64
2601 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
2602 294048 MPSADBW instruction not implemented
2603 294055 regtest none/tests/shell fails when locale is not set to C
2604 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
2605 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
2606 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
2607 294260 disInstr_AMD64: disInstr miscalculated next %rip
2608 294523 --partial-loads-ok=yes causes false negatives
2609 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
2610 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
2611 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
2612 295089 can not annotate source for both helgrind and drd
2613 295221 POWER Processor decimal floating point instruction support missing
2614 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
2615 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
2616 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
2617 295617 ARM - Add some missing syscalls
2618 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
2619 296229 Linux user input device ioctls missing wrappers
2620 296318 ELF Debug info improvements (more than one rx/rw mapping)
2621 296422 Add translation chaining support
2622 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
2623 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
2624 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
2625 297078 gdbserver signal handling problems [..]
2626 297147 drd false positives on newly allocated memory
2627 297329 disallow decoding of IBM Power DFP insns on some machines
2628 297497 POWER Processor decimal floating point instruction support missing
2629 297701 Another alias for strncasecmp_l in libc-2.13.so
2630 297911 'invalid write' not reported when using APIs for custom mem allocators.
2631 297976 s390x: revisit EX implementation
2632 297991 Valgrind interferes with mmap()+ftell()
2633 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
2634 297993 Fix compilation of valgrind with gcc -g3.
2635 298080 POWER Processor DFP support missing, part 3
2636 298227 == 273475 (Add support for AVX instructions)
2637 298335 == 273475 (Add support for AVX instructions)
2638 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
2639 298394 s390x: Don't bail out on an unknown machine model. [..]
2640 298421 accept4() syscall (366) support is missing for ARM
2641 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
2642 298732 valgrind installation problem in ubuntu with kernel version 3.x
2643 298862 POWER Processor DFP instruction support missing, part 4
2644 298864 DWARF reader mis-parses DW_FORM_ref_addr
2645 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
2646 299053 Support DWARF4 DW_AT_high_pc constant form
2647 299104 == 273475 (Add support for AVX instructions)
2648 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
2649 299629 dup3() syscall (358) support is missing for ARM
2650 299694 POWER Processor DFP instruction support missing, part 5
2651 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
2652 299803 == 273475 (Add support for AVX instructions)
2653 299804 == 273475 (Add support for AVX instructions)
2654 299805 == 273475 (Add support for AVX instructions)
2655 300140 ARM - Missing (T1) SMMUL
2656 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
2657 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
2658 300414 FCOM and FCOMP unimplemented for amd64 guest
2659 301204 infinite loop in canonicaliseSymtab with ifunc symbol
2660 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
2661 301265 add x86 support to Android build
2662 301984 configure script doesn't detect certain versions of clang
2663 302205 Fix compiler warnings for POWER VEX code and POWER test cases
2664 302287 Unhandled movbe instruction on Atom processors
2665 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
2666 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
2667 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
2668 302656 == 273475 (Add support for AVX instructions)
2669 302709 valgrind for ARM needs extra tls support for android emulator [..]
2670 302827 add wrapper for CDROM_GET_CAPABILITY
2671 302901 Valgrind crashes with dwz optimized debuginfo
2672 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
2673 303116 Add support for the POWER instruction popcntb
2674 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
2675 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
2676 303466 == 273475 (Add support for AVX instructions)
2677 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
2678 303963 strstr() function produces wrong results under valgrind callgrind
2679 304054 CALL_FN_xx macros need to enforce stack alignment
2680 304561 tee system call not supported
2681 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
2682 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
2683 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
2684 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
2685 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
2686 n-i-bz s390x: Shadow registers can now be examined using vgdb
2688 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
2689 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
2693 Release 3.7.0 (5 November 2011)
2694 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2695 3.7.0 is a feature release with many significant improvements and the
2696 usual collection of bug fixes.
2698 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2699 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
2700 Support for recent distros and toolchain components (glibc 2.14, gcc
2701 4.6, MacOSX 10.7) has been added.
2703 * ================== PLATFORM CHANGES =================
2705 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
2706 analyse 64-bit programs running on z/Architecture. Most user space
2707 instructions up to and including z10 are supported. Valgrind has
2708 been tested extensively on z9, z10, and z196 machines running SLES
2709 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
2710 known to work well. Callgrind, Helgrind, and DRD work reasonably
2711 well on z9 and later models. See README.s390 for more details.
2713 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
2714 64-bit processes are supported. Some complex threaded applications
2715 (Firefox) are observed to hang when run as 32 bit applications,
2716 whereas 64-bit versions run OK. The cause is unknown. Memcheck
2717 will likely report some false errors. In general, expect some rough
2718 spots. This release also supports MacOSX 10.6, but drops support
2721 * Preliminary support for Android (on ARM). Valgrind can now run
2722 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
2723 README.android for more details, plus instructions on how to get
2726 * Support for the IBM Power ISA 2.06 (Power7 instructions)
2728 * General correctness and performance improvements for ARM/Linux, and,
2729 by extension, ARM/Android.
2731 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
2732 instruction set support is under development but is not available in
2735 * Support for AIX5 has been removed.
2737 * ==================== TOOL CHANGES ====================
2739 * Memcheck: some incremental changes:
2741 - reduction of memory use in some circumstances
2743 - improved handling of freed memory, which in some circumstances
2744 can cause detection of use-after-free that would previously have
2747 - fix of a longstanding bug that could cause false negatives (missed
2748 errors) in programs doing vector saturated narrowing instructions.
2750 * Helgrind: performance improvements and major memory use reductions,
2751 particularly for large, long running applications which perform many
2752 synchronisation (lock, unlock, etc) events. Plus many smaller
2755 - display of locksets for both threads involved in a race
2757 - general improvements in formatting/clarity of error messages
2759 - addition of facilities and documentation regarding annotation
2760 of thread safe reference counted C++ classes
2762 - new flag --check-stack-refs=no|yes [yes], to disable race checking
2763 on thread stacks (a performance hack)
2765 - new flag --free-is-write=no|yes [no], to enable detection of races
2766 where one thread accesses heap memory but another one frees it,
2767 without any coordinating synchronisation event
2769 * DRD: enabled XML output; added support for delayed thread deletion
2770 in order to detect races that occur close to the end of a thread
2771 (--join-list-vol); fixed a memory leak triggered by repeated client
2772 memory allocatation and deallocation; improved Darwin support.
2774 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
2776 * exp-sgcheck: this tool has been reduced in scope so as to improve
2777 performance and remove checking that Memcheck does better.
2778 Specifically, the ability to check for overruns for stack and global
2779 arrays is unchanged, but the ability to check for overruns of heap
2780 blocks has been removed. The tool has accordingly been renamed to
2781 exp-sgcheck ("Stack and Global Array Checking").
