2 Release 3.16.0 (?? ?????? 2020)
3 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
5 3.16.0 is a feature release with many improvements and the usual collection of
8 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
9 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
10 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
11 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
12 support for X86/macOS 10.13, AMD64/macOS 10.13 and nanoMIPS/Linux.
14 * ==================== CORE CHANGES ===================
16 * It is now possible to dynamically change the value of many command
17 line options while your program (or its children) are running under
19 To have the list of dynamically changeable options, run
20 "valgrind --help-dyn-options".
21 You can change the options from the shell by using vgdb to launch
22 the monitor command "v.clo <clo option>...".
23 The same monitor command can be used from a gdb connected
24 to the valgrind gdbserver.
25 Your program can also change the dynamically changeable options using
26 the client request VALGRIND_CLO_CHANGE(option).
28 * ================== PLATFORM CHANGES =================
30 * mips: preliminary support for nanoMIPS instruction set has been added.
32 * ==================== TOOL CHANGES ====================
38 - cg_annotate's --auto and --show-percs options now default to 'yes', because
39 they are usually wanted.
43 - callgrind_annotate's --auto and --show-percs options now default to 'yes',
44 because they are usually wanted.
46 - The command option --collect-systime has been enhanced to specify
47 the unit used to record the elapsed time spent during system calls.
48 The command option now accepts the values no|yes|msec|usec|nsec,
49 where yes is a synonym of msec. When giving the value nsec, the
50 system cpu time of system calls is also recorded.
56 - Several memcheck options are now dynamically changeable.
57 Use valgrind --help-dyn-options to list them.
59 - The release 3.15 introduced a backward incompatible change for
60 some suppression entries related to preadv and pwritev syscalls.
61 When reading a suppression entry using the unsupported 3.14 format,
62 valgrind will now produce a warning to say the suppression entry will not
63 work, and suggest the needed change.
67 - The exprimental Stack and Global Array Checking tool has been removed.
68 It only ever worked on x86 and amd64, and even on those it had a
69 high false positive rate and was slow. An alternative for detecting
70 stack and global array overruns is using the AddressSanitizer (ASAN)
71 facility of the GCC and llvm compilers which require you to rebuild
72 your code with -fsanitize=address.
74 * ==================== OTHER CHANGES ====================
76 * New and modified GDB server monitor features:
78 - Option -T tells vgdb to output a timestamp in the vgdb information messages.
80 - The gdbserver monitor commands that require an address and an optional
81 length argument now accepts the alternate 'C like' syntax "address[length]".
82 For example, the memcheck command "monitor who_points_at 0x12345678 120"
83 can now also be given as "monitor who_points_at 0x12345678[120]".
85 * ==================== FIXED BUGS ====================
87 The following bugs have been fixed or resolved. Note that "n-i-bz"
88 stands for "not in bugzilla" -- that is, a bug that was reported to us
89 but never got a bugzilla entry. We encourage you to file bugs in
90 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
91 than mailing the developers (or mailing lists) directly -- bugs that
92 are not entered into bugzilla tend to get forgotten about or ignored.
94 To see details of a given bug, visit
95 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
96 where XXXXXX is the bug number as listed below.
98 343099 Linux setns syscall wrapper missing, unhandled syscall: 308
99 == 368923 WARNING: unhandled arm64-linux syscall: 268 (setns)
100 == 369031 WARNING: unhandled amd64-linux syscall: 308 (setns)
101 385386 Assertion failed "szB >= CACHE_ENTRY_SIZE" at m_debuginfo/image.c:517
102 400162 Patch: Guard against __GLIBC_PREREQ for musl libc
103 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
104 400872 Add nanoMIPS support to Valgrind
105 403212 drd/tests/trylock hangs on FreeBSD
106 404406 s390x: z14 miscellaneous instructions not implemented
107 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
108 406561 mcinfcallWSRU gdbserver_test fails on ppc64
109 406824 Unsupported baseline
110 407218 Add support for the copy_file_range syscall
111 407307 Intercept stpcpy also in ld.so for arm64
112 407376 Update Xen support to 4.12 (4.13, actually) and add more coverage
114 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
115 408009 Expose rdrand and f16c even on avx if host cpu supports them
116 408091 Missing pkey syscalls
117 408414 Add support for missing for preadv2 and pwritev2 syscalls
118 409141 Valgrind hangs when SIGKILLed
119 409206 Support for Linux PPS and PTP ioctls
120 409367 exit_group() after signal to thread waiting in futex() causes hangs
121 409429 amd64: recognize 'cmpeq' variants as a dependency breaking idiom
122 409780 References to non-existent configure.in
123 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
124 410599 Non-deterministic behaviour of pth_self_kill_15_other test
125 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
126 411134 Allow the user to change a set of command line options during execution
127 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
128 412344 Problem setting mips flags with specific paths
129 412408 unhandled arm-linux syscall: 124 - adjtime - on arm-linux
130 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
131 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
132 413603 callgrind_annotate/cg_annotate truncate function names at '#'
133 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
134 415136 ARMv8.1 Compare-and-Swap instructions are not supported
135 415757 vex x86->IR: 0x66 0xF 0xCE 0x4F (bswapw)
136 416239 valgrind crashes when handling clock_adjtime
137 416285 Use prlimit64 in VG_(getrlimit) and VG_(setrlimit)
138 416286 DRD reports "conflicting load" error on std::mutex::lock()
139 416301 s390x: "compare and signal" not supported
140 416387 finit_module and bpf syscalls are unhandled on arm64
141 416464 Fix false reports for uninitialized memory for PR_CAPBSET_READ/DROP
142 416667 gcc10 ppc64le impossible constraint in 'asm' in test_isa.
143 416753 new 32bit time syscalls for 2038+
144 417075 pwritev(vector[...]) suppression ignored
145 417075 is not fixed, but incompatible supp entries are detected
146 and a warning is produced for these.
147 417187 [MIPS] Conditional branch problem since 'grail' changes
148 417238 Test memcheck/tests/vbit-test fails on mips64 BE
149 417266 Make memcheck/tests/linux/sigqueue usable with musl
150 417281 s390x: /bin/true segfaults with "grail" enabled
151 417427 commit to fix vki_siginfo_t definition created numerous regression
153 417452 s390_insn_store_emit: dst->tag for HRcVec128
154 417578 Add suppressions for glibc DTV leaks
155 417906 clone with CLONE_VFORK and no CLONE_VM fails
156 418004 Grail code additions break ppc64.
157 418435 s390x: spurious "Conditional jump or move depends on uninitialised [..]"
158 418997 s390x: Support Iex_ITE for float and vector types
159 419503 s390x: Avoid modifying registers returned from isel functions
160 n-i-bz Fix minor one time leaks in dhat.
161 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
162 n-i-bz Add support for the Linux io_uring system calls
163 n-i-bz sys_statx: don't complain if both |filename| and |buf| are NULL.
164 n-i-bz Fix non-glibc build of test suite with s390x_features
165 n-i-bz MinGW, include/valgrind.h: Fix detection of 64-bit mode
168 Release 3.15.0 (12 April 2019)
169 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
171 3.15.0 is a feature release with many improvements and the usual collection of
174 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
175 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
176 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
177 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
178 support for X86/macOS 10.13 and AMD64/macOS 10.13.
180 * ==================== CORE CHANGES ===================
182 * The XTree Massif output format now makes use of the information obtained
183 when specifying --read-inline-info=yes.
185 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
187 * ==================== TOOL CHANGES ====================
191 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
192 it has been promoted from an experimental tool to a regular tool. Run it
193 with --tool=dhat instead of --tool=exp-dhat.
195 - DHAT now prints only minimal data when the program ends, instead writing
196 the bulk of the profiling data to a file. As a result, the --show-top-n
197 and --sort-by options have been removed.
199 - Profile results can be viewed with the new viewer, dh_view.html. When
200 a run ends, a short message is printed, explaining how to view the result.
202 - See the documentation for more details.
206 - cg_annotate has a new option, --show-percs, which prints percentages next
211 - callgrind_annotate has a new option, --show-percs, which prints percentages
212 next to all event counts.
214 - callgrind_annotate now inserts commas in call counts, and
215 sort the caller/callee lists in the call tree.
219 - The default value for --read-inline-info is now "yes" on
220 Linux/Android/Solaris. It is still "no" on other OS.
224 - The option --xtree-leak=yes (to output leak result in xtree format)
225 automatically activates the option --show-leak-kinds=all, as xtree
226 visualisation tools such as kcachegrind can in any case select what kind
227 of leak to visualise.
229 - There has been further work to avoid false positives. In particular,
230 integer equality on partially defined inputs (C == and !=) is now handled
233 * ==================== OTHER CHANGES ====================
235 * The new option --show-error-list=no|yes displays, at the end of the run, the
236 list of detected errors and the used suppressions. Prior to this change,
237 showing this information could only be done by specifying "-v -v", but that
238 also produced a lot of other possibly-non-useful messages. The option -s is
239 equivalent to --show-error-list=yes.
241 * ==================== FIXED BUGS ====================
243 The following bugs have been fixed or resolved. Note that "n-i-bz"
244 stands for "not in bugzilla" -- that is, a bug that was reported to us
245 but never got a bugzilla entry. We encourage you to file bugs in
246 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
247 than mailing the developers (or mailing lists) directly -- bugs that
248 are not entered into bugzilla tend to get forgotten about or ignored.
250 To see details of a given bug, visit
251 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
252 where XXXXXX is the bug number as listed below.
254 385411 s390x: z13 vector floating-point instructions not implemented
255 397187 z13 vector register support for vgdb gdbserver
256 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
257 398870 Please add support for instruction vcvtps2ph
258 399287 amd64 front end: Illegal Instruction vcmptrueps
259 399301 Use inlined frames in Massif XTree output.
260 399322 Improve callgrind_annotate output
261 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
262 400164 helgrind test encounters mips x-compiler warnings and assembler error
263 400490 s390x: VRs allocated as if separate from FPRs
264 400491 s390x: Operand of LOCH treated as unsigned integer
265 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
266 options, which specify a mips64 processor
267 401112 LLVM 5.0 generates comparison against partially initialized data
268 401277 More bugs in z13 support
269 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
270 401578 drd: crashes sometimes on fork()
271 401627 memcheck errors with glibc avx2 optimized wcsncmp
272 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
273 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
274 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
276 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
277 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
278 402123 invalid assembler opcodes for mips32r2
279 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
280 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
281 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
282 402351 mips64 libvexmultiarch_test fails on s390x
284 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
285 402480 Do not use %rsp in clobber list
286 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
287 402515 Implement new option --show-error-list=no|yes / -s
288 402519 POWER 3.0 addex instruction incorrectly implemented
289 402781 Redo the cache used to process indirect branch targets
290 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
291 403552 s390x: wrong facility bit checked for vector facility
292 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
293 404638 Add VG_(replaceIndexXA)
294 404843 s390x: backtrace sometimes ends prematurely
295 404888 autotools cleanup series
296 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
297 405182 Valgrind fails to build with Clang
298 405205 filter_libc: remove the line holding the futex syscall error entirely
299 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
300 the upper and lower 32-bits of the 64-bit result
301 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
302 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
303 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
305 405403 s390x disassembler cannot be used on x86
306 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
307 405458 MIPS mkFormVEC arguments swapped?
308 405716 drd: Fix an integer overflow in the stack margin calculation
309 405722 Support arm64 core dump
310 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
311 of the 64-bit destination field.
312 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
313 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
314 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
316 406256 PPC64, vector floating point instructions don't handle subnormal
317 according to VSCR[NJ] bit setting.
318 406352 cachegrind/callgrind fails ann tests because of missing a.c
319 406354 dhat is broken on x86 (32bit)
320 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
321 406357 gdbserver_tests fails because of gdb output change
322 406360 memcheck/tests/libstdc++.supp needs more supression variants
323 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
324 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
325 407340 PPC64, does not support the vlogefp, vexptefp instructions.
326 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
327 n-i-bz Fix callgrind_annotate non deterministic order for equal total
328 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
329 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
330 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
332 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
333 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
334 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
338 Release 3.14.0 (9 October 2018)
339 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
341 3.14.0 is a feature release with many improvements and the usual collection of
344 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
345 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
346 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
347 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
348 support for X86/macOS 10.13, AMD64/macOS 10.13.
350 * ==================== CORE CHANGES ===================
352 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
353 debug info for unloaded code. This allows saved stack traces (e.g. for
354 memory leaks) to include file/line info for code that has been dlclose'd (or
355 similar). See the user manual for more information and known limitations.
357 * Ability to specify suppressions based on source file name and line number.
359 * Majorly overhauled register allocator. No end-user changes, but the JIT
360 generates code a bit more quickly now.
362 * ================== PLATFORM CHANGES =================
364 * Preliminary support for macOS 10.13 has been added.
366 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
368 * mips: support for MIPS SIMD architecture (MSA) has been added.
370 * mips: support for MIPS N32 ABI has been added.
372 * s390: partial support for vector instructions (integer and string) has been
375 * ==================== TOOL CHANGES ====================
377 * Helgrind: Addition of a flag
378 --delta-stacktrace=no|yes [yes on linux amd64/x86]
379 which specifies how full history stack traces should be computed.
380 Setting this to =yes can speed up Helgrind by 25% when using
381 --history-level=full.
383 * Memcheck: reduced false positive rate for optimised code created by Clang 6
384 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
385 blocks more carefully to determine where it can avoid expensive definedness
386 checks without loss of precision. This is controlled by the flag
387 --expensive-definedness-checks=no|auto|yes [auto].
389 * ==================== OTHER CHANGES ====================
391 * Valgrind is now buildable with link-time optimisation (LTO). A new
392 configure option --enable-lto=yes allows building Valgrind with LTO. If the
393 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
394 Note that if you are doing Valgrind development, --enable-lto=yes massively
395 slows down the build process.
397 * ==================== FIXED BUGS ====================
399 The following bugs have been fixed or resolved. Note that "n-i-bz"
400 stands for "not in bugzilla" -- that is, a bug that was reported to us
401 but never got a bugzilla entry. We encourage you to file bugs in
402 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
403 than mailing the developers (or mailing lists) directly -- bugs that
404 are not entered into bugzilla tend to get forgotten about or ignored.
406 To see details of a given bug, visit
407 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
408 where XXXXXX is the bug number as listed below.
410 79362 Debug info is lost for .so files when they are dlclose'd
411 208052 strlcpy error when n = 0
412 255603 exp-sgcheck Assertion '!already_present' failed
413 338252 building valgrind with -flto (link time optimisation) fails
414 345763 MIPS N32 ABI support
415 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
416 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
417 372347 Replacement problem of the additional c++14/c++17 new/delete operators
418 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
419 376257 helgrind history full speed up using a cached stack
420 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
422 379748 Fix missing pselect syscall (OS X 10.11)
423 379754 Fix missing syscall ulock_wait (OS X 10.12)
424 380397 s390x: __GI_strcspn() replacemenet needed
425 381162 possible array overrun in VEX register allocator
426 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
427 381274 powerpc too chatty even with --sigill-diagnostics=no
428 381289 epoll_pwait can have a NULL sigmask
429 381553 VEX register allocator v3
430 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
431 381769 Use ucontext_t instead of struct ucontext
432 381805 arm32 needs ld.so index hardwire for new glibc security fixes
433 382256 gz compiler flag test doesn't work for gold
434 382407 vg_perf needs "--terse" command line option
435 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
436 382563 MIPS MSA ASE support
437 382998 xml-socket doesn't work
438 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
439 383723 Fix missing kevent_qos syscall (macOS 10.11)
440 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
441 384096 Mention AddrCheck at Memcheck's command line option [..]
442 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
443 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
444 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
445 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
446 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
447 384337 performance improvements to VEX register allocator v2 and v3
448 384526 reduce number of spill insns generated by VEX register allocator v3
449 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
450 384631 Sanitise client args as printed with -v
451 384633 Add a simple progress-reporting facility
452 384987 VEX regalloc: allocate caller-save registers for short lived vregs
453 385055 PPC VEX temporary storage exhausted
454 385182 PPC64 is missing support for the DSCR
455 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
456 385207 PPC64, generate_store_FPRF() generates too many Iops
457 385208 PPC64, xxperm instruction exhausts temporary memory
458 385210 PPC64, vpermr instruction could exhaust temporary memory
459 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
460 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
461 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
462 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
463 385408 s390x: z13 vector "support" instructions not implemented
464 385409 s390x: z13 vector integer instructions not implemented
465 385410 s390x: z13 vector string instructions not implemented
466 385412 s390x: new non-vector z13 instructions not implemented
467 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
468 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
469 385939 Optionally exit on the first error
470 386318 valgrind.org/info/tools.html is missing SGCheck
471 386425 running valgrind + wine on armv7l gives illegal opcode
472 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
473 387410 MIPSr6 support
474 387664 Memcheck: make expensive-definedness-checks be the default
475 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
476 387766 asm shifts cause false positive "Conditional jump or move depends
477 on uninitialised value"
478 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
479 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
480 388786 Support bpf syscall in amd64 Linux
481 388862 Add replacements for wmemchr and wcsnlen on Linux
482 389065 valgrind meets gcc flag -Wlogical-op
483 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
484 390471 suppression by specification of source-file line number
485 390723 make xtree dump files world wide readable, similar to log files
486 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
487 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
488 392118 unhandled amd64-linux syscall: 332 (statx)
489 392449 callgrind not clearing the number of calls properly
490 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
491 lxssp, stxssp and stxvl instructions.
492 393023 callgrind_control risks using the wrong vgdb
493 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
494 393099 posix_memalign() invalid write if alignment == 0
495 393146 failing assert "is_DebugInfo_active(di)"
496 395709 PPC64 is missing support for the xvnegsp instruction
497 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
499 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
500 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
501 396839 s390x: Trap instructions not implemented
502 396887 arch_prctl should return EINVAL on unknown option
503 == 397286 crash before launching binary (Unsupported arch_prctl option)
504 == 397393 valgrind: the 'impossible' happened: (Archlinux)
505 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
506 396906 compile tests failure on mips32-linux: broken inline asm in tests on
508 397012 glibc ld.so uses arch_prctl on i386
509 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
510 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
511 397424 glibc 2.27 and gdb_server tests
512 398028 Assertion `cfsi_fits` failing in simple C program
513 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
515 n-i-bz Fix missing workq_ops operations (macOS)
516 n-i-bz fix bug in strspn replacement
517 n-i-bz Add support for the Linux BLKFLSBUF ioctl
518 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
519 n-i-bz Fix possible stack trashing by semctl syscall wrapping
520 n-i-bz Add support for the Linux membarrier() system call
521 n-i-bz x86 front end: recognise and handle UD2 correctly
522 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
523 correctly aligned before entering the handler.
525 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
526 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
527 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
531 Release 3.13.0 (15 June 2017)
532 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
534 3.13.0 is a feature release with many improvements and the usual collection of
537 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
538 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
539 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
540 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
542 * ==================== CORE CHANGES ===================
544 * The translation cache size has been increased to keep up with the demands of
545 large applications. The maximum number of sectors has increased from 24 to
546 48. The default number of sectors has increased from 16 to 32 on all
547 targets except Android, where the increase is from 6 to 12.
549 * The amount of memory that Valgrind can use has been increased from 64GB to
550 128GB. In particular this means your application can allocate up to about
551 60GB when running on Memcheck.
553 * Valgrind's default load address has been changed from 0x3800'0000 to
554 0x5800'0000, so as to make it possible to load larger executables. This
555 should make it possible to load executables of size at least 1200MB.
557 * A massive spaceleak caused by reading compressed debuginfo files has been
558 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
561 * The C++ demangler has been updated.
563 * Support for demangling Rust symbols has been added.
565 * A new representation of stack traces, the "XTree", has been added. An XTree
566 is a tree of stacktraces with data associated with the stacktraces. This is
567 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
568 consumption of your program. Reporting is controlled by the new options
569 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
571 A report can also be produced on demand using the gdbserver monitor command
572 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
573 format' and 'massif format. The existing visualisers for these formats (e.g.
574 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
575 analyse these reports.
577 Memcheck can also produce XTree leak reports using the Callgrind file
578 format. For more details, see the user manual.
580 * ================== PLATFORM CHANGES =================
582 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
584 * amd64: fixes for JIT failure problems on long AVX2 code blocks
586 * amd64 and x86: support for CET prefixes has been added
588 * arm32: a few missing ARMv8 instructions have been implemented
590 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
591 Store-Conditional instructions has been added. This is to deal with
592 processor implementations that implement the LL/SC specifications strictly
593 and as a result cause Valgrind to hang in certain situations. The
594 alternative implementation is automatically enabled at startup, as required.
595 You can use the option --sim-hints=fallback-llsc to force-enable it if you
598 * Support for OSX 10.12 has been improved.
600 * On Linux, clone handling has been improved to honour CLONE_VFORK that
601 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
602 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
603 CLONE_VM exact semantics will (still) not work.
605 * The TileGX/Linux port has been removed because it appears to be both unused
608 * ==================== TOOL CHANGES ====================
612 - Memcheck should give fewer false positives when running optimised
613 Clang/LLVM generated code.
615 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
617 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
618 to produce the end of execution leak report in a xtree callgrind format
621 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
622 the leak report in an xtree file.
626 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
628 - For some workloads (typically, for big applications), Massif memory
629 consumption and CPU consumption has decreased significantly.
633 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
635 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
636 for Ada gnat compiled applications.
638 * ==================== OTHER CHANGES ====================
640 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
641 append the inner guest stacktrace to the inner host stacktrace. This helps
642 to investigate the errors reported by the outer, when they are caused by the
643 inner guest program (such as an inner regtest). See README_DEVELOPERS for
646 * To allow fast detection of callgrind files by desktop environments and file
647 managers, the format was extended to have an optional first line that
648 uniquely identifies the format ("# callgrind format"). Callgrind creates
649 this line now, as does the new xtree functionality.
651 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
652 have a new %n format letter that is replaced by a sequence number.
654 * "--version -v" now shows the SVN revision numbers from which Valgrind was
657 * ==================== FIXED BUGS ====================
659 The following bugs have been fixed or resolved. Note that "n-i-bz"
660 stands for "not in bugzilla" -- that is, a bug that was reported to us
661 but never got a bugzilla entry. We encourage you to file bugs in
662 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
663 than mailing the developers (or mailing lists) directly -- bugs that
664 are not entered into bugzilla tend to get forgotten about or ignored.
666 To see details of a given bug, visit
667 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
668 where XXXXXX is the bug number as listed below.
670 162848 --log-file output isn't split when a program forks
671 340777 Illegal instruction on mips (ar71xx)
672 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
673 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
674 to a different stack.
675 344139 x86 stack-seg overrides, needed by the Wine people
676 344524 store conditional of guest applications always fail - observed on
678 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
679 352395 Please provide SVN revision info in --version -v
680 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
681 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
682 INVALID_POSIX_THREADID' failed
683 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
684 barrier implementation
685 358697 valgrind.h: Some code remains even when defining NVALGRIND
686 359202 Add musl libc configure/compile
687 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
688 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
689 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
690 362223 assertion failed when .valgrindrc is a directory instead of a file
691 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
692 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
693 368507 can't malloc chunks larger than about 34GB
694 368529 Android arm target link error, missing atexit and pthread_atfork
695 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
696 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
697 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
698 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
699 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
700 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
701 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
702 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
703 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
704 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
705 370028 Reduce the number of compiler warnings on MIPS platforms
706 370635 arm64 missing syscall getcpu
707 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
708 371227 Clean AArch64 syscall table
709 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
710 371471 Valgrind complains about non legit memory leaks on placement new (C++)
711 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
712 371503 disInstr(arm64): unhandled instruction 0xF89F0000
713 371869 support '%' in symbol Z-encoding
714 371916 execution tree xtree concept
715 372120 c++ demangler demangles symbols which are not c++
716 372185 Support of valgrind on ARMv8 with 32 bit executable
717 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
718 372195 Power PC, xxsel instruction is not always recognized.
719 372504 Hanging on exit_group
720 372600 process loops forever when fatal signals are arriving quickly
721 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
722 373046 Stacks registered by core are never deregistered
723 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
724 373086 Implement additional Xen hypercalls
725 373192 Calling posix_spawn in glibc 2.24 completely broken
726 373488 Support for fanotify API on ARM64 architecture
727 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
728 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
729 373938 const IRExpr arguments for matchIRExpr()
730 374719 some spelling fixes
731 374963 increase valgrind's load address to prevent mmap failure
732 375514 valgrind_get_tls_addr() does not work in case of static TLS
733 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
734 for ML_(find_rx_mapping)()
735 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
736 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
737 == 377159 "vex: the `impossible' happened" still present
738 == 375150 Assertion 'tres.status == VexTransOK' failed
739 == 378068 valgrind crashes on AVX2 function in FFmpeg
740 376142 Segfaults on MIPS Cavium Octeon boards
741 376279 disInstr(arm64): unhandled instruction 0xD50320FF
742 376455 Solaris: unhandled syscall lgrpsys(180)
743 376518 Solaris: unhandled fast trap getlgrp(6)
744 376611 ppc64 and arm64 don't know about prlimit64 syscall
745 376729 PPC64, remove R2 from the clobber list
747 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
748 to be wrongly marked as addressable
749 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
750 PIE enabled by default
751 377376 memcheck/tests/linux/getregset fails with glibc2.24
752 377427 PPC64, lxv instruction failing on odd destination register
753 377478 PPC64: ISA 3.0 setup fixes
754 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
755 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
756 and 2 args for FUTEX_TRYLOCK_PI
757 377717 Fix massive space leak when reading compressed debuginfo sections
758 377891 Update Xen 4.6 domctl wrappers
759 377930 fcntl syscall wrapper is missing flock structure check
760 378524 libvexmultiarch_test regression on s390x and ppc64
761 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
762 378673 Update libiberty demangler
763 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
764 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
765 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
766 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
767 (task_register_dyld_image_infos)
768 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
769 (task_register_dyld_shared_cache_image_info)
770 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
771 379473 MIPS: add support for rdhwr cycle counter register
772 379504 remove TileGX/Linux port
773 379525 Support more x86 nop opcodes
774 379838 disAMode(x86): not an addr!
775 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
776 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
777 379895 clock_gettime does not execute POST syscall wrapper
778 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
779 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
780 380200 xtree generated callgrind files refer to files without directory name
781 380202 Assertion failure for cache line size (cls == 64) on aarch64.
782 380397 s390x: __GI_strcspn() replacement needed
783 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
785 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
786 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
787 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
791 Release 3.12.0 (20 October 2016)
792 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
794 3.12.0 is a feature release with many improvements and the usual
795 collection of bug fixes.
797 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
798 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
799 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
800 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
801 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
802 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
804 * ================== PLATFORM CHANGES =================
806 * POWER: Support for ISA 3.0 has been added
808 * mips: support for O32 FPXX ABI has been added.
809 * mips: improved recognition of different processors
810 * mips: determination of page size now done at run time
812 * amd64: Partial support for AMD FMA4 instructions.
814 * arm, arm64: Support for v8 crypto and CRC instructions.
816 * Improvements and robustification of the Solaris port.
818 * Preliminary support for MacOS 10.12 (Sierra) has been added.
820 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
821 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
822 where possible. Valgrind's support for 32-bit x86 has stagnated in
823 recent years and has fallen far behind that for 64-bit x86
824 instructions. By contrast 64-bit x86 is well supported, up to and
827 * ==================== TOOL CHANGES ====================
831 - Added meta mempool support for describing a custom allocator which:
832 - Auto-frees all chunks assuming that destroying a pool destroys all
834 - Uses itself to allocate other memory blocks
836 - New flag --ignore-range-below-sp to ignore memory accesses below
837 the stack pointer, if you really have to. The related flag
838 --workaround-gcc296-bugs=yes is now deprecated. Use
839 --ignore-range-below-sp=1024-1 as a replacement.
843 - Improved thread startup time significantly on non-Linux platforms.
847 - Added collection of the metric "tot-blocks-allocd"
849 * ==================== OTHER CHANGES ====================
851 * Replacement/wrapping of malloc/new related functions is now done not just
852 for system libraries by default, but for any globally defined malloc/new
853 related function (both in shared libraries and statically linked alternative
854 malloc implementations). The dynamic (runtime) linker is excluded, though.
855 To only intercept malloc/new related functions in
856 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
857 "nouserintercepts" can be any non-existing library name).
858 This new functionality is not implemented for MacOS X.
860 * The maximum number of callers in a suppression entry is now equal to
861 the maximum size for --num-callers (500).
862 Note that --gen-suppressions=yes|all similarly generates suppressions
863 containing up to --num-callers frames.
865 * New and modified GDB server monitor features:
867 - Valgrind's gdbserver now accepts the command 'catch syscall'.
868 Note that you must have GDB >= 7.11 to use 'catch syscall' with
871 * New option --run-cxx-freeres=<yes|no> can be used to change whether
872 __gnu_cxx::__freeres() cleanup function is called or not. Default is
875 * Valgrind is able to read compressed debuginfo sections in two formats:
876 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
877 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
879 * Modest JIT-cost improvements: the cost of instrumenting code blocks
880 for the most common use case (x86_64-linux, Memcheck) has been
883 * Improved performance for programs that do a lot of discarding of
884 instruction address ranges of 8KB or less.
886 * The C++ symbol demangler has been updated.
888 * More robustness against invalid syscall parameters on Linux.
890 * ==================== FIXED BUGS ====================
892 The following bugs have been fixed or resolved. Note that "n-i-bz"
893 stands for "not in bugzilla" -- that is, a bug that was reported to us
894 but never got a bugzilla entry. We encourage you to file bugs in
895 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
896 than mailing the developers (or mailing lists) directly -- bugs that
897 are not entered into bugzilla tend to get forgotten about or ignored.