2783 * ==================== OTHER CHANGES ====================
2785 * GDB server: Valgrind now has an embedded GDB server. That means it
2786 is possible to control a Valgrind run from GDB, doing all the usual
2787 things that GDB can do (single stepping, breakpoints, examining
2788 data, etc). Tool-specific functionality is also available. For
2789 example, it is possible to query the definedness state of variables
2790 or memory from within GDB when running Memcheck; arbitrarily large
2791 memory watchpoints are supported, etc. To use the GDB server, start
2792 Valgrind with the flag --vgdb-error=0 and follow the on-screen
2795 * Improved support for unfriendly self-modifying code: a new option
2796 --smc-check=all-non-file is available. This adds the relevant
2797 consistency checks only to code that originates in non-file-backed
2798 mappings. In effect this confines the consistency checking only to
2799 code that is or might be JIT generated, and avoids checks on code
2800 that must have been compiled ahead of time. This significantly
2801 improves performance on applications that generate code at run time.
2803 * It is now possible to build a working Valgrind using Clang-2.9 on
2806 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
2807 These enable and disable error reporting on a per-thread, and
2808 nestable, basis. This is useful for hiding errors in particularly
2809 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
2810 now uses this facility.
2812 * Added the --mod-funcname option to cg_diff.
2814 * ==================== FIXED BUGS ====================
2816 The following bugs have been fixed or resolved. Note that "n-i-bz"
2817 stands for "not in bugzilla" -- that is, a bug that was reported to us
2818 but never got a bugzilla entry. We encourage you to file bugs in
2819 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2820 mailing the developers (or mailing lists) directly -- bugs that are
2821 not entered into bugzilla tend to get forgotten about or ignored.
2823 To see details of a given bug, visit
2824 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2825 where XXXXXX is the bug number as listed below.
2827 79311 malloc silly arg warning does not give stack trace
2828 210935 port valgrind.h (not valgrind) to win32 to support client requests
2829 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
2830 243404 Port to zSeries
2831 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
2832 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
2833 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
2834 253206 Some fixes for the faultstatus testcase
2835 255223 capget testcase fails when running as root
2836 256703 xlc_dbl_u32.c testcase broken
2837 256726 Helgrind tests have broken inline asm
2838 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
2839 264800 testcase compile failure on zseries
2840 265762 make public VEX headers compilable by G++ 3.x
2841 265771 assertion in jumps.c (r11523) fails with glibc-2.3
2842 266753 configure script does not give the user the option to not use QtCore
2843 266931 gen_insn_test.pl is broken
2844 266961 ld-linux.so.2 i?86-linux strlen issues
2845 266990 setns instruction causes false positive
2846 267020 Make directory for temporary files configurable at run-time.
2847 267342 == 267997 (segmentation fault on Mac OS 10.6)
2848 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
2849 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
2850 267488 regtest: darwin support for 64-bit build
2851 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
2852 267630 Add support for IBM Power ISA 2.06 -- stage 1
2853 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
2854 267819 Add client request for informing the core about reallocation
2855 267925 laog data structure quadratic for a single sequence of lock
2856 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
2857 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
2858 268513 missed optimizations in fold_Expr
2859 268619 s390x: fpr - gpr transfer facility
2860 268620 s390x: reconsider "long displacement" requirement
2861 268621 s390x: improve IR generation for XC
2862 268715 s390x: FLOGR is not universally available
2863 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
2864 268930 s390x: MHY is not universally available
2865 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
2866 269079 Support ptrace system call on ARM
2867 269144 missing "Bad option" error message
2868 269209 conditional load and store facility (z196)
2869 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
2870 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
2871 269736 s390x: minor code generation tweaks
2872 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
2873 269863 s390x: remove unused function parameters
2874 269864 s390x: tweak s390_emit_load_cc
2875 269884 == 250101 (overhead for huge blocks exhausts space too soon)
2876 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
2877 270115 s390x: rewrite some testcases
2878 270309 == 267997 (valgrind crash on startup)
2879 270320 add support for Linux FIOQSIZE ioctl() call
2880 270326 segfault while trying to sanitize the environment passed to execle
2881 270794 IBM POWER7 support patch causes regression in none/tests
2882 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
2883 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
2884 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
2885 270959 s390x: invalid use of R0 as base register
2886 271042 VSX configure check fails when it should not
2887 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
2888 271259 s390x: fix code confusion
2889 271337 == 267997 (Valgrind segfaults on MacOS X)
2890 271385 s390x: Implement Ist_MBE
2891 271501 s390x: misc cleanups
2892 271504 s390x: promote likely and unlikely
2893 271579 ppc: using wrong enum type
2894 271615 unhandled instruction "popcnt" (arch=amd10h)
2895 271730 Fix bug when checking ioctls: duplicate check
2896 271776 s390x: provide STFLE instruction support
2897 271779 s390x: provide clock instructions like STCK
2898 271799 Darwin: ioctls without an arg report a memory error
2899 271820 arm: fix type confusion
2900 271917 pthread_cond_timedwait failure leads to not-locked false positive
2901 272067 s390x: fix DISP20 macro
2902 272615 A typo in debug output in mc_leakcheck.c
2903 272661 callgrind_annotate chokes when run from paths containing regex chars
2904 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
2905 272955 Unhandled syscall error for pwrite64 on ppc64 arch
2906 272967 make documentation build-system more robust
2907 272986 Fix gcc-4.6 warnings with valgrind.h
2908 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
2909 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
2910 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
2911 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
2912 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
2913 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
2914 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
2915 273778 exp-ptrcheck: unhandled sysno == 259
2916 274089 exp-ptrcheck: unhandled sysno == 208
2917 274378 s390x: Various dispatcher tweaks
2918 274447 WARNING: unhandled syscall: 340
2919 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
2920 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
2921 274926 valgrind does not build against linux-3
2922 275148 configure FAIL with glibc-2.14
2923 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
2924 275168 Make Valgrind work for MacOSX 10.7 Lion
2925 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
2926 275278 valgrind does not build on Linux kernel 3.0.* due to silly
2927 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
2928 275308 Fix implementation for ppc64 fres instruc
2929 275339 s390x: fix testcase compile warnings
2930 275517 s390x: Provide support for CKSM instruction
2931 275710 s390x: get rid of redundant address mode calculation
2932 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
2933 275852 == 250101 (valgrind uses all swap space and is killed)
2934 276784 Add support for IBM Power ISA 2.06 -- stage 3
2935 276987 gdbsrv: fix tests following recent commits
2936 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
2937 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
2938 277471 Unhandled syscall: 340
2939 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
2940 277653 ARM: support Thumb2 PLD instruction
2941 277663 ARM: NEON float VMUL by scalar incorrect
2942 277689 ARM: tests for VSTn with register post-index are broken
2943 277694 ARM: BLX LR instruction broken in ARM mode
2944 277780 ARM: VMOV.F32 (immediate) instruction is broken
2945 278057 fuse filesystem syscall deadlocks
2946 278078 Unimplemented syscall 280 on ppc32
2947 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
2948 278454 VALGRIND_STACK_DEREGISTER has wrong output type
2949 278502 == 275284 (Valgrind confuses memcpy() and memmove())
2950 278892 gdbsrv: factorize gdb version handling, fix doc and typos
2951 279027 Support for MVCL and CLCL instruction
2952 279027 s390x: Provide support for CLCL and MVCL instructions
2953 279062 Remove a redundant check in the insn selector for ppc.
2954 279071 JDK creates PTEST with redundant REX.W prefix
2955 279212 gdbsrv: add monitor cmd v.info scheduler.
2956 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
2957 279698 memcheck discards valid-bits for packuswb
2958 279795 memcheck reports uninitialised values for mincore on amd64
2959 279994 Add support for IBM Power ISA 2.06 -- stage 3
2960 280083 mempolicy syscall check errors
2961 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
2962 280710 s390x: config files for nightly builds
2963 280757 /tmp dir still used by valgrind even if TMPDIR is specified
2964 280965 Valgrind breaks fcntl locks when program does mmap
2965 281138 WARNING: unhandled syscall: 340
2966 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
2967 281304 == 275168 (Darwin: dyld "cannot load inserted library")
2968 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
2969 281468 s390x: handle do_clone and gcc clones in call traces
2970 281488 ARM: VFP register corruption
2971 281828 == 275284 (false memmove warning: "Source and destination overlap")
2972 281883 s390x: Fix system call wrapper for "clone".
2973 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
2974 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
2975 282238 SLES10: make check fails
2976 282979 strcasestr needs replacement with recent(>=2.12) glibc
2977 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
2978 283243 Regression in ppc64 memcheck tests
2979 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
2980 283427 re-connect epoll_pwait syscall on ARM linux
2981 283600 gdbsrv: android: port vgdb.c
2982 283709 none/tests/faultstatus needs to account for page size
2983 284305 filter_gdb needs enhancement to work on ppc64
2984 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
2985 284472 Thumb2 ROR.W encoding T2 not implemented
2986 284621 XML-escape process command line in XML output
2987 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
2988 that have non-power-of-2 sizes (also AMDs)
2989 n-i-bz don't be spooked by libraries mashed by elfhack
2990 n-i-bz don't be spooked by libxul.so linked with gold
2991 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
2993 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
2994 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
2995 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
2999 Release 3.6.1 (16 February 2011)
3000 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3001 3.6.1 is a bug fix release. It adds support for some SSE4
3002 instructions that were omitted in 3.6.0 due to lack of time. Initial
3003 support for glibc-2.13 has been added. A number of bugs causing
3004 crashing or assertion failures have been fixed.
3006 The following bugs have been fixed or resolved. Note that "n-i-bz"
3007 stands for "not in bugzilla" -- that is, a bug that was reported to us
3008 but never got a bugzilla entry. We encourage you to file bugs in
3009 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3010 mailing the developers (or mailing lists) directly -- bugs that are
3011 not entered into bugzilla tend to get forgotten about or ignored.
3013 To see details of a given bug, visit
3014 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3015 where XXXXXX is the bug number as listed below.
3017 188572 Valgrind on Mac should suppress setenv() mem leak
3018 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
3019 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
3020 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3021 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
3022 254420 memory pool tracking broken
3023 254957 Test code failing to compile due to changes in memcheck.h
3024 255009 helgrind/drd: crash on chmod with invalid parameter
3025 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
3026 255355 helgrind/drd: crash on threaded programs doing fork
3028 255418 (SSE4.x) rint call compiled with ICC
3029 255822 --gen-suppressions can create invalid files: "too many callers [...]"
3030 255888 closing valgrindoutput tag outputted to log-stream on error
3031 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
3032 255966 Slowness when using mempool annotations
3033 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
3034 256600 super-optimized strcasecmp() false positive
3035 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
3036 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
3037 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
3038 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
3039 257276 Missing case in memcheck --track-origins=yes
3040 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
3041 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
3042 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
3043 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
3044 263099 callgrind_annotate counts Ir improperly [...]
3045 263877 undefined coprocessor instruction on ARMv7
3046 265964 configure FAIL with glibc-2.13
3047 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
3048 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
3049 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
3050 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
3051 n-i-bz DRD: disable --free-is-write due to implementation difficulties
3053 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
3057 Release 3.6.0 (21 October 2010)
3058 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3059 3.6.0 is a feature release with many significant improvements and the
3060 usual collection of bug fixes.
3062 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
3063 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
3064 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
3066 -------------------------
3068 Here are some highlights. Details are shown further down:
3070 * Support for ARM/Linux.
3072 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3074 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3076 * Support for the SSE4.2 instruction set.
3078 * Enhancements to the Callgrind profiler, including the ability to
3079 handle CPUs with three levels of cache.
3081 * A new experimental heap profiler, DHAT.
3083 * A huge number of bug fixes and small enhancements.
3085 -------------------------
3087 Here are details of the above changes, together with descriptions of
3088 many other changes, and a list of fixed bugs.
3090 * ================== PLATFORM CHANGES =================
3092 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3093 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3094 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3096 This requires a CPU capable of running the ARMv7-A instruction set
3097 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3098 of the user space instruction set, including ARM and Thumb integer
3099 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3100 Cachegrind and Massif tools work properly; other tools work to
3103 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3104 with support for recent releases of the underlying toolchain
3105 components, notably gcc-4.5 and glibc-2.12.
3107 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3108 support also works much better on OS X 10.5, and is as solid as
3111 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3112 64-bit mode. In 32-bit mode, support is only available up to and
3113 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3114 supported in 64-bit mode, and 32-bit mode does in fact support the
3115 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3116 10.6 on 32-bit targets.
3118 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3119 and including version 2.05 is supported.
3121 * ==================== TOOL CHANGES ====================
3123 * Cachegrind has a new processing script, cg_diff, which finds the
3124 difference between two profiles. It's very useful for evaluating
3125 the performance effects of a change in a program.
3127 Related to this change, the meaning of cg_annotate's (rarely-used)
3128 --threshold option has changed; this is unlikely to affect many
3129 people, if you do use it please see the user manual for details.
3131 * Callgrind now can do branch prediction simulation, similar to
3132 Cachegrind. In addition, it optionally can count the number of
3133 executed global bus events. Both can be used for a better
3134 approximation of a "Cycle Estimation" as derived event (you need to
3135 update the event formula in KCachegrind yourself).
3137 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3138 rather than the L2 cache. This is to accommodate machines with
3139 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3140 cache configuration of such a machine it will run the simulation as
3141 if the L2 cache isn't present. This means the results are less
3142 likely to match the true result for the machine, but
3143 Cachegrind/Callgrind's results are already only approximate, and
3144 should not be considered authoritative. The results are still
3145 useful for giving a general idea about a program's locality.
3147 * Massif has a new option, --pages-as-heap, which is disabled by
3148 default. When enabled, instead of tracking allocations at the level
3149 of heap blocks (as allocated with malloc/new/new[]), it instead
3150 tracks memory allocations at the level of memory pages (as mapped by
3151 mmap, brk, etc). Each mapped page is treated as its own block.
3152 Interpreting the page-level output is harder than the heap-level
3153 output, but this option is useful if you want to account for every
3154 byte of memory used by a program.
3156 * DRD has two new command-line options: --free-is-write and
3157 --trace-alloc. The former allows to detect reading from already freed
3158 memory, and the latter allows tracing of all memory allocations and
3161 * DRD has several new annotations. Custom barrier implementations can
3162 now be annotated, as well as benign races on static variables.
3164 * DRD's happens before / happens after annotations have been made more
3165 powerful, so that they can now also be used to annotate e.g. a smart
3166 pointer implementation.