899 To see details of a given bug, visit
900 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
901 where XXXXXX is the bug number as listed below.
903 191069 Exiting due to signal not reported in XML output
904 199468 Suppressions: stack size limited to 25
905 while --num-callers allows more frames
906 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
907 278744 cvtps2pd with redundant RexW
908 303877 valgrind doesn't support compressed debuginfo sections.
909 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
910 348345 Assertion fails for negative lineno
911 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
912 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
913 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
914 351804 Crash on generating suppressions for "printf" call on OS X 10.10
915 352197 mips: mmap2() not wrapped correctly for page size > 4096
916 353083 arm64 doesn't implement various xattr system calls
917 353084 arm64 doesn't support sigpending system call
918 353137 www: update info for Supported Platforms
919 353138 www: update "The Valgrind Developers" page
920 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
923 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
924 353398 WARNING: unhandled amd64-solaris syscall: 207
925 353660 XML in auxwhat tag not escaping reserved symbols properly
926 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
927 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
928 353802 ELF debug info reader confused with multiple .rodata sections
929 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
930 353917 unhandled amd64-solaris syscall fchdir(120)
931 353920 unhandled amd64-solaris syscall: 170
932 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
933 354392 unhandled amd64-solaris syscall: 171
934 354797 Vbit test does not include Iops for Power 8 instruction support
935 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
939 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
940 355188 valgrind should intercept all malloc related global functions
941 355454 do not intercept malloc related symbols from the runtime linker
942 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
943 356044 Dwarf line info reader misinterprets is_stmt register
944 356112 mips: replace addi with addiu
945 356393 valgrind (vex) crashes because isZeroU happened
948 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
949 356678 arm64-linux: unhandled syscall 232 (mincore)
950 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
951 356823 Unsupported ARM instruction: stlex
952 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
953 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
954 357673 crash if I try to run valgrind with a binary link with libcurl
955 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
956 357871 pthread_spin_destroy not properly wrapped
957 357887 Calls to VG_(fclose) do not close the file descriptor
958 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
959 358030 support direct socket calls on x86 32bit (new in linux 4.3)
960 358478 drd/tests/std_thread.cpp doesn't build with GCC6
961 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
962 359181 Buffer Overflow during Demangling
963 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
964 359289 s390x: popcnt (B9E1) not implemented
965 359472 The Power PC vsubuqm instruction doesn't always give the correct result
966 359503 Add missing syscalls for aarch64 (arm64)
967 359645 "You need libc6-dbg" help message could be more helpful
968 359703 s390: wire up separate socketcalls system calls
969 359724 getsockname might crash - deref_UInt should call safe_to_deref
970 359733 amd64 implement ld.so strchr/index override like x86
971 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
972 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
974 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
975 359871 Incorrect mask handling in ppoll
976 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
977 360008 Contents of Power vr registers contents is not printed correctly when
978 the --vgdb-shadow-registers=yes option is used
979 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
981 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
982 360425 arm64 unsupported instruction ldpsw
984 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
985 360571 Error about the Android Runtime reading below the stack pointer on ARM
986 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
987 360749 kludge for multiple .rodata sections on Solaris no longer needed
988 360752 raise the number of reserved fds in m_main.c from 10 to 12
989 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
990 361226 s390x: risbgn (EC59) not implemented
991 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
992 361354 ppc64[le]: wire up separate socketcalls system calls
993 361615 Inconsistent termination for multithreaded process terminated by signal
994 361926 Unhandled Solaris syscall: sysfs(84)
995 362009 V dumps core on unimplemented functionality before threads are created
996 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
997 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
998 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
999 362953 Request for an update to the Valgrind Developers page
1000 363680 add renameat2() support
1001 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
1002 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
1003 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
1004 364058 clarify in manual limitations of array overruns detections
1005 364413 pselect sycallwrapper mishandles NULL sigmask
1006 364728 Power PC, missing support for several HW registers in
1007 get_otrack_shadow_offset_wrk()
1008 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
1009 365273 Invalid write to stack location reported after signal handler runs
1010 365912 ppc64BE segfault during jm-insns test (RELRO)
1011 366079 FPXX Support for MIPS32 Valgrind
1012 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
1013 366344 Multiple unhandled instruction for Aarch64
1014 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
1015 367995 Integration of memcheck with custom memory allocator
1016 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
1017 368412 False positive result for altivec capability check
1018 368416 Add tc06_two_races_xml.exp output for ppc64
1019 368419 Perf Events ioctls not implemented
1020 368461 mmapunmap test fails on ppc64
1021 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
1022 369000 AMD64 fma4 instructions unsupported.
1023 369169 ppc64 fails jm_int_isa_2_07 test
1024 369175 jm_vec_isa_2_07 test crashes on ppc64
1025 369209 valgrind loops and eats up all memory if cwd doesn't exist.
1026 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
1027 369359 msghdr_foreachfield can crash when handling bad iovec
1028 369360 Bad sigprocmask old or new sets can crash valgrind
1029 369361 vmsplice syscall wrapper crashes on bad iovec
1030 369362 Bad sigaction arguments crash valgrind
1031 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
1032 369402 Bad set/get_thread_area pointer crashes valgrind
1033 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
1034 369446 valgrind crashes on unknown fcntl command
1035 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
1036 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
1037 370265 ISA 3.0 HW cap stuff needs updating
1038 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
1039 372195 Power PC, xxsel instruction is not always recognized
1041 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
1042 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
1043 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
1044 n-i-bz Fix ppoll_alarm exclusion on OS X
1045 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
1046 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
1047 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
1048 n-i-bz amd64: memcheck false positive with shr %edx
1049 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
1050 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
1051 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
1052 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
1054 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
1055 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
1056 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
1060 Release 3.11.0 (22 September 2015)
1061 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1063 3.11.0 is a feature release with many improvements and the usual
1064 collection of bug fixes.
1066 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1067 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1068 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1069 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1070 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1071 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1073 * ================== PLATFORM CHANGES =================
1075 * Support for Solaris/x86 and Solaris/amd64 has been added.
1077 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1079 * Preliminary support for the Tilera TileGX architecture has been added.
1081 * s390x: It is now required for the host to have the "long displacement"
1082 facility. The oldest supported machine model is z990.
1084 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1085 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1087 * The JIT's register allocator is significantly faster, making the JIT
1088 as a whole somewhat faster, so JIT-intensive activities, for example
1089 program startup, are modestly faster, around 5%.
1091 * There have been changes to the default settings of several command
1092 line flags, as detailed below.
1094 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1095 capable hosts, the simulated CPUID will now indicate AVX2 support.
1097 * ==================== TOOL CHANGES ====================
1101 - The default value for --leak-check-heuristics has been changed from
1102 "none" to "all". This helps to reduce the number of possibly
1103 lost blocks, in particular for C++ applications.
1105 - The default value for --keep-stacktraces has been changed from
1106 "malloc-then-free" to "malloc-and-free". This has a small cost in
1107 memory (one word per malloc-ed block) but allows Memcheck to show the
1108 3 stacktraces of a dangling reference: where the block was allocated,
1109 where it was freed, and where it is acccessed after being freed.
1111 - The default value for --partial-loads-ok has been changed from "no" to
1112 "yes", so as to avoid false positive errors resulting from some kinds
1113 of vectorised loops.
1115 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1116 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1117 than get_vbits when you need to associate byte data value with
1118 their corresponding validity bits.
1120 - The 'block_list' monitor command has been enhanced:
1121 o it can print a range of loss records
1122 o it now accepts an optional argument 'limited <max_blocks>'
1123 to control the number of blocks printed.
1124 o if a block has been found using a heuristic, then
1125 'block_list' now shows the heuristic after the block size.
1126 o the loss records/blocks to print can be limited to the blocks
1127 found via specified heuristics.
1129 - The C helper functions used to instrument loads on
1130 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1131 replaced by handwritten assembly sequences. This gives speedups
1132 in the region of 0% to 7% for those targets only.
1134 - A new command line option, --expensive-definedness-checks=yes|no,
1135 has been added. This is useful for avoiding occasional invalid
1136 uninitialised-value errors in optimised code. Watch out for
1137 runtime degradation, as this can be up to 25%. As always, though,
1138 the slowdown is highly application specific. The default setting
1143 - A new monitor command 'all_snapshots <filename>' dumps all
1144 snapshots taken so far.
1148 - Significant memory reduction and moderate speedups for
1149 --history-level=full for applications accessing a lot of memory
1150 with many different stacktraces.
1152 - The default value for --conflict-cache-size=N has been doubled to
1153 2000000. Users that were not using the default value should
1154 preferably also double the value they give.
1156 The default was changed due to the changes in the "full history"
1157 implementation. Doubling the value gives on average a slightly more
1158 complete history and uses similar memory (or significantly less memory
1159 in the worst case) than the previous implementation.
1161 - The Helgrind monitor command 'info locks' now accepts an optional
1162 argument 'lock_addr', which shows information about the lock at the
1165 - When using --history-level=full, the new Helgrind monitor command
1166 'accesshistory <addr> [<len>]' will show the recorded accesses for
1167 <len> (or 1) bytes at <addr>.
1169 * ==================== OTHER CHANGES ====================
1171 * The default value for the --smc-check option has been changed from
1172 "stack" to "all-non-file" on targets that provide automatic D-I
1173 cache coherence (x86, amd64 and s390x). The result is to provide,
1174 by default, transparent support for JIT generated and self-modifying
1175 code on all targets.
1177 * Mac OS X only: the default value for the --dsymutil option has been
1178 changed from "no" to "yes", since any serious usage on Mac OS X
1179 always required it to be "yes".
1181 * The command line options --db-attach and --db-command have been removed.
1182 They were deprecated in 3.10.0.
1184 * When a process dies due to a signal, Valgrind now shows the signal
1185 and the stacktrace at default verbosity (i.e. verbosity 1).
1187 * The address description logic used by Memcheck and Helgrind now
1188 describes addresses in anonymous segments, file mmap-ed segments,
1189 shared memory segments and the brk data segment.
1191 * The new option --error-markers=<begin>,<end> can be used to mark the
1192 begin/end of errors in textual output mode, to facilitate
1193 searching/extracting errors in output files that mix valgrind errors
1194 with program output.
1196 * The new option --max-threads=<number> can be used to change the number
1197 of threads valgrind can handle. The default is 500 threads which
1198 should be more than enough for most applications.
1200 * The new option --valgrind-stacksize=<number> can be used to change the
1201 size of the private thread stacks used by Valgrind. This is useful
1202 for reducing memory use or increasing the stack size if Valgrind
1203 segfaults due to stack overflow.
1205 * The new option --avg-transtab-entry-size=<number> can be used to specify
1206 the expected instrumented block size, either to reduce memory use or
1207 to avoid excessive retranslation.
1209 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1211 * New and modified GDB server monitor features:
1213 - When a signal is reported in GDB, you can now use the GDB convenience
1214 variable $_siginfo to examine detailed signal information.
1216 - Valgrind's gdbserver now allows the user to change the signal
1217 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1218 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1219 continue without passing the signal to the process.
1221 - With GDB >= 7.10, the command 'target remote'
1222 will automatically load the executable file of the process running
1223 under Valgrind. This means you do not need to specify the executable
1224 file yourself, GDB will discover it itself. See GDB documentation about
1225 'qXfer:exec-file:read' packet for more info.
1227 * ==================== FIXED BUGS ====================
1229 The following bugs have been fixed or resolved. Note that "n-i-bz"
1230 stands for "not in bugzilla" -- that is, a bug that was reported to us
1231 but never got a bugzilla entry. We encourage you to file bugs in
1232 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1233 than mailing the developers (or mailing lists) directly -- bugs that
1234 are not entered into bugzilla tend to get forgotten about or ignored.
1236 To see details of a given bug, visit
1237 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1238 where XXXXXX is the bug number as listed below.
1240 116002 VG_(printf): Problems with justification of strings and integers
1241 155125 avoid cutting away file:lineno after long function name
1242 197259 Unsupported arch_prtctl PR_SET_GS option
1243 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
1244 201216 Fix Valgrind does not support pthread_sigmask() on OS X
1245 201435 Fix Darwin: -v does not show kernel version
1246 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
1247 211256 Fixed an outdated comment regarding the default platform.
1248 211529 Incomplete call stacks for code compiled by newer versions of MSVC
1249 211926 Avoid compilation warnings in valgrind.h with -pedantic
1250 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
1252 226609 Crediting upstream authors in man page
1253 231257 Valgrind omits path when executing script from shebang line
1254 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
1255 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
1256 269360 s390x: Fix addressing mode selection for compare-and-swap
1257 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
1259 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
1260 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
1261 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
1262 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
1263 330147 libmpiwrap PMPI_Get_count returns undefined value
1264 333051 mmap of huge pages fails due to incorrect alignment
1266 334802 valgrind does not always explain why a given option is bad
1267 335618 mov.w rN, pc/sp (ARM32)
1268 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
1273 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
1274 338602 AVX2 bit in CPUID missing
1275 338606 Strange message for scripts with invalid interpreter
1276 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
1277 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
1278 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
1280 339156 gdbsrv not called for fatal signal
1281 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
1282 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
1283 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
1284 339442 Fix testsuite build failure on OS X 10.9
1285 339542 Enable compilation with Intel's ICC compiler
1286 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
1287 339688 Mac-specific ASM does not support .version directive (cpuid,
1288 tronical and pushfpopf tests)
1289 339745 Valgrind crash when check Marmalade app (partial fix)
1290 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
1291 339778 Linux/TileGx platform support for Valgrind
1292 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
1293 339789 Fix none/tests/execve test on Mac OS X 10.9
1294 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
1295 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
1296 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
1297 340392 Allow user to select more accurate definedness checking in memcheck
1298 to avoid invalid complaints on optimised code
1299 340430 Fix some grammatical weirdness in the manual.
1300 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
1301 341419 Signal handler ucontext_t not filled out correctly on OS X
1302 341539 VG_(describe_addr) should not describe address as belonging to client
1303 segment if it is past the heap end
1304 341613 Enable building of manythreads and thread-exits tests on Mac OS X
1305 341615 Fix none/tests/darwin/access_extended test on Mac OS X
1306 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
1307 341789 aarch64: shmat fails with valgrind on ARMv8
1308 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
1309 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
1310 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
1311 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
1312 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
1313 342221 socket connect false positive uninit memory for unknown af family
1314 342353 Allow dumping full massif output while valgrind is still running
1315 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
1319 342603 Add I2C_SMBUS ioctl support
1320 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
1321 342683 Mark memory past the initial brk limit as unaddressable
1322 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
1323 342795 Internal glibc __GI_mempcpy call should be intercepted
1324 342841 s390x: Support instructions fiebr(a) and fidbr(a)
1325 343012 Unhandled syscall 319 (memfd_create)
1326 343069 Patch updating v4l2 API support
1327 343173 helgrind crash during stack unwind
1328 343219 fix GET_STARTREGS for arm
1329 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
1330 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
1331 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
1332 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
1333 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
1334 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
1335 343597 ppc64le: incorrect use of offseof macro
1336 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
1337 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
1338 343732 Unhandled syscall 144 (setgid) on aarch64
1339 343733 Unhandled syscall 187 (msgctl and related) on aarch64
1340 343802 s390x: False positive "conditional jump or move depends on [..]
1341 343902 --vgdb=yes doesn't break when --xml=yes is used
1342 343967 Don't warn about setuid/setgid/setcap executable for directories
1343 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
1344 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
1345 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
1346 344054 www - update info for Solaris/illumos
1347 344416 'make regtest' does not work cleanly on OS X
1348 344235 Remove duplicate include of pub_core_aspacemgr.h
1349 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
1350 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
1351 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
1352 344314 callgrind_annotate ... warnings about commands containing newlines
1353 344318 socketcall should wrap recvmmsg and sendmmsg
1354 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
1355 344416 Fix 'make regtest' does not work cleanly on OS X
1356 344499 Fix compilation for Linux kernel >= 4.0.0
1357 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
1358 unix:349 (__pthread_fchdir)
1359 344559 Garbage collection of unused segment names in address space manager
1360 344560 Fix stack traces missing penultimate frame on OS X
1361 344621 Fix memcheck/tests/err_disable4 test on OS X
1362 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
1363 344702 Fix missing libobjc suppressions on OS X 10.10
1365 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
1366 344939 Fix memcheck/tests/xml1 on OS X 10.10
1367 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
1368 345079 Fix build problems in VEX/useful/test_main.c
1369 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
1370 345177 arm64: prfm (reg) not implemented
1371 345215 Performance improvements for the register allocator
1372 345248 add support for Solaris OS in valgrind
1373 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
1374 345394 Fix memcheck/tests/strchr on OS X
1375 345637 Fix memcheck/tests/sendmsg on OS X
1376 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
1377 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
1378 345887 Fix an assertion in the address space manager
1379 345928 amd64: callstack only contains current function for small stacks
1380 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1381 345987 MIPS64: Implement cavium LHX instruction
1382 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1383 346185 Fix typo saving altivec register v24
1384 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1385 and LibVEX_GuestPPC64_get_CR()
1386 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1387 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1388 346307 fuse filesystem syscall deadlocks
1389 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1390 346411 MIPS: SysRes::_valEx handling is incorrect
1391 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1392 346474 PPC64 Power 8, spr TEXASRU register not supported
1393 346487 Compiler generates "note" about a future ABI change for PPC64
1394 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1395 and causing spurious "invalid read of size 8" warnings
1396 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1397 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1398 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1399 347322 Power PC regression test cleanup
1400 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1402 347389 unhandled syscall: 373 (Linux ARM syncfs)
1403 347686 Patch set to cleanup PPC64 regtests
1404 347978 Remove bash dependencies where not needed
1405 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1406 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1408 348102 Patch updating v4l2 API support
1409 348247 amd64 front end: jno jumps wrongly when overflow is not set
1410 348269 Improve mmap MAP_HUGETLB support.
1411 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1412 348345 Assertion fails for negative lineno
1413 348377 Unsupported ARM instruction: yield
1414 348565 Fix detection of command line option availability for clang
1415 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1416 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1417 348748 Fix redundant condition
1418 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1419 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1420 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1421 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1422 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1423 349626 Implemented additional Xen hypercalls
1424 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1425 349790 Clean up of the hardware capability checking utilities.
1426 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1427 349874 Fix typos in source code
1428 349879 memcheck: add handwritten assembly for helperc_LOADV*
1429 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1430 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1431 350202 Add limited param to 'monitor block_list'
1432 350290 s390x: Support instructions fixbr(a)
1433 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1434 350809 Fix none/tests/async-sigs for Solaris
1435 350811 Remove reference to --db-attach which has been removed.
1436 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1437 350854 hard-to-understand code in VG_(load_ELF)()
1438 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1439 351386 Solaris: Cannot run ld.so.1 under Valgrind
1440 351474 Fix VG_(iseqsigset) as obvious
1441 351531 Typo in /include/vki/vki-xen-physdev.h header guard
1442 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
1443 351858 ldsoexec support on Solaris
1444 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
1445 352130 helgrind reports false races for printfs using mempcpy on FILE* state
1446 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
1447 352320 arm64 crash on none/tests/nestedfs
1448 352765 Vbit test fails on Power 6
1449 352768 The mbar instruction is missing from the Power PC support
1450 352769 Power PC program priority register (PPR) is not supported
1451 n-i-bz Provide implementations of certain compiler builtins to support
1452 compilers that may not provide those
1453 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
1454 n-i-bz Fix compilation on distros with glibc < 2.5
1455 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
1456 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
1457 n-i-bz mremap did not work properly on shared memory
1458 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
1459 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
1461 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
1462 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
1463 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
1467 Release 3.10.1 (25 November 2014)
1468 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1469 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
1470 and backports fixes for all reported missing AArch64 ARMv8 instructions
1471 and syscalls from the trunk. If you package or deliver 3.10.0 for others
1472 to use, you might want to consider upgrading to 3.10.1 instead.
1474 The following bugs have been fixed or resolved. Note that "n-i-bz"
1475 stands for "not in bugzilla" -- that is, a bug that was reported to us
1476 but never got a bugzilla entry. We encourage you to file bugs in
1477 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1478 than mailing the developers (or mailing lists) directly -- bugs that
1479 are not entered into bugzilla tend to get forgotten about or ignored.
1481 To see details of a given bug, visit
1482 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1483 where XXXXXX is the bug number as listed below.
1485 335440 arm64: ld1 (single structure) is not implemented
1486 335713 arm64: unhanded instruction: prfm (immediate)
1487 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
1488 339182 ppc64: AvSplat ought to load destination vector register with [..]
1489 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
1490 339433 ppc64 lxvw4x instruction uses four 32-byte loads
1491 339645 Use correct tag names in sys_getdents/64 wrappers
1492 339706 Fix false positive for ioctl(TIOCSIG) on linux
1493 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
1494 339853 arm64 times syscall unknown
1495 339855 arm64 unhandled getsid/setsid syscalls
1496 339858 arm64 dmb sy not implemented
1497 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
1498 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
1499 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
1501 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
1502 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
1503 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
1504 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
1505 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
1506 340509 arm64: unhandled instruction fcvtas
1507 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
1508 340632 arm64: unhandled instruction fcvtas
1509 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
1510 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
1511 340788 warning: unhandled syscall: 318 (getrandom)
1512 340807 disInstr(arm): unhandled instruction: 0xEE989B20
1513 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
1514 340922 arm64: unhandled getgroups/setgroups syscalls
1515 350251 Fix typo in VEX utility program (test_main.c).
1516 350407 arm64: unhandled instruction ucvtf (vector, integer)
1517 350809 none/tests/async-sigs breaks when run under cron on Solaris
1518 350811 update README.solaris after r15445
1519 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
1520 350854 strange code in VG_(load_ELF)()
1521 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1522 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
1523 n-i-bz Add missing ]] to terminate CDATA.
1524 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
1525 n-i-bz Enable sys_fadvise64_64 on arm32.
1526 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
1527 n-i-bz Add test cases for all known arm64 load/store instructions.
1528 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
1529 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
1530 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
1531 n-i-bz arm64: Implement "fcvtpu w, s".
1532 n-i-bz arm64: implement ADDP and various others
1533 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
1534 n-i-bz arm64: enable FCVT{A,N}S X,S.
1536 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
1540 Release 3.10.0 (10 September 2014)
1541 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1543 3.10.0 is a feature release with many improvements and the usual
1544 collection of bug fixes.
1546 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1547 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1548 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
1549 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
1550 significantly improved relative to the 3.9.0 release.
1552 * ================== PLATFORM CHANGES =================
1554 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
1555 is mostly complete, and is usable, but some SIMD instructions are as
1558 * Support for little-endian variant of the 64-bit POWER architecture.
1560 * Support for Android on MIPS32.
1562 * Support for 64bit FPU on MIPS32 platforms.
1564 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
1566 * Configuration for and running on Android targets has changed.
1567 See README.android in the source tree for details.
1569 * ================== DEPRECATED FEATURES =================
1571 * --db-attach is now deprecated and will be removed in the next
1572 valgrind feature release. The built-in GDB server capabilities are
1573 superior and should be used instead. Learn more here:
1574 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
1576 * ==================== TOOL CHANGES ====================
1580 - Client code can now selectively disable and re-enable reporting of
1581 invalid address errors in specific ranges using the new client
1582 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
1583 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
1585 - Leak checker: there is a new leak check heuristic called
1586 "length64". This is used to detect interior pointers pointing 8
1587 bytes inside a block, on the assumption that the first 8 bytes
1588 holds the value "block size - 8". This is used by
1589 sqlite3MemMalloc, for example.
1591 - Checking of system call parameters: if a syscall parameter
1592 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
1593 several fields not initialised, an error is now reported for each
1594 field. Previously, an error was reported only for the first
1595 uninitialised field.
1597 - Mismatched alloc/free checking: a new flag
1598 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
1599 such checks if necessary.
1603 - Improvements to error messages:
1605 o Race condition error message involving heap allocated blocks also
1606 show the thread number that allocated the raced-on block.
1608 o All locks referenced by an error message are now announced.
1609 Previously, some error messages only showed the lock addresses.
1611 o The message indicating where a lock was first observed now also
1612 describes the address/location of the lock.
1614 - Helgrind now understands the Ada task termination rules and
1615 creates a happens-before relationship between a terminated task
1616 and its master. This avoids some false positives and avoids a big
1617 memory leak when a lot of Ada tasks are created and terminated.
1618 The interceptions are only activated with forthcoming releases of
1619 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
1621 - A new GDB server monitor command "info locks" giving the list of
1622 locks, their location, and their status.
1626 - callgrind_control now supports the --vgdb-prefix argument,
1627 which is needed if valgrind was started with this same argument.
1629 * ==================== OTHER CHANGES ====================
1631 * Unwinding through inlined function calls. Stack unwinding can now
1632 make use of Dwarf3 inlined-unwind information if it is available.
1633 The practical effect is that inlined calls become visible in stack
1634 traces. The suppression matching machinery has been adjusted
1635 accordingly. This is controlled by the new option
1636 --read-inline-info=yes|no. Currently this is enabled by default
1637 only on Linux and Android targets and only for the tools Memcheck,
1640 * Valgrind can now read EXIDX unwind information on 32-bit ARM
1641 targets. If an object contains both CFI and EXIDX unwind
1642 information, Valgrind will prefer the CFI over the EXIDX. This
1643 facilitates unwinding through system libraries on arm-android
1646 * Address description logic has been improved and is now common
1647 between Memcheck and Helgrind, resulting in better address
1648 descriptions for some kinds of error messages.
1650 * Error messages about dubious arguments (eg, to malloc or calloc) are
1651 output like other errors. This means that they can be suppressed
1652 and they have a stack trace.
1654 * The C++ demangler has been updated for better C++11 support.
1656 * New and modified GDB server monitor features:
1658 - Thread local variables/storage (__thread) can now be displayed.
1660 - The GDB server monitor command "v.info location <address>"
1661 displays information about an address. The information produced
1662 depends on the tool and on the options given to valgrind.
1663 Possibly, the following are described: global variables, local
1664 (stack) variables, allocated or freed blocks, ...
1666 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
1667 ask the GDB server to stop at the start of program execution, at
1668 the end of the program execution and on Valgrind internal errors.
1670 - A new monitor command "v.info stats" shows various Valgrind core
1671 and tool statistics.
1673 - A new monitor command "v.set hostvisibility" allows the GDB server
1674 to provide access to Valgrind internal host status/memory.
1676 * A new option "--aspace-minaddr=<address>" can in some situations
1677 allow the use of more memory by decreasing the address above which
1678 Valgrind maps memory. It can also be used to solve address
1679 conflicts with system libraries by increasing the default value.
1680 See user manual for details.
1682 * The amount of memory used by Valgrind to store debug info (unwind
1683 info, line number information and symbol data) has been
1684 significantly reduced, even though Valgrind now reads more
1685 information in order to support unwinding of inlined function calls.
1687 * Dwarf3 handling with --read-var-info=yes has been improved:
1689 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
1691 - Code compiled with
1692 -ffunction-sections -fdata-sections -Wl,--gc-sections
1693 no longer causes assertion failures.
1695 * Improved checking for the --sim-hints= and --kernel-variant=
1696 options. Unknown strings are now detected and reported to the user
1699 * The semantics of stack start/end boundaries in the valgrind.h
1700 VALGRIND_STACK_REGISTER client request has been clarified and
1701 documented. The convention is that start and end are respectively
1702 the lowest and highest addressable bytes of the stack.
1704 * ==================== FIXED BUGS ====================
1706 The following bugs have been fixed or resolved. Note that "n-i-bz"
1707 stands for "not in bugzilla" -- that is, a bug that was reported to us
1708 but never got a bugzilla entry. We encourage you to file bugs in
1709 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1710 than mailing the developers (or mailing lists) directly -- bugs that
1711 are not entered into bugzilla tend to get forgotten about or ignored.
1713 To see details of a given bug, visit
1714 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1715 where XXXXXX is the bug number as listed below.
1717 175819 Support for ipv6 socket reporting with --track-fds
1718 232510 make distcheck fails
1719 249435 Analyzing wine programs with callgrind triggers a crash
1720 278972 support for inlined function calls in stacktraces and suppression
1722 291310 FXSAVE instruction marks memory as undefined on amd64
1723 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
1724 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
1725 315199 vgcore file for threaded app does not show which thread crashed
1726 315952 tun/tap ioctls are not supported
1727 323178 Unhandled instruction: PLDW register (ARM)
1728 323179 Unhandled instruction: PLDW immediate (ARM)
1729 324050 Helgrind: SEGV because of unaligned stack when using movdqa
1730 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
1731 325124 [MIPSEL] Compilation error
1732 325477 Phase 4 support for IBM Power ISA 2.07
1733 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
1734 325628 Phase 5 support for IBM Power ISA 2.07
1735 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
1736 325751 Missing the two privileged Power PC Transactional Memory Instructions
1737 325816 Phase 6 support for IBM Power ISA 2.07
1738 325856 Make SGCheck fail gracefully on unsupported platforms
1739 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
1740 326436 DRD: False positive in libstdc++ std::list::push_back
1741 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
1742 326462 Refactor vgdb to isolate invoker stuff into separate module
1743 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
1744 326623 DRD: false positive conflict report in a field assignment
1745 326724 Valgrind does not compile on OSX 1.9 Mavericks
1746 326816 Intercept for __strncpy_sse2_unaligned missing?
1747 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
1748 326983 Clear direction flag after tests on amd64.
1749 327212 Do not prepend the current directory to absolute path names.