3168 * Helgrind's annotation set has also been drastically improved, so as
3169 to provide to users a general set of annotations to describe locks,
3170 semaphores, barriers and condition variables. Annotations to
3171 describe thread-safe reference counted heap objects have also been
3174 * Memcheck has a new command-line option, --show-possibly-lost, which
3175 is enabled by default. When disabled, the leak detector will not
3176 show possibly-lost blocks.
3178 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3179 has been added. DHAT keeps track of allocated heap blocks, and also
3180 inspects every memory reference to see which block (if any) is being
3181 accessed. This gives a lot of insight into block lifetimes,
3182 utilisation, turnover, liveness, and the location of hot and cold
3183 fields. You can use DHAT to do hot-field profiling.
3185 * ==================== OTHER CHANGES ====================
3187 * Improved support for unfriendly self-modifying code: the extra
3188 overhead incurred by --smc-check=all has been reduced by
3189 approximately a factor of 5 as compared with 3.5.0.
3191 * Ability to show directory names for source files in error messages.
3192 This is combined with a flexible mechanism for specifying which
3193 parts of the paths should be shown. This is enabled by the new flag
3196 * A new flag, --require-text-symbol, which will stop the run if a
3197 specified symbol is not found it a given shared object when it is
3198 loaded into the process. This makes advanced working with function
3199 intercepting and wrapping safer and more reliable.
3201 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3202 and control of Valgrind is now available for the tools Memcheck and
3203 Helgrind. XML output from Valgrind is available for Memcheck,
3204 Helgrind and exp-Ptrcheck.
3206 * More reliable stack unwinding on amd64-linux, particularly in the
3207 presence of function wrappers, and with gcc-4.5 compiled code.
3209 * Modest scalability (performance improvements) for massive
3210 long-running applications, particularly for those with huge amounts
3213 * Support for analyzing programs running under Wine with has been
3214 improved. The header files <valgrind/valgrind.h>,
3215 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3216 Windows-programs compiled with MinGW or one of the Microsoft Visual
3219 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3220 The 32-bit simulator was not affected. This did not occur often,
3221 but when it did would usually crash the program under test.
3224 * A large number of bugs were fixed. These are shown below.
3226 * A number of bugs were investigated, and were candidates for fixing,
3227 but are not fixed in 3.6.0, due to lack of developer time. They may
3228 get fixed in later releases. They are:
3230 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3231 212419 false positive "lock order violated" (A+B vs A)
3232 213685 Undefined value propagates past dependency breaking instruction
3233 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3234 237920 valgrind segfault on fork failure
3235 242137 support for code compiled by LLVM-2.8
3236 242423 Another unknown Intel cache config value
3237 243232 Inconsistent Lock Orderings report with trylock
3238 243483 ppc: callgrind triggers VEX assertion failure
3239 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3240 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3242 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3243 249435 Analyzing wine programs with callgrind triggers a crash
3244 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3245 250065 Handling large allocations
3246 250101 huge "free" memory usage due to m_mallocfree.c
3247 "superblocks fragmentation"
3248 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3249 252091 Callgrind on ARM does not detect function returns correctly
3250 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3251 254420 memory pool tracking broken
3252 n-i-bz support for adding symbols for JIT generated code
3255 The following bugs have been fixed or resolved. Note that "n-i-bz"
3256 stands for "not in bugzilla" -- that is, a bug that was reported to us
3257 but never got a bugzilla entry. We encourage you to file bugs in
3258 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3259 mailing the developers (or mailing lists) directly -- bugs that are
3260 not entered into bugzilla tend to get forgotten about or ignored.
3262 To see details of a given bug, visit
3263 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3264 where XXXXXX is the bug number as listed below.
3266 135264 dcbzl instruction missing
3268 153699 Valgrind should report unaligned reads with movdqa
3270 190429 Valgrind reports lost of errors in ld.so
3271 with x86_64 2.9.90 glibc
3272 197266 valgrind appears to choke on the xmms instruction
3274 197988 Crash when demangling very large symbol names
3275 202315 unhandled syscall: 332 (inotify_init1)
3276 203256 Add page-level profiling to Massif
3277 205093 dsymutil=yes needs quotes, locking (partial fix)
3278 205241 Snow Leopard 10.6 support (partial fix)
3279 206600 Leak checker fails to upgrade indirect blocks when their
3280 parent becomes reachable
3281 210935 port valgrind.h (not valgrind) to win32 so apps run under
3282 wine can make client requests
3283 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3284 within Linux ip-stack checksum functions
3285 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3287 213685 Undefined value propagates past dependency breaking instruction
3289 215914 Valgrind inserts bogus empty environment variable
3291 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3292 222545 shmat fails under valgind on some arm targets
3293 222560 ARM NEON support
3296 232509 Docs build fails with formatting inside <title></title> elements
3298 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3299 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3300 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3301 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3302 237485 symlink (syscall 57) is not supported on Mac OS
3303 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3305 238208 is_just_below_ESP doesn't take into account red-zone
3306 238345 valgrind passes wrong $0 when executing a shell script
3307 238679 mq_timedreceive syscall doesn't flag the reception buffer
3309 238696 fcntl command F_DUPFD_CLOEXEC not supported
3310 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3311 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3312 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
3314 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
3320 242606 unhandled syscall: setegid (in Ptrcheck)
3321 242814 Helgrind "Impossible has happened" during
3322 QApplication::initInstance();
3323 243064 Valgrind attempting to read debug information from iso
3324 243270 Make stack unwinding in Valgrind wrappers more reliable
3325 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
3326 sysno = 277 (mq_open)
3327 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
3328 244493 ARM VFP d16-d31 registers support
3329 244670 add support for audit_session_self syscall on Mac OS 10.6
3330 244921 The xml report of helgrind tool is not well format
3331 244923 In the xml report file, the <preamble> not escape the
3332 xml char, eg '<','&','>'
3333 245535 print full path names in plain text reports
3334 245925 x86-64 red zone handling problem
3335 246258 Valgrind not catching integer underruns + new [] s
3336 246311 reg/reg cmpxchg doesn't work on amd64
3337 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
3338 246888 Improve Makefile.vex.am
3339 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
3340 to [f]chmod_extended
3341 247526 IBM POWER6 (ISA 2.05) support is incomplete
3342 247561 Some leak testcases fails due to reachable addresses in
3344 247875 sizeofIRType to handle Ity_I128
3345 247894 [PATCH] unhandled syscall sys_readahead
3346 247980 Doesn't honor CFLAGS passed to configure
3347 248373 darwin10.supp is empty in the trunk
3348 248822 Linux FIBMAP ioctl has int parameter instead of long
3349 248893 [PATCH] make readdwarf.c big endianess safe to enable
3350 unwinding on big endian systems
3351 249224 Syscall 336 not supported (SYS_proc_info)
3353 249775 Incorrect scheme for detecting NEON capabilities of host CPU
3354 249943 jni JVM init fails when using valgrind
3355 249991 Valgrind incorrectly declares AESKEYGENASSIST support
3357 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
3358 250799 frexp$fenv_access_off function generates SIGILL
3359 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
3360 251251 support pclmulqdq insn
3361 251362 valgrind: ARM: attach to debugger either fails or provokes
3363 251674 Unhandled syscall 294
3366 254257 Add support for debugfiles found by build-id
3367 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
3368 254646 Wrapped functions cause stack misalignment on OS X
3369 (and possibly Linux)
3370 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
3372 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
3376 Release 3.5.0 (19 August 2009)
3377 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3378 3.5.0 is a feature release with many significant improvements and the
3379 usual collection of bug fixes. The main improvement is that Valgrind
3380 now works on Mac OS X.