1750 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
1751 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
1752 327284 s390x: Fix translation of the risbg instruction
1753 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
1754 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
1755 327916 DW_TAG_typedef may have no name
1756 327943 s390x: add a redirection for the 'index' function
1757 328100 XABORT not implemented
1758 328205 Implement additional Xen hypercalls
1759 328454 add support Backtraces with ARM unwind tables (EXIDX)
1760 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
1761 328711 valgrind.1 manpage "memcheck options" section is badly generated
1762 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
1763 329612 Incorrect handling of AT_BASE for image execution
1764 329694 clang warns about using uninitialized variable
1765 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
1766 330228 mmap must align to VKI_SHMLBA on mips32
1767 330257 LLVM does not support `-mno-dynamic-no-pic` option
1768 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
1769 330459 --track-fds=yes doesn't track eventfds
1770 330469 Add clock_adjtime syscall support
1771 330594 Missing sysalls on PowerPC / uClibc
1772 330622 Add test to regression suite for POWER instruction: dcbzl
1773 330939 Support for AMD's syscall instruction on x86
1775 330941 Typo in PRE(poll) syscall wrapper
1776 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
1777 331254 Fix expected output for memcheck/tests/dw4
1778 331255 Fix race condition in test none/tests/coolo_sigaction
1779 331257 Fix type of jump buffer in test none/tests/faultstatus
1780 331305 configure uses bash specific syntax
1781 331337 s390x WARNING: unhandled syscall: 326 (dup3)
1782 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
1783 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
1784 331829 Unexpected ioctl opcode sign extension
1785 331830 ppc64: WARNING: unhandled syscall: 96/97
1786 331839 drd/tests/sem_open specifies invalid semaphore name
1787 331847 outcome of drd/tests/thread_name is nondeterministic
1788 332037 Valgrind cannot handle Thumb "add pc, reg"
1789 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
1790 consistency checks enabled
1791 332263 intercepts for pthread_rwlock_timedrdlock and
1792 pthread_rwlock_timedwrlock are incorrect
1793 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
1795 332276 Implement additional Xen hypercalls
1796 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
1797 332765 Fix ms_print to create temporary files in a proper directory
1798 333072 drd: Add semaphore annotations
1799 333145 Tests for missaligned PC+#imm access for arm
1800 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
1801 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
1802 333248 WARNING: unhandled syscall: unix:443
1803 333428 ldr.w pc [rD, #imm] instruction leads to assertion
1804 333501 cachegrind: assertion: Cache set count is not a power of two.
1807 333666 Recognize MPX instructions and bnd prefix.
1808 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
1809 333817 Valgrind reports the memory areas written to by the SG_IO
1811 334049 lzcnt fails silently (x86_32)
1812 334384 Valgrind does not have support Little Endian support for
1814 334585 recvmmsg unhandled (+patch) (arm)
1815 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
1816 334727 Build fails with -Werror=format-security
1817 334788 clarify doc about --log-file initial program directory
1818 334834 PPC64 Little Endian support, patch 2
1819 334836 PPC64 Little Endian support, patch 3 testcase fixes
1820 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
1821 335034 Unhandled ioctl: HCIGETDEVLIST
1822 335155 vgdb, fix error print statement.
1823 335262 arm64: movi 8bit version is not supported
1824 335263 arm64: dmb instruction is not implemented
1825 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
1826 335496 arm64: sbc/abc instructions are not implemented
1827 335554 arm64: unhandled instruction: abs
1828 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
1829 335735 arm64: unhandled instruction: cnt
1830 335736 arm64: unhandled instruction: uaddlv
1831 335848 arm64: unhandled instruction: {s,u}cvtf
1832 335902 arm64: unhandled instruction: sli
1833 335903 arm64: unhandled instruction: umull (vector)
1834 336055 arm64: unhandled instruction: mov (element)
1835 336062 arm64: unhandled instruction: shrn{,2}
1836 336139 mip64: [...] valgrind hangs and spins on a single core [...]
1837 336189 arm64: unhandled Instruction: mvn
1838 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
1839 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
1840 336772 Make moans about unknown ioctls more informative
1841 336957 Add a section about the Solaris/illumos port on the webpage
1842 337094 ifunc wrapper is broken on ppc64
1843 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
1844 337528 leak check heuristic for block prefixed by length as 64bit number
1845 337740 Implement additional Xen hypercalls
1846 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
1847 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
1848 337871 deprecate --db-attach
1849 338023 Add support for all V4L2/media ioctls
1850 338024 inlined functions are not shown if DW_AT_ranges is used
1851 338106 Add support for 'kcmp' syscall
1852 338115 DRD: computed conflict set differs from actual after fork
1853 338160 implement display of thread local storage in gdbsrv
1854 338205 configure.ac and check for -Wno-tautological-compare
1855 338300 coredumps are missing one byte of every segment
1856 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
1857 338499 --sim-hints parsing broken due to wrong order in tokens
1858 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
1859 338681 Unable to unwind through clone thread created on i386-linux
1860 338698 race condition between gdbsrv and vgdb on startup
1861 338703 helgrind on arm-linux gets false positives in dynamic loader
1862 338791 alt dwz files can be relative of debug/main file
1863 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
1864 338932 build V-trunk with gcc-trunk
1865 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
1866 345079 Fix build problems in VEX/useful/test_main.c
1867 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
1868 n-i-bz s390x: Fix memory corruption for multithreaded applications
1869 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
1870 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
1871 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
1872 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
1873 n-i-bz Add command-line option --default-suppressions
1874 n-i-bz Add support for BLKDISCARDZEROES ioctl
1875 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
1876 n-i-bz Add support for sys_pivot_root and sys_unshare
1878 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
1879 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
1880 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
1884 Release 3.9.0 (31 October 2013)
1885 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1886 3.9.0 is a feature release with many improvements and the usual
1887 collection of bug fixes.
1889 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
1890 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
1891 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
1892 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
1894 * ================== PLATFORM CHANGES =================
1896 * Support for MIPS64 LE and BE running Linux. Valgrind has been
1897 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
1899 * Support for MIPS DSP ASE on MIPS32 platforms.
1901 * Support for s390x Decimal Floating Point instructions on hosts that
1902 have the DFP facility installed.
1904 * Support for POWER8 (Power ISA 2.07) instructions
1906 * Support for Intel AVX2 instructions. This is available only on 64
1909 * Initial support for Intel Transactional Synchronization Extensions,
1912 * Initial support for Hardware Transactional Memory on POWER.
1914 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
1915 run large GUI apps tolerably well.
1917 * ==================== TOOL CHANGES ====================
1921 - Improvements in handling of vectorised code, leading to
1922 significantly fewer false error reports. You need to use the flag
1923 --partial-loads-ok=yes to get the benefits of these changes.
1925 - Better control over the leak checker. It is now possible to
1926 specify which leak kinds (definite/indirect/possible/reachable)
1927 should be displayed, which should be regarded as errors, and which
1928 should be suppressed by a given leak suppression. This is done
1929 using the options --show-leak-kinds=kind1,kind2,..,
1930 --errors-for-leak-kinds=kind1,kind2,.. and an optional
1931 "match-leak-kinds:" line in suppression entries, respectively.
1933 Note that generated leak suppressions contain this new line and
1934 are therefore more specific than in previous releases. To get the
1935 same behaviour as previous releases, remove the "match-leak-kinds:"
1936 line from generated suppressions before using them.
1938 - Reduced "possible leak" reports from the leak checker by the use
1939 of better heuristics. The available heuristics provide detection
1940 of valid interior pointers to std::stdstring, to new[] allocated
1941 arrays with elements having destructors and to interior pointers
1942 pointing to an inner part of a C++ object using multiple
1943 inheritance. They can be selected individually using the
1944 option --leak-check-heuristics=heur1,heur2,...
1946 - Better control of stacktrace acquisition for heap-allocated
1947 blocks. Using the --keep-stacktraces option, it is possible to
1948 control independently whether a stack trace is acquired for each
1949 allocation and deallocation. This can be used to create better
1950 "use after free" errors or to decrease Valgrind's resource
1951 consumption by recording less information.
1953 - Better reporting of leak suppression usage. The list of used
1954 suppressions (shown when the -v option is given) now shows, for
1955 each leak suppressions, how many blocks and bytes it suppressed
1956 during the last leak search.
1960 - False errors resulting from the use of statically initialised
1961 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
1964 - False errors resulting from the use of pthread_cond_waits that
1965 timeout, have been removed.
1967 * ==================== OTHER CHANGES ====================
1969 * Some attempt to tune Valgrind's space requirements to the expected
1970 capabilities of the target:
1972 - The default size of the translation cache has been reduced from 8
1973 sectors to 6 on Android platforms, since each sector occupies
1974 about 40MB when using Memcheck.
1976 - The default size of the translation cache has been increased to 16
1977 sectors on all other platforms, reflecting the fact that large
1978 applications require instrumentation and storage of huge amounts
1979 of code. For similar reasons, the number of memory mapped
1980 segments that can be tracked has been increased by a factor of 6.
1982 - In all cases, the maximum number of sectors in the translation
1983 cache can be controlled by the new flag --num-transtab-sectors.
1985 * Changes in how debug info (line numbers, etc) is read:
1987 - Valgrind no longer temporarily mmaps the entire object to read
1988 from it. Instead, reading is done through a small fixed sized
1989 buffer. This avoids virtual memory usage spikes when Valgrind
1990 reads debuginfo from large shared objects.
1992 - A new experimental remote debug info server. Valgrind can read
1993 debug info from a different machine (typically, a build host)
1994 where debuginfo objects are stored. This can save a lot of time
1995 and hassle when running Valgrind on resource-constrained targets
1996 (phones, tablets) when the full debuginfo objects are stored
1997 somewhere else. This is enabled by the --debuginfo-server=
2000 - Consistency checking between main and debug objects can be
2001 disabled using the --allow-mismatched-debuginfo option.
2003 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
2004 scanning can recover stack traces in some cases when the normal
2005 unwind mechanisms fail. Stack scanning is best described as "a
2006 nasty, dangerous and misleading hack" and so is disabled by default.
2007 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
2010 * Detection and merging of recursive stack frame cycles. When your
2011 program has recursive algorithms, this limits the memory used by
2012 Valgrind for recorded stack traces and avoids recording
2013 uninteresting repeated calls. This is controlled by the command
2014 line option --merge-recursive-frame and by the monitor command
2015 "v.set merge-recursive-frames".
2017 * File name and line numbers for used suppressions. The list of used
2018 suppressions (shown when the -v option is given) now shows, for each
2019 used suppression, the file name and line number where the suppression
2022 * New and modified GDB server monitor features:
2024 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
2025 that can be used to execute gdbserver monitor commands from the
2028 - A new monitor command, "v.info open_fds", that gives the list of
2029 open file descriptors and additional details.
2031 - An optional message in the "v.info n_errs_found" monitor command,
2032 for example "v.info n_errs_found test 1234 finished", allowing a
2033 comment string to be added to the process output, perhaps for the
2034 purpose of separating errors of different tests or test phases.
2036 - A new monitor command "v.info execontext" that shows information
2037 about the stack traces recorded by Valgrind.
2039 - A new monitor command "v.do expensive_sanity_check_general" to run
2040 some internal consistency checks.
2042 * New flag --sigill-diagnostics to control whether a diagnostic
2043 message is printed when the JIT encounters an instruction it can't
2044 translate. The actual behavior -- delivery of SIGILL to the
2045 application -- is unchanged.
2047 * The maximum amount of memory that Valgrind can use on 64 bit targets
2048 has been increased from 32GB to 64GB. This should make it possible
2049 to run applications on Memcheck that natively require up to about 35GB.
2051 * ==================== FIXED BUGS ====================
2053 The following bugs have been fixed or resolved. Note that "n-i-bz"
2054 stands for "not in bugzilla" -- that is, a bug that was reported to us
2055 but never got a bugzilla entry. We encourage you to file bugs in
2056 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2057 than mailing the developers (or mailing lists) directly -- bugs that
2058 are not entered into bugzilla tend to get forgotten about or ignored.
2060 To see details of a given bug, visit
2061 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2062 where XXXXXX is the bug number as listed below.
2064 123837 system call: 4th argument is optional, depending on cmd
2065 135425 memcheck should tell you where Freed blocks were Mallocd
2066 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2067 207815 Adds some of the drm ioctls to syswrap-linux.c
2068 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2069 252955 Impossible to compile with ccache
2070 253519 Memcheck reports auxv pointer accesses as invalid reads.
2071 263034 Crash when loading some PPC64 binaries
2072 269599 Increase deepest backtrace
2073 274695 s390x: Support "compare to/from logical" instructions (z196)
2074 275800 s390x: Autodetect cache info (part 2)
2075 280271 Valgrind reports possible memory leaks on still-reachable std::string
2076 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2077 289578 Backtraces with ARM unwind tables (stack scan flags)
2078 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2079 304832 ppc32: build failure
2080 305431 Use find_buildid shdr fallback for separate .debug files
2081 305728 Add support for AVX2 instructions
2082 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2083 306035 s390x: Fix IR generation for LAAG and friends
2084 306054 s390x: Condition code computation for convert-to-int/logical
2085 306098 s390x: alternate opcode form for convert to/from fixed
2086 306587 Fix cache line detection from auxiliary vector for PPC.
2087 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2088 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2089 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2090 307101 sys_capget second argument can be NULL
2091 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2092 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2093 307113 s390x: DFP support
2094 307141 valgrind does't work in mips-linux system
2095 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2096 307285 x86_amd64 feature test for avx in test suite is wrong
2097 307290 memcheck overlap testcase needs memcpy version filter
2098 307463 Please add "&limit=0" to the "all open bugs" link
2099 307465 --show-possibly-lost=no should reduce the error count / exit code
2100 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2101 307729 pkgconfig support broken valgrind.pc
2102 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2103 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2104 308089 Unhandled syscall on ppc64: prctl
2105 308135 PPC32 MPC8xx has 16 bytes cache size
2106 308321 testsuite memcheck filter interferes with gdb_filter
2108 308341 vgdb should report process exit (or fatal signal)
2109 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2110 308495 Remove build dependency on installed Xen headers
2111 308573 Internal error on 64-bit instruction executed in 32-bit mode
2113 308627 pmovmskb validity bit propagation is imprecise
2114 308644 vgdb command for having the info for the track-fds option
2115 308711 give more info about aspacemgr and arenas in out_of_memory
2116 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2117 308718 ARM implement SMLALBB family of instructions
2118 308886 Missing support for PTRACE_SET/GETREGSET
2119 308930 syscall name_to_handle_at (303 on amd64) not handled
2120 309229 V-bit tester does not report number of tests generated
2121 309323 print unrecognized instuction on MIPS
2122 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2123 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2124 309430 Self hosting ppc64 encounters a vassert error on operand type
2125 309600 valgrind is a bit confused about 0-sized sections
2126 309823 Generate errors for still reachable blocks
2127 309921 PCMPISTRI validity bit propagation is imprecise
2128 309922 none/tests/ppc64/test_dfp5 sometimes fails
2129 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2130 310424 --read-var-info does not properly describe static variables
2131 310792 search additional path for debug symbols
2132 310931 s390x: Message-security assist (MSA) instruction extension [..]
2133 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2134 311318 ARM: "128-bit constant is not implemented" error message
2135 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2136 311690 V crashes because it redirects branches inside of a redirected function
2137 311880 x86_64: make regtest hangs at shell_valid1
2138 311922 WARNING: unhandled syscall: 170
2140 312171 ppc: insn selection for DFP
2141 312571 Rounding mode call wrong for the DFP Iops [..]
2142 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2143 312913 Dangling pointers error should also report the alloc stack trace
2144 312980 Building on Mountain Lion generates some compiler warnings
2145 313267 Adding MIPS64/Linux port to Valgrind
2148 313811 Buffer overflow in assert_fail
2149 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2150 314269 ppc: dead code in insn selection
2151 314718 ARM: implement integer divide instruction (sdiv and udiv)
2152 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2153 315441 sendmsg syscall should ignore unset msghdr msg_flags
2154 315534 msgrcv inside a thread causes valgrind to hang (block)
2155 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2156 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2157 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2158 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2159 316144 valgrind.1 manpage contains unknown ??? strings [..]
2160 316145 callgrind command line options in manpage reference (unknown) [..]
2161 316145 callgrind command line options in manpage reference [..]
2162 316181 drd: Fixed a 4x slowdown for certain applications
2163 316503 Valgrind does not support SSE4 "movntdqa" instruction
2164 316535 Use of |signed int| instead of |size_t| in valgrind messages
2165 316696 fluidanimate program of parsec 2.1 stuck
2166 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2167 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2168 317186 "Impossible happens" when occurs VCVT instruction on ARM
2169 317318 Support for Threading Building Blocks "scalable_malloc"
2170 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2171 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2172 317463 bmi testcase IR SANITY CHECK FAILURE
2173 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2174 318050 libmpiwrap fails to compile with out-of-source build
2175 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2176 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2177 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2178 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2179 318932 Add missing PPC64 and PPC32 system call support
2180 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2181 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2182 319494 VEX Makefile-gcc standalone build update after r2702
2183 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2184 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2185 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2186 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2187 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2188 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2189 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2190 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2191 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2192 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2193 320895 add fanotify support (patch included)
2194 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2195 321065 Valgrind updates for Xen 4.3
2196 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2197 321363 Unhandled instruction: SSAX (ARM + Thumb)
2198 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2199 321466 Unhandled instruction: SHASX (ARM + Thumb)
2200 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2201 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2202 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2203 321620 Unhandled instruction: UASX (ARM + Thumb)
2204 321621 Unhandled instruction: USAX (ARM + Thumb)
2205 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2206 321693 Unhandled instruction: LDRSBT (Thumb)
2207 321694 Unhandled instruction: UQASX (ARM + Thumb)
2208 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2209 321697 Unhandled instruction: UHASX (ARM + Thumb)
2210 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2211 321704 Unhandled instruction: REVSH (ARM + Thumb)
2212 321730 Add cg_diff and cg_merge man pages
2213 321738 Add vgdb and valgrind-listener man pages
2215 321891 Unhandled instruction: LDRHT (Thumb)
2216 321960 pthread_create() then alloca() causing invalid stack write errors
2217 321969 ppc32 and ppc64 don't support [lf]setxattr
2218 322254 Show threadname together with tid if set by application
2219 322294 Add initial support for IBM Power ISA 2.07
2220 322368 Assertion failure in wqthread_hijack under OS X 10.8
2221 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2222 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2223 322851 0bXXX binary literal syntax is not standard
2224 323035 Unhandled instruction: LDRSHT(Thumb)
2225 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2226 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2227 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2228 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2229 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2230 323437 Phase 2 support for IBM Power ISA 2.07
2231 323713 Support mmxext (integer sse) subset on i386 (athlon)
2232 323803 Transactional memory instructions are not supported for Power
2233 323893 SSE3 not available on amd cpus in valgrind
2234 323905 Probable false positive from Valgrind/drd on close()
2235 323912 valgrind.h header isn't compatible for mingw64
2236 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
2237 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
2238 324181 mmap does not handle MAP_32BIT
2239 324227 memcheck false positive leak when a thread calls exit+block [..]
2240 324421 Support for fanotify API on ARM architecture
2241 324514 gdbserver monitor cmd output behaviour consistency [..]
2242 324518 ppc64: Emulation of dcbt instructions does not handle [..]
2243 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
2244 324582 When access is made to freed memory, report both allocation [..]
2245 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
2246 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
2247 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
2248 324834 Unhandled instructions in Microsoft C run-time for x86_64
2249 324894 Phase 3 support for IBM Power ISA 2.07
2250 326091 drd: Avoid false race reports from optimized strlen() impls
2251 326113 valgrind libvex hwcaps error on AMD64
2252 n-i-bz Some wrong command line options could be ignored
2253 n-i-bz patch to allow fair-sched on android
2254 n-i-bz report error for vgdb snapshot requested before execution
2255 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
2257 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
2261 Release 3.8.1 (19 September 2012)
2262 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2263 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
2264 that occur moderately frequently in real use cases, adds support for
2265 some missing instructions on ARM, and fixes a deadlock condition on
2266 MacOSX. If you package or deliver 3.8.0 for others to use, you might
2267 want to consider upgrading to 3.8.1 instead.
2269 The following bugs have been fixed or resolved. Note that "n-i-bz"
2270 stands for "not in bugzilla" -- that is, a bug that was reported to us
2271 but never got a bugzilla entry. We encourage you to file bugs in
2272 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2273 than mailing the developers (or mailing lists) directly -- bugs that
2274 are not entered into bugzilla tend to get forgotten about or ignored.
2276 To see details of a given bug, visit
2277 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2278 where XXXXXX is the bug number as listed below.
2281 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
2282 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
2283 298281 wcslen causes false(?) uninitialised value warnings
2284 301281 valgrind hangs on OS X when the process calls system()
2285 304035 disInstr(arm): unhandled instruction 0xE1023053
2286 304867 implement MOVBE instruction in x86 mode
2287 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
2288 305042 amd64: implement 0F 7F encoding of movq between two registers
2289 305199 ARM: implement QDADD and QDSUB
2290 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
2291 305513 killed by fatal signal: SIGSEGV
2292 305690 DRD reporting invalid semaphore when sem_trywait fails
2293 305926 Invalid alignment checks for some AVX instructions
2294 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
2295 306310 3.8.0 release tarball missing some files
2296 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
2297 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
2298 n-i-bz shmat of a segment > 4Gb does not work
2299 n-i-bz simulate_control_c script wrong USR1 signal number on mips
2300 n-i-bz vgdb ptrace calls wrong on mips [...]
2301 n-i-bz Fixes for more MPI false positives
2302 n-i-bz exp-sgcheck's memcpy causes programs to segfault
2303 n-i-bz OSX build w/ clang: asserts at startup
2304 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
2305 n-i-bz fix a couple of union tag-vs-field mixups
2306 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
2308 The following bugs were fixed in 3.8.0 but not listed in this NEWS
2311 254088 Valgrind should know about UD2 instruction
2314 304754 NEWS blows TeX's little mind
2316 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
2320 Release 3.8.0 (10 August 2012)
2321 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2322 3.8.0 is a feature release with many improvements and the usual
2323 collection of bug fixes.
2325 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2326 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
2327 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
2328 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
2329 There is initial support for MacOSX 10.8, but it is not usable for
2330 serious work at present.
2332 * ================== PLATFORM CHANGES =================
2334 * Support for MIPS32 platforms running Linux. Valgrind has been
2335 tested on MIPS32 and MIPS32r2 platforms running different Debian
2336 Squeeze and MeeGo distributions. Both little-endian and big-endian
2337 cores are supported. The tools Memcheck, Massif and Lackey have
2338 been tested and are known to work. See README.mips for more details.
2340 * Preliminary support for Android running on x86.
2342 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
2344 * Support for Intel AVX instructions and for AES instructions. This
2345 support is available only for 64 bit code.
2347 * Support for POWER Decimal Floating Point instructions.
2349 * ==================== TOOL CHANGES ====================
2351 * Non-libc malloc implementations are now supported. This is useful
2352 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
2353 Using the new option --soname-synonyms, such tools can be informed
2354 that the malloc implementation is either linked statically into the
2355 executable, or is present in some other shared library different
2356 from libc.so. This makes it possible to process statically linked
2357 programs, and programs using other malloc libraries, for example
2358 TCMalloc or JEMalloc.
2360 * For tools that provide their own replacement for malloc et al, the
2361 option --redzone-size=<number> allows users to specify the size of
2362 the padding blocks (redzones) added before and after each client
2363 allocated block. Smaller redzones decrease the memory needed by
2364 Valgrind. Bigger redzones increase the chance to detect blocks
2365 overrun or underrun. Prior to this change, the redzone size was
2366 hardwired to 16 bytes in Memcheck.
2370 - The leak_check GDB server monitor command now can
2371 control the maximum nr of loss records to output.
2373 - Reduction of memory use for applications allocating
2374 many blocks and/or having many partially defined bytes.
2376 - Addition of GDB server monitor command 'block_list' that lists
2377 the addresses/sizes of the blocks of a leak search loss record.
2379 - Addition of GDB server monitor command 'who_points_at' that lists
2380 the locations pointing at a block.
2382 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2383 detect an invalid access of these redzones, by marking them
2384 noaccess. Similarly, if a redzone size is given for a memory
2385 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2386 This still allows to find some bugs if the user has forgotten to
2387 mark the pool superblock noaccess.
2389 - Performance of memory leak check has been improved, especially in
2390 cases where there are many leaked blocks and/or many suppression
2391 rules used to suppress leak reports.
2393 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2394 more precise analysis, which is important for LLVM/Clang
2395 generated code. This is at the cost of somewhat reduced
2396 performance. Note there is no change to analysis precision or
2397 costs on Linux targets.
2401 - Added even more facilities that can help finding the cause of a data
2402 race, namely the command-line option --ptrace-addr and the macro
2403 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2405 - Fixed a subtle bug that could cause false positive data race reports.
2407 * ==================== OTHER CHANGES ====================
2409 * The C++ demangler has been updated so as to work well with C++
2410 compiled by up to at least g++ 4.6.
2412 * Tool developers can make replacement/wrapping more flexible thanks
2413 to the new option --soname-synonyms. This was reported above, but
2414 in fact is very general and applies to all function
2415 replacement/wrapping, not just to malloc-family functions.
2417 * Round-robin scheduling of threads can be selected, using the new
2418 option --fair-sched= yes. Prior to this change, the pipe-based
2419 thread serialisation mechanism (which is still the default) could
2420 give very unfair scheduling. --fair-sched=yes improves
2421 responsiveness of interactive multithreaded applications, and
2422 improves repeatability of results from the thread checkers Helgrind
2425 * For tool developers: support to run Valgrind on Valgrind has been
2426 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2428 * gdbserver now shows the float shadow registers as integer
2429 rather than float values, as the shadow values are mostly
2430 used as bit patterns.
2432 * Increased limit for the --num-callers command line flag to 500.
2434 * Performance improvements for error matching when there are many
2435 suppression records in use.
2437 * Improved support for DWARF4 debugging information (bug 284184).
2439 * Initial support for DWZ compressed Dwarf debug info.
2441 * Improved control over the IR optimiser's handling of the tradeoff
2442 between performance and precision of exceptions. Specifically,
2443 --vex-iropt-precise-memory-exns has been removed and replaced by
2444 --vex-iropt-register-updates, with extended functionality. This
2445 allows the Valgrind gdbserver to always show up to date register
2448 * Modest performance gains through the use of translation chaining for
2451 * ==================== FIXED BUGS ====================
2453 The following bugs have been fixed or resolved. Note that "n-i-bz"
2454 stands for "not in bugzilla" -- that is, a bug that was reported to us
2455 but never got a bugzilla entry. We encourage you to file bugs in
2456 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2457 than mailing the developers (or mailing lists) directly -- bugs that
2458 are not entered into bugzilla tend to get forgotten about or ignored.
2460 To see details of a given bug, visit
2461 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2462 where XXXXXX is the bug number as listed below.
2464 197914 Building valgrind from svn now requires automake-1.10
2465 203877 increase to 16Mb maximum allowed alignment for memalign et al
2466 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
2467 247386 make perf does not run all performance tests
2468 270006 Valgrind scheduler unfair
2469 270777 Adding MIPS/Linux port to Valgrind
2470 270796 s390x: Removed broken support for the TS insn
2471 271438 Fix configure for proper SSE4.2 detection
2472 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
2473 273475 Add support for AVX instructions
2474 274078 improved configure logic for mpicc
2475 276993 fix mremap 'no thrash checks'
2476 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
2477 281482 memcheck incorrect byte allocation count in realloc() for silly argument
2478 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
2479 283413 Fix wrong sanity check
2480 283671 Robustize alignment computation in LibVEX_Alloc
2481 283961 Adding support for some HCI IOCTLs
2482 284124 parse_type_DIE: confused by: DWARF 4
2483 284864 == 273475 (Add support for AVX instructions)
2484 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
2485 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
2486 285725 == 273475 (Add support for AVX instructions)
2487 286261 add wrapper for linux I2C_RDWR ioctl
2488 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
2489 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
2490 286384 configure fails "checking for a supported version of gcc"
2491 286497 == 273475 (Add support for AVX instructions)
2492 286596 == 273475 (Add support for AVX instructions)
2493 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
2494 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
2495 287260 Incorrect conditional jump or move depends on uninitialised value(s)
2496 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
2497 287307 == 273475 (Add support for AVX instructions)
2498 287858 VG_(strerror): unknown error
2499 288298 (MacOSX) unhandled syscall shm_unlink
2500 288995 == 273475 (Add support for AVX instructions)
2501 289470 Loading of large Mach-O thin binaries fails.
2502 289656 == 273475 (Add support for AVX instructions)
2503 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
2504 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
2505 289839 s390x: Provide support for unicode conversion instructions
2506 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
2507 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
2508 290655 Add support for AESKEYGENASSIST instruction
2509 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
2510 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
2511 291253 ES register not initialised in valgrind simulation
2512 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
2513 291865 s390x: Support the "Compare Double and Swap" family of instructions
2514 292300 == 273475 (Add support for AVX instructions)
2515 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
2516 292493 == 273475 (Add support for AVX instructions)
2517 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
2518 292627 Missing support for some SCSI ioctls
2519 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
2520 292841 == 273475 (Add support for AVX instructions)
2521 292993 implement the getcpu syscall on amd64-linux
2522 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
2523 293088 Add some VEX sanity checks for ppc64 unhandled instructions
2524 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
2525 293754 PCMPxSTRx not implemented for 16-bit characters
2526 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
2527 293808 CLFLUSH not supported by latest VEX for amd64
2528 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
2529 294048 MPSADBW instruction not implemented
2530 294055 regtest none/tests/shell fails when locale is not set to C
2531 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
2532 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
2533 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
2534 294260 disInstr_AMD64: disInstr miscalculated next %rip
2535 294523 --partial-loads-ok=yes causes false negatives
2536 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
2537 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
2538 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
2539 295089 can not annotate source for both helgrind and drd
2540 295221 POWER Processor decimal floating point instruction support missing
2541 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
2542 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
2543 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
2544 295617 ARM - Add some missing syscalls
2545 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
2546 296229 Linux user input device ioctls missing wrappers
2547 296318 ELF Debug info improvements (more than one rx/rw mapping)
2548 296422 Add translation chaining support
2549 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
2550 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
2551 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
2552 297078 gdbserver signal handling problems [..]