3382 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
3383 and X86/Darwin. Support for recent distros and toolchain components
3384 (glibc 2.10, gcc 4.5) has been added.
3386 -------------------------
3388 Here is a short summary of the changes. Details are shown further
3391 * Support for Mac OS X (10.5.x).
3393 * Improvements and simplifications to Memcheck's leak checker.
3395 * Clarification and simplifications in various aspects of Valgrind's
3398 * XML output for Helgrind and Ptrcheck.
3400 * Performance and stability improvements for Helgrind and DRD.
3402 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
3404 * A new experimental tool, BBV, useful for computer architecture
3407 * Improved Wine support, including ability to read Windows PDB
3410 -------------------------
3412 Here are details of the above changes, followed by descriptions of
3413 many other minor changes, and a list of fixed bugs.
3416 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
3417 called "Darwin" because that is the name of the OS core, which is the
3418 level that Valgrind works at.)
3422 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
3423 because it would require work and 10.4 is only becoming less common.
3425 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
3426 fairly well. For 10.5.x, 32-bit programs are the default even on
3427 64-bit machines, so it handles most current programs.
3429 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
3430 officially supported, but simple programs at least will probably work.
3431 However, start-up is slow.
3433 - PowerPC machines are not supported.
3435 Things that don't work:
3437 - The Ptrcheck tool.
3439 - Objective-C garbage collection.
3443 - If you have Rogue Amoeba's "Instant Hijack" program installed,
3444 Valgrind will fail with a SIGTRAP at start-up. See
3445 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
3450 - You will likely find --dsymutil=yes a useful option, as error
3451 messages may be imprecise without it.
3453 - Mac OS X support is new and therefore will be less robust than the
3454 Linux support. Please report any bugs you find.
3456 - Threaded programs may run more slowly than on Linux.
3458 Many thanks to Greg Parker for developing this port over several years.
3461 * Memcheck's leak checker has been improved.
3463 - The results for --leak-check=summary now match the summary results
3464 for --leak-check=full. Previously they could differ because
3465 --leak-check=summary counted "indirectly lost" blocks and
3466 "suppressed" blocks as "definitely lost".
3468 - Blocks that are only reachable via at least one interior-pointer,
3469 but are directly pointed to by a start-pointer, were previously
3470 marked as "still reachable". They are now correctly marked as
3473 - The default value for the --leak-resolution option has been
3474 changed from "low" to "high". In general, this means that more
3475 leak reports will be produced, but each leak report will describe
3476 fewer leaked blocks.
3478 - With --leak-check=full, "definitely lost" and "possibly lost"
3479 leaks are now considered as proper errors, ie. they are counted
3480 for the "ERROR SUMMARY" and affect the behaviour of
3481 --error-exitcode. These leaks are not counted as errors if
3482 --leak-check=summary is specified, however.
3484 - Documentation for the leak checker has been improved.
3487 * Various aspects of Valgrind's text output have changed.
3489 - Valgrind's start-up message has changed. It is shorter but also
3490 includes the command being run, which makes it easier to use
3491 --trace-children=yes. An example:
3493 - Valgrind's shut-down messages have also changed. This is most
3494 noticeable with Memcheck, where the leak summary now occurs before
3495 the error summary. This change was necessary to allow leaks to be
3496 counted as proper errors (see the description of the leak checker
3497 changes above for more details). This was also necessary to fix a
3498 longstanding bug in which uses of suppressions against leaks were
3499 not "counted", leading to difficulties in maintaining suppression
3500 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3502 - Behavior of -v has changed. In previous versions, -v printed out
3503 a mixture of marginally-user-useful information, and tool/core
3504 statistics. The statistics printing has now been moved to its own
3505 flag, --stats=yes. This means -v is less verbose and more likely
3506 to convey useful end-user information.
3508 - The format of some (non-XML) stack trace entries has changed a
3509 little. Previously there were six possible forms:
3511 0x80483BF: really (a.c:20)
3512 0x80483BF: really (in /foo/a.out)
3514 0x80483BF: (within /foo/a.out)
3515 0x80483BF: ??? (a.c:20)
3518 The third and fourth of these forms have been made more consistent
3519 with the others. The six possible forms are now:
3521 0x80483BF: really (a.c:20)
3522 0x80483BF: really (in /foo/a.out)
3523 0x80483BF: really (in ???)
3524 0x80483BF: ??? (in /foo/a.out)
3525 0x80483BF: ??? (a.c:20)
3528 Stack traces produced when --xml=yes is specified are different
3532 * Helgrind and Ptrcheck now support XML output, so they can be used
3533 from GUI tools. Also, the XML output mechanism has been
3536 - The XML format has been overhauled and generalised, so it is more
3537 suitable for error reporting tools in general. The Memcheck
3538 specific aspects of it have been removed. The new format, which
3539 is an evolution of the old format, is described in
3540 docs/internals/xml-output-protocol4.txt.
3542 - Memcheck has been updated to use the new format.
3544 - Helgrind and Ptrcheck are now able to emit output in this format.
3546 - The XML output mechanism has been overhauled. XML is now output
3547 to its own file descriptor, which means that:
3549 * Valgrind can output text and XML independently.
3551 * The longstanding problem of XML output being corrupted by
3552 unexpected un-tagged text messages is solved.
3554 As before, the destination for text output is specified using
3555 --log-file=, --log-fd= or --log-socket=.
3557 As before, XML output for a tool is enabled using --xml=yes.
3559 Because there's a new XML output channel, the XML output
3560 destination is now specified by --xml-file=, --xml-fd= or
3563 Initial feedback has shown this causes some confusion. To
3564 clarify, the two envisaged usage scenarios are:
3566 (1) Normal text output. In this case, do not specify --xml=yes
3567 nor any of --xml-file=, --xml-fd= or --xml-socket=.
3569 (2) XML output. In this case, specify --xml=yes, and one of
3570 --xml-file=, --xml-fd= or --xml-socket= to select the XML
3571 destination, one of --log-file=, --log-fd= or --log-socket=
3572 to select the destination for any remaining text messages,
3573 and, importantly, -q.
3575 -q makes Valgrind completely silent on the text channel,
3576 except in the case of critical failures, such as Valgrind
3577 itself segfaulting, or failing to read debugging information.
3578 Hence, in this scenario, it suffices to check whether or not
3579 any output appeared on the text channel. If yes, then it is
3580 likely to be a critical error which should be brought to the
3581 attention of the user. If no (the text channel produced no
3582 output) then it can be assumed that the run was successful.
3584 This allows GUIs to make the critical distinction they need to
3585 make (did the run fail or not?) without having to search or
3586 filter the text output channel in any way.