2553 297147 drd false positives on newly allocated memory
2554 297329 disallow decoding of IBM Power DFP insns on some machines
2555 297497 POWER Processor decimal floating point instruction support missing
2556 297701 Another alias for strncasecmp_l in libc-2.13.so
2557 297911 'invalid write' not reported when using APIs for custom mem allocators.
2558 297976 s390x: revisit EX implementation
2559 297991 Valgrind interferes with mmap()+ftell()
2560 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
2561 297993 Fix compilation of valgrind with gcc -g3.
2562 298080 POWER Processor DFP support missing, part 3
2563 298227 == 273475 (Add support for AVX instructions)
2564 298335 == 273475 (Add support for AVX instructions)
2565 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
2566 298394 s390x: Don't bail out on an unknown machine model. [..]
2567 298421 accept4() syscall (366) support is missing for ARM
2568 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
2569 298732 valgrind installation problem in ubuntu with kernel version 3.x
2570 298862 POWER Processor DFP instruction support missing, part 4
2571 298864 DWARF reader mis-parses DW_FORM_ref_addr
2572 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
2573 299053 Support DWARF4 DW_AT_high_pc constant form
2574 299104 == 273475 (Add support for AVX instructions)
2575 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
2576 299629 dup3() syscall (358) support is missing for ARM
2577 299694 POWER Processor DFP instruction support missing, part 5
2578 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
2579 299803 == 273475 (Add support for AVX instructions)
2580 299804 == 273475 (Add support for AVX instructions)
2581 299805 == 273475 (Add support for AVX instructions)
2582 300140 ARM - Missing (T1) SMMUL
2583 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
2584 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
2585 300414 FCOM and FCOMP unimplemented for amd64 guest
2586 301204 infinite loop in canonicaliseSymtab with ifunc symbol
2587 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
2588 301265 add x86 support to Android build
2589 301984 configure script doesn't detect certain versions of clang
2590 302205 Fix compiler warnings for POWER VEX code and POWER test cases
2591 302287 Unhandled movbe instruction on Atom processors
2592 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
2593 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
2594 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
2595 302656 == 273475 (Add support for AVX instructions)
2596 302709 valgrind for ARM needs extra tls support for android emulator [..]
2597 302827 add wrapper for CDROM_GET_CAPABILITY
2598 302901 Valgrind crashes with dwz optimized debuginfo
2599 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
2600 303116 Add support for the POWER instruction popcntb
2601 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
2602 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
2603 303466 == 273475 (Add support for AVX instructions)
2604 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
2605 303963 strstr() function produces wrong results under valgrind callgrind
2606 304054 CALL_FN_xx macros need to enforce stack alignment
2607 304561 tee system call not supported
2608 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
2609 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
2610 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
2611 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
2612 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
2613 n-i-bz s390x: Shadow registers can now be examined using vgdb
2615 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
2616 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
2620 Release 3.7.0 (5 November 2011)
2621 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2622 3.7.0 is a feature release with many significant improvements and the
2623 usual collection of bug fixes.
2625 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2626 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
2627 Support for recent distros and toolchain components (glibc 2.14, gcc
2628 4.6, MacOSX 10.7) has been added.
2630 * ================== PLATFORM CHANGES =================
2632 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
2633 analyse 64-bit programs running on z/Architecture. Most user space
2634 instructions up to and including z10 are supported. Valgrind has
2635 been tested extensively on z9, z10, and z196 machines running SLES
2636 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
2637 known to work well. Callgrind, Helgrind, and DRD work reasonably
2638 well on z9 and later models. See README.s390 for more details.
2640 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
2641 64-bit processes are supported. Some complex threaded applications
2642 (Firefox) are observed to hang when run as 32 bit applications,
2643 whereas 64-bit versions run OK. The cause is unknown. Memcheck
2644 will likely report some false errors. In general, expect some rough
2645 spots. This release also supports MacOSX 10.6, but drops support
2648 * Preliminary support for Android (on ARM). Valgrind can now run
2649 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
2650 README.android for more details, plus instructions on how to get
2653 * Support for the IBM Power ISA 2.06 (Power7 instructions)
2655 * General correctness and performance improvements for ARM/Linux, and,
2656 by extension, ARM/Android.
2658 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
2659 instruction set support is under development but is not available in
2662 * Support for AIX5 has been removed.
2664 * ==================== TOOL CHANGES ====================
2666 * Memcheck: some incremental changes:
2668 - reduction of memory use in some circumstances
2670 - improved handling of freed memory, which in some circumstances
2671 can cause detection of use-after-free that would previously have
2674 - fix of a longstanding bug that could cause false negatives (missed
2675 errors) in programs doing vector saturated narrowing instructions.
2677 * Helgrind: performance improvements and major memory use reductions,
2678 particularly for large, long running applications which perform many
2679 synchronisation (lock, unlock, etc) events. Plus many smaller
2682 - display of locksets for both threads involved in a race
2684 - general improvements in formatting/clarity of error messages
2686 - addition of facilities and documentation regarding annotation
2687 of thread safe reference counted C++ classes
2689 - new flag --check-stack-refs=no|yes [yes], to disable race checking
2690 on thread stacks (a performance hack)
2692 - new flag --free-is-write=no|yes [no], to enable detection of races
2693 where one thread accesses heap memory but another one frees it,
2694 without any coordinating synchronisation event
2696 * DRD: enabled XML output; added support for delayed thread deletion
2697 in order to detect races that occur close to the end of a thread
2698 (--join-list-vol); fixed a memory leak triggered by repeated client
2699 memory allocatation and deallocation; improved Darwin support.
2701 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
2703 * exp-sgcheck: this tool has been reduced in scope so as to improve
2704 performance and remove checking that Memcheck does better.
2705 Specifically, the ability to check for overruns for stack and global
2706 arrays is unchanged, but the ability to check for overruns of heap
2707 blocks has been removed. The tool has accordingly been renamed to
2708 exp-sgcheck ("Stack and Global Array Checking").
2710 * ==================== OTHER CHANGES ====================
2712 * GDB server: Valgrind now has an embedded GDB server. That means it
2713 is possible to control a Valgrind run from GDB, doing all the usual
2714 things that GDB can do (single stepping, breakpoints, examining
2715 data, etc). Tool-specific functionality is also available. For
2716 example, it is possible to query the definedness state of variables
2717 or memory from within GDB when running Memcheck; arbitrarily large
2718 memory watchpoints are supported, etc. To use the GDB server, start
2719 Valgrind with the flag --vgdb-error=0 and follow the on-screen
2722 * Improved support for unfriendly self-modifying code: a new option
2723 --smc-check=all-non-file is available. This adds the relevant
2724 consistency checks only to code that originates in non-file-backed
2725 mappings. In effect this confines the consistency checking only to
2726 code that is or might be JIT generated, and avoids checks on code
2727 that must have been compiled ahead of time. This significantly
2728 improves performance on applications that generate code at run time.
2730 * It is now possible to build a working Valgrind using Clang-2.9 on
2733 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
2734 These enable and disable error reporting on a per-thread, and
2735 nestable, basis. This is useful for hiding errors in particularly
2736 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
2737 now uses this facility.
2739 * Added the --mod-funcname option to cg_diff.
2741 * ==================== FIXED BUGS ====================
2743 The following bugs have been fixed or resolved. Note that "n-i-bz"
2744 stands for "not in bugzilla" -- that is, a bug that was reported to us
2745 but never got a bugzilla entry. We encourage you to file bugs in
2746 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2747 mailing the developers (or mailing lists) directly -- bugs that are
2748 not entered into bugzilla tend to get forgotten about or ignored.
2750 To see details of a given bug, visit
2751 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2752 where XXXXXX is the bug number as listed below.
2754 79311 malloc silly arg warning does not give stack trace
2755 210935 port valgrind.h (not valgrind) to win32 to support client requests
2756 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
2757 243404 Port to zSeries
2758 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
2759 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
2760 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
2761 253206 Some fixes for the faultstatus testcase
2762 255223 capget testcase fails when running as root
2763 256703 xlc_dbl_u32.c testcase broken
2764 256726 Helgrind tests have broken inline asm
2765 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
2766 264800 testcase compile failure on zseries
2767 265762 make public VEX headers compilable by G++ 3.x
2768 265771 assertion in jumps.c (r11523) fails with glibc-2.3
2769 266753 configure script does not give the user the option to not use QtCore
2770 266931 gen_insn_test.pl is broken
2771 266961 ld-linux.so.2 i?86-linux strlen issues
2772 266990 setns instruction causes false positive
2773 267020 Make directory for temporary files configurable at run-time.
2774 267342 == 267997 (segmentation fault on Mac OS 10.6)
2775 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
2776 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
2777 267488 regtest: darwin support for 64-bit build
2778 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
2779 267630 Add support for IBM Power ISA 2.06 -- stage 1
2780 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
2781 267819 Add client request for informing the core about reallocation
2782 267925 laog data structure quadratic for a single sequence of lock
2783 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
2784 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
2785 268513 missed optimizations in fold_Expr
2786 268619 s390x: fpr - gpr transfer facility
2787 268620 s390x: reconsider "long displacement" requirement
2788 268621 s390x: improve IR generation for XC
2789 268715 s390x: FLOGR is not universally available
2790 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
2791 268930 s390x: MHY is not universally available
2792 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
2793 269079 Support ptrace system call on ARM
2794 269144 missing "Bad option" error message
2795 269209 conditional load and store facility (z196)
2796 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
2797 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
2798 269736 s390x: minor code generation tweaks
2799 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
2800 269863 s390x: remove unused function parameters
2801 269864 s390x: tweak s390_emit_load_cc
2802 269884 == 250101 (overhead for huge blocks exhausts space too soon)
2803 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
2804 270115 s390x: rewrite some testcases
2805 270309 == 267997 (valgrind crash on startup)
2806 270320 add support for Linux FIOQSIZE ioctl() call
2807 270326 segfault while trying to sanitize the environment passed to execle
2808 270794 IBM POWER7 support patch causes regression in none/tests
2809 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
2810 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
2811 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
2812 270959 s390x: invalid use of R0 as base register
2813 271042 VSX configure check fails when it should not
2814 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
2815 271259 s390x: fix code confusion
2816 271337 == 267997 (Valgrind segfaults on MacOS X)
2817 271385 s390x: Implement Ist_MBE
2818 271501 s390x: misc cleanups
2819 271504 s390x: promote likely and unlikely
2820 271579 ppc: using wrong enum type
2821 271615 unhandled instruction "popcnt" (arch=amd10h)
2822 271730 Fix bug when checking ioctls: duplicate check
2823 271776 s390x: provide STFLE instruction support
2824 271779 s390x: provide clock instructions like STCK
2825 271799 Darwin: ioctls without an arg report a memory error
2826 271820 arm: fix type confusion
2827 271917 pthread_cond_timedwait failure leads to not-locked false positive
2828 272067 s390x: fix DISP20 macro
2829 272615 A typo in debug output in mc_leakcheck.c
2830 272661 callgrind_annotate chokes when run from paths containing regex chars
2831 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
2832 272955 Unhandled syscall error for pwrite64 on ppc64 arch
2833 272967 make documentation build-system more robust
2834 272986 Fix gcc-4.6 warnings with valgrind.h
2835 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
2836 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
2837 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
2838 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
2839 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
2840 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
2841 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
2842 273778 exp-ptrcheck: unhandled sysno == 259
2843 274089 exp-ptrcheck: unhandled sysno == 208
2844 274378 s390x: Various dispatcher tweaks
2845 274447 WARNING: unhandled syscall: 340
2846 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
2847 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
2848 274926 valgrind does not build against linux-3
2849 275148 configure FAIL with glibc-2.14
2850 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
2851 275168 Make Valgrind work for MacOSX 10.7 Lion
2852 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
2853 275278 valgrind does not build on Linux kernel 3.0.* due to silly
2854 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
2855 275308 Fix implementation for ppc64 fres instruc
2856 275339 s390x: fix testcase compile warnings
2857 275517 s390x: Provide support for CKSM instruction
2858 275710 s390x: get rid of redundant address mode calculation
2859 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
2860 275852 == 250101 (valgrind uses all swap space and is killed)
2861 276784 Add support for IBM Power ISA 2.06 -- stage 3
2862 276987 gdbsrv: fix tests following recent commits
2863 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
2864 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
2865 277471 Unhandled syscall: 340
2866 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
2867 277653 ARM: support Thumb2 PLD instruction
2868 277663 ARM: NEON float VMUL by scalar incorrect
2869 277689 ARM: tests for VSTn with register post-index are broken
2870 277694 ARM: BLX LR instruction broken in ARM mode
2871 277780 ARM: VMOV.F32 (immediate) instruction is broken
2872 278057 fuse filesystem syscall deadlocks
2873 278078 Unimplemented syscall 280 on ppc32
2874 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
2875 278454 VALGRIND_STACK_DEREGISTER has wrong output type
2876 278502 == 275284 (Valgrind confuses memcpy() and memmove())
2877 278892 gdbsrv: factorize gdb version handling, fix doc and typos
2878 279027 Support for MVCL and CLCL instruction
2879 279027 s390x: Provide support for CLCL and MVCL instructions
2880 279062 Remove a redundant check in the insn selector for ppc.
2881 279071 JDK creates PTEST with redundant REX.W prefix
2882 279212 gdbsrv: add monitor cmd v.info scheduler.
2883 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
2884 279698 memcheck discards valid-bits for packuswb
2885 279795 memcheck reports uninitialised values for mincore on amd64
2886 279994 Add support for IBM Power ISA 2.06 -- stage 3
2887 280083 mempolicy syscall check errors
2888 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
2889 280710 s390x: config files for nightly builds
2890 280757 /tmp dir still used by valgrind even if TMPDIR is specified
2891 280965 Valgrind breaks fcntl locks when program does mmap
2892 281138 WARNING: unhandled syscall: 340
2893 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
2894 281304 == 275168 (Darwin: dyld "cannot load inserted library")
2895 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
2896 281468 s390x: handle do_clone and gcc clones in call traces
2897 281488 ARM: VFP register corruption
2898 281828 == 275284 (false memmove warning: "Source and destination overlap")
2899 281883 s390x: Fix system call wrapper for "clone".