3588 It is also recommended to use --child-silent-after-fork=yes in
3592 * Improvements and changes in Helgrind:
3594 - XML output, as described above
3596 - Checks for consistent association between pthread condition
3597 variables and their associated mutexes are now performed.
3599 - pthread_spinlock functions are supported.
3601 - Modest performance improvements.
3603 - Initial (skeletal) support for describing the behaviour of
3604 non-POSIX synchronisation objects through ThreadSanitizer
3605 compatible ANNOTATE_* macros.
3607 - More controllable tradeoffs between performance and the level of
3608 detail of "previous" accesses in a race. There are now three
3611 * --history-level=full. This is the default, and was also the
3612 default in 3.4.x. It shows both stacks involved in a race, but
3613 requires a lot of memory and can be very slow in programs that
3614 do many inter-thread synchronisation events.
3616 * --history-level=none. This only shows the later stack involved
3617 in a race. This can be much faster than --history-level=full,
3618 but makes it much more difficult to find the other access
3619 involved in the race.
3621 The new intermediate setting is
3623 * --history-level=approx
3625 For the earlier (other) access, two stacks are presented. The
3626 earlier access is guaranteed to be somewhere in between the two
3627 program points denoted by those stacks. This is not as useful
3628 as showing the exact stack for the previous access (as per
3629 --history-level=full), but it is better than nothing, and it's
3630 almost as fast as --history-level=none.
3633 * New features and improvements in DRD:
3635 - The error messages printed by DRD are now easier to interpret.
3636 Instead of using two different numbers to identify each thread
3637 (Valgrind thread ID and DRD thread ID), DRD does now identify
3638 threads via a single number (the DRD thread ID). Furthermore
3639 "first observed at" information is now printed for all error
3640 messages related to synchronization objects.
3642 - Added support for named semaphores (sem_open() and sem_close()).
3644 - Race conditions between pthread_barrier_wait() and
3645 pthread_barrier_destroy() calls are now reported.
3647 - Added support for custom allocators through the macros
3648 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
3649 in <valgrind/valgrind.h>). An alternative for these two macros is
3650 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
3653 - Added support for annotating non-POSIX synchronization objects
3654 through several new ANNOTATE_*() macros.
3656 - OpenMP: added support for the OpenMP runtime (libgomp) included
3657 with gcc versions 4.3.0 and 4.4.0.
3661 - Added two new command-line options (--first-race-only and
3662 --segment-merging-interval).
3665 * Genuinely atomic support for x86/amd64/ppc atomic instructions
3667 Valgrind will now preserve (memory-access) atomicity of LOCK-
3668 prefixed x86/amd64 instructions, and any others implying a global
3669 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
3671 This means that Valgrinded processes will "play nicely" in
3672 situations where communication with other processes, or the kernel,
3673 is done through shared memory and coordinated with such atomic
3674 instructions. Prior to this change, such arrangements usually
3675 resulted in hangs, races or other synchronisation failures, because
3676 Valgrind did not honour atomicity of such instructions.
3679 * A new experimental tool, BBV, has been added. BBV generates basic
3680 block vectors for use with the SimPoint analysis tool, which allows
3681 a program's overall behaviour to be approximated by running only a
3682 fraction of it. This is useful for computer architecture
3683 researchers. You can run BBV by specifying --tool=exp-bbv (the
3684 "exp-" prefix is short for "experimental"). BBV was written by
3688 * Modestly improved support for running Windows applications under
3689 Wine. In particular, initial support for reading Windows .PDB debug
3690 information has been added.
3693 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
3694 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
3698 * The Valgrind client requests VALGRIND_PRINTF and
3699 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
3700 the string was always printed immediately on its own line. Now, the
3701 string will be added to a buffer but not printed until a newline is
3702 encountered, or other Valgrind output is printed (note that for
3703 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
3704 "other Valgrind output"). This allows you to use multiple
3705 VALGRIND_PRINTF calls to build up a single output line, and also to
3706 print multiple output lines with a single request (by embedding
3707 multiple newlines in the string).
3710 * The graphs drawn by Massif's ms_print program have changed slightly:
3712 - The half-height chars '.' and ',' are no longer drawn, because
3713 they are confusing. The --y option can be used if the default
3714 y-resolution is not high enough.
3716 - Horizontal lines are now drawn after the top of a snapshot if
3717 there is a gap until the next snapshot. This makes it clear that
3718 the memory usage has not dropped to zero between snapshots.
3721 * Something that happened in 3.4.0, but wasn't clearly announced: the
3722 option --read-var-info=yes can be used by some tools (Memcheck,
3723 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
3724 variable type and location information. This makes those tools
3725 start up more slowly and increases memory consumption, but
3726 descriptions of data addresses in error messages become more
3730 * exp-Omega, an experimental instantaneous leak-detecting tool, was
3731 disabled in 3.4.0 due to a lack of interest and maintenance,
3732 although the source code was still in the distribution. The source
3733 code has now been removed from the distribution. For anyone
3734 interested, the removal occurred in SVN revision r10247.
3737 * Some changes have been made to the build system.
3739 - VEX/ is now integrated properly into the build system. This means
3740 that dependency tracking within VEX/ now works properly, "make
3741 install" will work without requiring "make" before it, and
3742 parallel builds (ie. 'make -j') now work (previously a
3743 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
3744 was effectively ignored).
3746 - The --with-vex configure option has been removed. It was of
3747 little use and removing it simplified the build system.
3749 - The location of some install files has changed. This should not
3750 affect most users. Those who might be affected:
3752 * For people who use Valgrind with MPI programs, the installed
3753 libmpiwrap.so library has moved from
3754 $(INSTALL)/<platform>/libmpiwrap.so to
3755 $(INSTALL)/libmpiwrap-<platform>.so.
3757 * For people who distribute standalone Valgrind tools, the
3758 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
3759 have moved to $(INSTALL)/libcoregrind-<platform>.a.
3761 These changes simplify the build system.
3763 - Previously, all the distributed suppression (*.supp) files were
3764 installed. Now, only default.supp is installed. This should not
3765 affect users as the other installed suppression files were not
3766 read; the fact that they were installed was a mistake.
3769 * KNOWN LIMITATIONS:
3771 - Memcheck is unusable with the Intel compiler suite version 11.1,
3772 when it generates code for SSE2-and-above capable targets. This
3773 is because of icc's use of highly optimised inlined strlen
3774 implementations. It causes Memcheck to report huge numbers of
3775 false errors even in simple programs. Helgrind and DRD may also
3778 Versions 11.0 and earlier may be OK, but this has not been
3782 The following bugs have been fixed or resolved. Note that "n-i-bz"
3783 stands for "not in bugzilla" -- that is, a bug that was reported to us
3784 but never got a bugzilla entry. We encourage you to file bugs in
3785 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3786 mailing the developers (or mailing lists) directly -- bugs that are
3787 not entered into bugzilla tend to get forgotten about or ignored.
3789 To see details of a given bug, visit
3790 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3791 where XXXXXX is the bug number as listed below.
3793 84303 How about a LockCheck tool?
3794 91633 dereference of null ptr in vgPlain_st_basetype
3795 97452 Valgrind doesn't report any pthreads problems
3796 100628 leak-check gets assertion failure when using
3797 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
3798 108528 NPTL pthread cleanup handlers not called
3799 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
3800 110128 mallinfo is not implemented...
3801 110770 VEX: Generated files not always updated when making valgrind
3802 111102 Memcheck: problems with large (memory footprint) applications
3803 115673 Vex's decoder should never assert
3804 117564 False positive: Syscall param clone(child_tidptr) contains
3805 uninitialised byte(s)
3806 119404 executing ssh from inside valgrind fails
3807 133679 Callgrind does not write path names to sources with dwarf debug
3809 135847 configure.in problem with non gnu compilers (and possible fix)
3810 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
3811 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
3812 136230 memcheck reports "possibly lost", should be "still reachable"
3813 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
3814 137904 Valgrind reports a memory leak when using POSIX threads,
3816 139076 valgrind VT_GETSTATE error
3817 142228 complaint of elf_dynamic_do_rela in trivial usage
3818 145347 spurious warning with USBDEVFS_REAPURB
3819 148441 (wine) can't find memory leak in Wine, win32 binary
3821 148742 Leak-check fails assert on exit
3822 149878 add (proper) check for calloc integer overflow
3823 150606 Call graph is broken when using callgrind control
3824 152393 leak errors produce an exit code of 0. I need some way to
3825 cause leak errors to result in a nonzero exit code.
3826 157154 documentation (leak-resolution doc speaks about num-callers
3827 def=4) + what is a loss record
3828 159501 incorrect handling of ALSA ioctls
3829 162020 Valgrinding an empty/zero-byte file crashes valgrind
3830 162482 ppc: Valgrind crashes while reading stabs information
3831 162718 x86: avoid segment selector 0 in sys_set_thread_area()
3832 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
3833 163560 VEX/test_main.c is missing from valgrind-3.3.1
3834 164353 malloc_usable_size() doesn't return a usable size
3835 165468 Inconsistent formatting in memcheck manual -- please fix
3836 169505 main.c:286 (endOfInstr):
3837 Assertion 'ii->cost_offset == *cost_offset' failed
3838 177206 Generate default.supp during compile instead of configure
3839 177209 Configure valt_load_address based on arch+os
3840 177305 eventfd / syscall 323 patch lost
3841 179731 Tests fail to build because of inlining of non-local asm labels
3842 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
3843 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
3844 181594 Bogus warning for empty text segment
3845 181707 dwarf doesn't require enumerations to have name
3846 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3847 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3848 Assertion '!already_present' failed.
3849 185359 exp-ptrcheck: unhandled syscall getresuid()
3850 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
3851 185816 Valgrind is unable to handle debug info for files with split
3852 debug info that are prelinked afterwards
3853 185980 [darwin] unhandled syscall: sem_open
3854 186238 bbToIR_AMD64: disInstr miscalculated next %rip
3855 186507 exp-ptrcheck unhandled syscalls prctl, etc.
3856 186790 Suppression pattern used for leaks are not reported
3857 186796 Symbols with length>200 in suppression files are ignored
3858 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
3859 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
3860 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
3861 188046 bashisms in the configure script
3862 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
3863 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
3864 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
3865 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
3866 assertion fail "!lock->heldBy"
3867 188427 Add support for epoll_create1 (with patch)
3868 188530 Support for SIOCGSTAMPNS
3869 188560 Include valgrind.spec in the tarball
3870 188572 Valgrind on Mac should suppress setenv() mem leak
3871 189054 Valgrind fails to build because of duplicate non-local asm labels
3872 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
3873 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
3874 189763 drd assertion failure: s_threadinfo[tid].is_recording
3875 190219 unhandled syscall: 328 (x86-linux)
3876 190391 dup of 181394; see above
3877 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
3878 190820 No debug information on powerpc-linux
3879 191095 PATCH: Improve usbdevfs ioctl handling
3880 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
3882 191189 --xml=yes should obey --gen-suppressions=all
3883 191192 syslog() needs a suppression on macosx
3884 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
3885 191761 getrlimit on MacOSX
3886 191992 multiple --fn-skip only works sometimes; dependent on order
3887 192634 V. reports "aspacem sync_check_mapping_callback:
3888 segment mismatch" on Darwin
3889 192954 __extension__ missing on 2 client requests
3890 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
3891 194474 "INSTALL" file has different build instructions than "README"
3892 194671 Unhandled syscall (sem_wait?) from mac valgrind
3893 195069 memcheck: reports leak (memory still reachable) for
3895 195169 drd: (vgDrd_barrier_post_wait):
3896 Assertion 'r->sg[p->post_iteration]' failed.
3897 195268 valgrind --log-file doesn't accept ~/...
3898 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
3899 195860 WARNING: unhandled syscall: unix:223
3900 196528 need a error suppression for pthread_rwlock_init under os x?
3901 197227 Support aio_* syscalls on Darwin
3902 197456 valgrind should reject --suppressions=(directory)
3903 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
3904 197591 unhandled syscall 27 (mincore)
3905 197793 Merge DCAS branch to the trunk == 85756, 142103
3906 197794 Avoid duplicate filenames in Vex
3907 197898 make check fails on current SVN
3908 197901 make check fails also under exp-ptrcheck in current SVN
3909 197929 Make --leak-resolution=high the default
3910 197930 Reduce spacing between leak reports
3911 197933 Print command line of client at start-up, and shorten preamble
3912 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
3913 198395 add BBV to the distribution as an experimental tool
3914 198624 Missing syscalls on Darwin: 82, 167, 281, 347
3915 198649 callgrind_annotate doesn't cumulate counters
3916 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
3917 199977 Valgrind complains about an unrecognized instruction in the
3918 atomic_incs test program
3919 200029 valgrind isn't able to read Fedora 12 debuginfo
3920 200760 darwin unhandled syscall: unix:284
3921 200827 DRD doesn't work on Mac OS X
3922 200990 VG_(read_millisecond_timer)() does not work correctly
3923 201016 Valgrind does not support pthread_kill() on Mac OS
3924 201169 Document --read-var-info
3925 201323 Pre-3.5.0 performance sanity checking
3926 201384 Review user manual for the 3.5.0 release
3927 201585 mfpvr not implemented on ppc
3928 201708 tests failing because x86 direction flag is left set
3929 201757 Valgrind doesn't handle any recent sys_futex additions
3930 204377 64-bit valgrind can not start a shell script
3931 (with #!/path/to/shell) if the shell is a 32-bit executable
3932 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
3933 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
3934 about memory allocation events with memory access tracing enabled
3935 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
3937 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
3941 Release 3.4.1 (28 February 2009)
3942 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3943 3.4.1 is a bug-fix release that fixes some regressions and assertion
3944 failures in debug info reading in 3.4.0, most notably incorrect stack
3945 traces on amd64-linux on older (glibc-2.3 based) systems. Various
3946 other debug info problems are also fixed. A number of bugs in the
3947 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
3949 In view of the fact that 3.4.0 contains user-visible regressions
3950 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
3951 encouraged to ship 3.4.1 in preference to 3.4.0.
3953 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
3954 bugzilla" -- that is, a bug that was reported to us but never got a
3955 bugzilla entry. We encourage you to file bugs in bugzilla
3956 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
3957 developers (or mailing lists) directly -- bugs that are not entered
3958 into bugzilla tend to get forgotten about or ignored.
3960 n-i-bz Fix various bugs reading icc-11 generated debug info
3961 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
3962 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
3963 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
3964 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
3965 179618 exp-ptrcheck crashed / exit prematurely
3966 179624 helgrind: false positive races with pthread_create and
3967 recv/open/close/read
3968 134207 pkg-config output contains @VG_PLATFORM@
3969 176926 floating point exception at valgrind startup with PPC 440EPX
3970 181594 Bogus warning for empty text segment
3971 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
3972 181707 Dwarf3 doesn't require enumerations to have name
3973 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3974 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3975 Assertion '!already_present' failed.
3976 185359 exp-ptrcheck unhandled syscall getresuid()
3978 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
3979 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
3983 Release 3.4.0 (2 January 2009)
3984 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3985 3.4.0 is a feature release with many significant improvements and the
3986 usual collection of bug fixes. This release supports X86/Linux,
3987 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
3988 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
3990 3.4.0 brings some significant tool improvements. Memcheck can now
3991 report the origin of uninitialised values, the thread checkers
3992 Helgrind and DRD are much improved, and we have a new experimental
3993 tool, exp-Ptrcheck, which is able to detect overruns of stack and
3994 global arrays. In detail:
3996 * Memcheck is now able to track the origin of uninitialised values.
3997 When it reports an uninitialised value error, it will try to show
3998 the origin of the value, as either a heap or stack allocation.
3999 Origin tracking is expensive and so is not enabled by default. To
4000 use it, specify --track-origins=yes. Memcheck's speed will be
4001 essentially halved, and memory usage will be significantly
4002 increased. Nevertheless it can drastically reduce the effort
4003 required to identify the root cause of uninitialised value errors,
4004 and so is often a programmer productivity win, despite running more
4007 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
4008 3.4.0, will be released shortly.
4010 * Helgrind's race detection algorithm has been completely redesigned
4011 and reimplemented, to address usability and scalability concerns:
4013 - The new algorithm has a lower false-error rate: it is much less
4014 likely to report races that do not really exist.
4016 - Helgrind will display full call stacks for both accesses involved
4017 in a race. This makes it easier to identify the root causes of
4020 - Limitations on the size of program that can run have been removed.
4022 - Performance has been modestly improved, although that is very
4025 - Direct support for Qt4 threading has been added.
4027 - pthread_barriers are now directly supported.
4029 - Helgrind works well on all supported Linux targets.
4031 * The DRD thread debugging tool has seen major improvements:
4033 - Greatly improved performance and significantly reduced memory
4036 - Support for several major threading libraries (Boost.Thread, Qt4,
4037 glib, OpenMP) has been added.
4039 - Support for atomic instructions, POSIX semaphores, barriers and
4040 reader-writer locks has been added.
4042 - Works now on PowerPC CPUs too.
4044 - Added support for printing thread stack usage at thread exit time.
4046 - Added support for debugging lock contention.
4048 - Added a manual for Drd.
4050 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
4051 checks for misuses of pointers. In that sense it is a bit like
4052 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
4053 detect overruns of stack and global arrays, it can detect
4054 arbitrarily far out-of-bounds accesses to heap blocks, and it can
4055 detect accesses heap blocks that have been freed a very long time
4056 ago (millions of blocks in the past).
4058 Ptrcheck currently works only on x86-linux and amd64-linux. To use
4059 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
4060 of the main Valgrind documentation. As this is an experimental
4061 tool, we would be particularly interested in hearing about your
4062 experiences with it.
4064 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
4065 longer built by default, although the code remains in the repository
4066 and the tarball. This is due to three factors: a perceived lack of
4067 users, a lack of maintenance, and concerns that it may not be
4068 possible to achieve reliable operation using the existing design.
4070 * As usual, support for the latest Linux distros and toolchain
4071 components has been added. It should work well on Fedora Core 10,
4072 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4073 state) is supported, as is glibc-2.9. The C++ demangler has been
4074 updated so as to work well with C++ compiled by even the most recent
4077 * You can now use frame-level wildcards in suppressions. This was a
4078 frequently-requested enhancement. A line "..." in a suppression now
4079 matches zero or more frames. This makes it easier to write
4080 suppressions which are precise yet insensitive to changes in
4083 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4085 * Very basic support for IBM Power6 has been added (64-bit processes only).
4087 * Valgrind is now cross-compilable. For example, it is possible to
4088 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4089 on a ppc32/64-linux target.
4091 * You can set the main thread's stack size at startup using the
4092 new --main-stacksize= flag (subject of course to ulimit settings).
4093 This is useful for running apps that need a lot of stack space.
4095 * The limitation that you can't use --trace-children=yes together
4096 with --db-attach=yes has been removed.
4098 * The following bugs have been fixed. Note that "n-i-bz" stands for
4099 "not in bugzilla" -- that is, a bug that was reported to us but
4100 never got a bugzilla entry. We encourage you to file bugs in
4101 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4102 mailing the developers (or mailing lists) directly.
4104 n-i-bz Make return types for some client requests 64-bit clean
4105 n-i-bz glibc 2.9 support
4106 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4107 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4108 n-i-bz Building in an env without gdb gives bogus gdb attach
4109 92456 Tracing the origin of uninitialised memory
4110 106497 Valgrind does not demangle some C++ template symbols
4112 151612 Suppression with "..." (frame-level wildcards in .supp files)
4113 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4114 159285 unhandled syscall:25 (stime, on x86-linux)
4115 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4116 160954 ppc build of valgrind crashes with illegal instruction (isel)
4117 160956 mallinfo implementation, w/ patch
4118 162092 Valgrind fails to start gnome-system-monitor
4119 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4120 163794 assertion failure with "--track-origins=yes"
4121 163933 sigcontext.err and .trapno must be set together
4122 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4123 164476 Missing kernel module loading system calls
4124 164669 SVN regression: mmap() drops posix file locks
4125 166581 Callgrind output corruption when program forks
4126 167288 Patch file for missing system calls on Cell BE
4127 168943 unsupported scas instruction pentium
4128 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4129 172417 x86->IR: 0x82 ...
4130 172563 amd64->IR: 0xD9 0xF5 - fprem1
4131 173099 .lds linker script generation error
4132 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4133 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4135 174908 --log-file value not expanded correctly for core file
4136 175044 Add lookup_dcookie for amd64
4137 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4139 Developer-visible changes:
4141 * Valgrind's debug-info reading machinery has been majorly overhauled.
4142 It can now correctly establish the addresses for ELF data symbols,
4143 which is something that has never worked properly before now.
4145 Also, Valgrind can now read DWARF3 type and location information for
4146 stack and global variables. This makes it possible to use the
4147 framework to build tools that rely on knowing the type and locations
4148 of stack and global variables, for example exp-Ptrcheck.
4150 Reading of such information is disabled by default, because most
4151 tools don't need it, and because it is expensive in space and time.
4152 However, you can force Valgrind to read it, using the
4153 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4154 make use of such information, if present, to provide source-level
4155 descriptions of data addresses in the error messages they create.
4157 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4158 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).