2900 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
2901 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
2902 282238 SLES10: make check fails
2903 282979 strcasestr needs replacement with recent(>=2.12) glibc
2904 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
2905 283243 Regression in ppc64 memcheck tests
2906 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
2907 283427 re-connect epoll_pwait syscall on ARM linux
2908 283600 gdbsrv: android: port vgdb.c
2909 283709 none/tests/faultstatus needs to account for page size
2910 284305 filter_gdb needs enhancement to work on ppc64
2911 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
2912 284472 Thumb2 ROR.W encoding T2 not implemented
2913 284621 XML-escape process command line in XML output
2914 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
2915 that have non-power-of-2 sizes (also AMDs)
2916 n-i-bz don't be spooked by libraries mashed by elfhack
2917 n-i-bz don't be spooked by libxul.so linked with gold
2918 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
2920 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
2921 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
2922 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
2926 Release 3.6.1 (16 February 2011)
2927 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2928 3.6.1 is a bug fix release. It adds support for some SSE4
2929 instructions that were omitted in 3.6.0 due to lack of time. Initial
2930 support for glibc-2.13 has been added. A number of bugs causing
2931 crashing or assertion failures have been fixed.
2933 The following bugs have been fixed or resolved. Note that "n-i-bz"
2934 stands for "not in bugzilla" -- that is, a bug that was reported to us
2935 but never got a bugzilla entry. We encourage you to file bugs in
2936 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2937 mailing the developers (or mailing lists) directly -- bugs that are
2938 not entered into bugzilla tend to get forgotten about or ignored.
2940 To see details of a given bug, visit
2941 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2942 where XXXXXX is the bug number as listed below.
2944 188572 Valgrind on Mac should suppress setenv() mem leak
2945 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
2946 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
2947 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
2948 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
2949 254420 memory pool tracking broken
2950 254957 Test code failing to compile due to changes in memcheck.h
2951 255009 helgrind/drd: crash on chmod with invalid parameter
2952 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
2953 255355 helgrind/drd: crash on threaded programs doing fork
2955 255418 (SSE4.x) rint call compiled with ICC
2956 255822 --gen-suppressions can create invalid files: "too many callers [...]"
2957 255888 closing valgrindoutput tag outputted to log-stream on error
2958 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
2959 255966 Slowness when using mempool annotations
2960 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
2961 256600 super-optimized strcasecmp() false positive
2962 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
2963 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
2964 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
2965 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
2966 257276 Missing case in memcheck --track-origins=yes
2967 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
2968 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
2969 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
2970 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
2971 263099 callgrind_annotate counts Ir improperly [...]
2972 263877 undefined coprocessor instruction on ARMv7
2973 265964 configure FAIL with glibc-2.13
2974 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
2975 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
2976 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
2977 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
2978 n-i-bz DRD: disable --free-is-write due to implementation difficulties
2980 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
2984 Release 3.6.0 (21 October 2010)
2985 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2986 3.6.0 is a feature release with many significant improvements and the
2987 usual collection of bug fixes.
2989 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2990 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
2991 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
2993 -------------------------
2995 Here are some highlights. Details are shown further down:
2997 * Support for ARM/Linux.
2999 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
3001 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
3003 * Support for the SSE4.2 instruction set.
3005 * Enhancements to the Callgrind profiler, including the ability to
3006 handle CPUs with three levels of cache.
3008 * A new experimental heap profiler, DHAT.
3010 * A huge number of bug fixes and small enhancements.
3012 -------------------------
3014 Here are details of the above changes, together with descriptions of
3015 many other changes, and a list of fixed bugs.
3017 * ================== PLATFORM CHANGES =================
3019 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
3020 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
3021 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
3023 This requires a CPU capable of running the ARMv7-A instruction set
3024 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
3025 of the user space instruction set, including ARM and Thumb integer
3026 code, VFPv3, NEON and V6 media instructions. The Memcheck,
3027 Cachegrind and Massif tools work properly; other tools work to
3030 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
3031 with support for recent releases of the underlying toolchain
3032 components, notably gcc-4.5 and glibc-2.12.
3034 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
3035 support also works much better on OS X 10.5, and is as solid as
3038 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
3039 64-bit mode. In 32-bit mode, support is only available up to and
3040 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
3041 supported in 64-bit mode, and 32-bit mode does in fact support the
3042 bare minimum SSE4 instructions to needed to run programs on Mac OS X
3043 10.6 on 32-bit targets.
3045 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
3046 and including version 2.05 is supported.
3048 * ==================== TOOL CHANGES ====================
3050 * Cachegrind has a new processing script, cg_diff, which finds the
3051 difference between two profiles. It's very useful for evaluating
3052 the performance effects of a change in a program.
3054 Related to this change, the meaning of cg_annotate's (rarely-used)
3055 --threshold option has changed; this is unlikely to affect many
3056 people, if you do use it please see the user manual for details.
3058 * Callgrind now can do branch prediction simulation, similar to
3059 Cachegrind. In addition, it optionally can count the number of
3060 executed global bus events. Both can be used for a better
3061 approximation of a "Cycle Estimation" as derived event (you need to
3062 update the event formula in KCachegrind yourself).
3064 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3065 rather than the L2 cache. This is to accommodate machines with
3066 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3067 cache configuration of such a machine it will run the simulation as
3068 if the L2 cache isn't present. This means the results are less
3069 likely to match the true result for the machine, but
3070 Cachegrind/Callgrind's results are already only approximate, and
3071 should not be considered authoritative. The results are still
3072 useful for giving a general idea about a program's locality.
3074 * Massif has a new option, --pages-as-heap, which is disabled by
3075 default. When enabled, instead of tracking allocations at the level
3076 of heap blocks (as allocated with malloc/new/new[]), it instead
3077 tracks memory allocations at the level of memory pages (as mapped by
3078 mmap, brk, etc). Each mapped page is treated as its own block.
3079 Interpreting the page-level output is harder than the heap-level
3080 output, but this option is useful if you want to account for every
3081 byte of memory used by a program.
3083 * DRD has two new command-line options: --free-is-write and
3084 --trace-alloc. The former allows to detect reading from already freed
3085 memory, and the latter allows tracing of all memory allocations and
3088 * DRD has several new annotations. Custom barrier implementations can
3089 now be annotated, as well as benign races on static variables.
3091 * DRD's happens before / happens after annotations have been made more
3092 powerful, so that they can now also be used to annotate e.g. a smart
3093 pointer implementation.
3095 * Helgrind's annotation set has also been drastically improved, so as
3096 to provide to users a general set of annotations to describe locks,
3097 semaphores, barriers and condition variables. Annotations to
3098 describe thread-safe reference counted heap objects have also been
3101 * Memcheck has a new command-line option, --show-possibly-lost, which
3102 is enabled by default. When disabled, the leak detector will not
3103 show possibly-lost blocks.
3105 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3106 has been added. DHAT keeps track of allocated heap blocks, and also
3107 inspects every memory reference to see which block (if any) is being
3108 accessed. This gives a lot of insight into block lifetimes,
3109 utilisation, turnover, liveness, and the location of hot and cold
3110 fields. You can use DHAT to do hot-field profiling.
3112 * ==================== OTHER CHANGES ====================
3114 * Improved support for unfriendly self-modifying code: the extra
3115 overhead incurred by --smc-check=all has been reduced by
3116 approximately a factor of 5 as compared with 3.5.0.
3118 * Ability to show directory names for source files in error messages.
3119 This is combined with a flexible mechanism for specifying which
3120 parts of the paths should be shown. This is enabled by the new flag
3123 * A new flag, --require-text-symbol, which will stop the run if a
3124 specified symbol is not found it a given shared object when it is
3125 loaded into the process. This makes advanced working with function
3126 intercepting and wrapping safer and more reliable.
3128 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3129 and control of Valgrind is now available for the tools Memcheck and
3130 Helgrind. XML output from Valgrind is available for Memcheck,
3131 Helgrind and exp-Ptrcheck.
3133 * More reliable stack unwinding on amd64-linux, particularly in the
3134 presence of function wrappers, and with gcc-4.5 compiled code.
3136 * Modest scalability (performance improvements) for massive
3137 long-running applications, particularly for those with huge amounts
3140 * Support for analyzing programs running under Wine with has been
3141 improved. The header files <valgrind/valgrind.h>,
3142 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3143 Windows-programs compiled with MinGW or one of the Microsoft Visual
3146 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3147 The 32-bit simulator was not affected. This did not occur often,
3148 but when it did would usually crash the program under test.
3151 * A large number of bugs were fixed. These are shown below.
3153 * A number of bugs were investigated, and were candidates for fixing,
3154 but are not fixed in 3.6.0, due to lack of developer time. They may
3155 get fixed in later releases. They are:
3157 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3158 212419 false positive "lock order violated" (A+B vs A)
3159 213685 Undefined value propagates past dependency breaking instruction
3160 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3161 237920 valgrind segfault on fork failure
3162 242137 support for code compiled by LLVM-2.8
3163 242423 Another unknown Intel cache config value
3164 243232 Inconsistent Lock Orderings report with trylock
3165 243483 ppc: callgrind triggers VEX assertion failure
3166 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3167 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3169 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3170 249435 Analyzing wine programs with callgrind triggers a crash
3171 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3172 250065 Handling large allocations
3173 250101 huge "free" memory usage due to m_mallocfree.c
3174 "superblocks fragmentation"
3175 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3176 252091 Callgrind on ARM does not detect function returns correctly
3177 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3178 254420 memory pool tracking broken
3179 n-i-bz support for adding symbols for JIT generated code
3182 The following bugs have been fixed or resolved. Note that "n-i-bz"
3183 stands for "not in bugzilla" -- that is, a bug that was reported to us
3184 but never got a bugzilla entry. We encourage you to file bugs in
3185 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3186 mailing the developers (or mailing lists) directly -- bugs that are
3187 not entered into bugzilla tend to get forgotten about or ignored.
3189 To see details of a given bug, visit
3190 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3191 where XXXXXX is the bug number as listed below.
3193 135264 dcbzl instruction missing
3195 153699 Valgrind should report unaligned reads with movdqa
3197 190429 Valgrind reports lost of errors in ld.so
3198 with x86_64 2.9.90 glibc
3199 197266 valgrind appears to choke on the xmms instruction
3201 197988 Crash when demangling very large symbol names
3202 202315 unhandled syscall: 332 (inotify_init1)
3203 203256 Add page-level profiling to Massif
3204 205093 dsymutil=yes needs quotes, locking (partial fix)
3205 205241 Snow Leopard 10.6 support (partial fix)
3206 206600 Leak checker fails to upgrade indirect blocks when their
3207 parent becomes reachable
3208 210935 port valgrind.h (not valgrind) to win32 so apps run under
3209 wine can make client requests
3210 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3211 within Linux ip-stack checksum functions
3212 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3214 213685 Undefined value propagates past dependency breaking instruction
3216 215914 Valgrind inserts bogus empty environment variable
3218 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3219 222545 shmat fails under valgind on some arm targets
3220 222560 ARM NEON support
3223 232509 Docs build fails with formatting inside <title></title> elements
3225 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3226 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3227 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3228 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3229 237485 symlink (syscall 57) is not supported on Mac OS
3230 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3232 238208 is_just_below_ESP doesn't take into account red-zone
3233 238345 valgrind passes wrong $0 when executing a shell script
3234 238679 mq_timedreceive syscall doesn't flag the reception buffer
3236 238696 fcntl command F_DUPFD_CLOEXEC not supported
3237 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3238 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3239 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
3241 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
3247 242606 unhandled syscall: setegid (in Ptrcheck)
3248 242814 Helgrind "Impossible has happened" during
3249 QApplication::initInstance();
3250 243064 Valgrind attempting to read debug information from iso
3251 243270 Make stack unwinding in Valgrind wrappers more reliable
3252 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
3253 sysno = 277 (mq_open)
3254 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
3255 244493 ARM VFP d16-d31 registers support
3256 244670 add support for audit_session_self syscall on Mac OS 10.6
3257 244921 The xml report of helgrind tool is not well format
3258 244923 In the xml report file, the <preamble> not escape the
3259 xml char, eg '<','&','>'
3260 245535 print full path names in plain text reports
3261 245925 x86-64 red zone handling problem
3262 246258 Valgrind not catching integer underruns + new [] s
3263 246311 reg/reg cmpxchg doesn't work on amd64
3264 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
3265 246888 Improve Makefile.vex.am
3266 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
3267 to [f]chmod_extended
3268 247526 IBM POWER6 (ISA 2.05) support is incomplete
3269 247561 Some leak testcases fails due to reachable addresses in
3271 247875 sizeofIRType to handle Ity_I128
3272 247894 [PATCH] unhandled syscall sys_readahead
3273 247980 Doesn't honor CFLAGS passed to configure
3274 248373 darwin10.supp is empty in the trunk
3275 248822 Linux FIBMAP ioctl has int parameter instead of long
3276 248893 [PATCH] make readdwarf.c big endianess safe to enable
3277 unwinding on big endian systems
3278 249224 Syscall 336 not supported (SYS_proc_info)
3280 249775 Incorrect scheme for detecting NEON capabilities of host CPU
3281 249943 jni JVM init fails when using valgrind
3282 249991 Valgrind incorrectly declares AESKEYGENASSIST support
3284 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
3285 250799 frexp$fenv_access_off function generates SIGILL
3286 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
3287 251251 support pclmulqdq insn
3288 251362 valgrind: ARM: attach to debugger either fails or provokes
3290 251674 Unhandled syscall 294
3293 254257 Add support for debugfiles found by build-id
3294 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
3295 254646 Wrapped functions cause stack misalignment on OS X
3296 (and possibly Linux)
3297 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
3299 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
3303 Release 3.5.0 (19 August 2009)
3304 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3305 3.5.0 is a feature release with many significant improvements and the
3306 usual collection of bug fixes. The main improvement is that Valgrind
3307 now works on Mac OS X.
3309 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
3310 and X86/Darwin. Support for recent distros and toolchain components
3311 (glibc 2.10, gcc 4.5) has been added.
3313 -------------------------
3315 Here is a short summary of the changes. Details are shown further
3318 * Support for Mac OS X (10.5.x).
3320 * Improvements and simplifications to Memcheck's leak checker.
3322 * Clarification and simplifications in various aspects of Valgrind's
3325 * XML output for Helgrind and Ptrcheck.
3327 * Performance and stability improvements for Helgrind and DRD.
3329 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
3331 * A new experimental tool, BBV, useful for computer architecture
3334 * Improved Wine support, including ability to read Windows PDB
3337 -------------------------
3339 Here are details of the above changes, followed by descriptions of
3340 many other minor changes, and a list of fixed bugs.
3343 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
3344 called "Darwin" because that is the name of the OS core, which is the
3345 level that Valgrind works at.)
3349 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
3350 because it would require work and 10.4 is only becoming less common.
3352 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
3353 fairly well. For 10.5.x, 32-bit programs are the default even on
3354 64-bit machines, so it handles most current programs.
3356 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
3357 officially supported, but simple programs at least will probably work.
3358 However, start-up is slow.
3360 - PowerPC machines are not supported.
3362 Things that don't work:
3364 - The Ptrcheck tool.
3366 - Objective-C garbage collection.
3370 - If you have Rogue Amoeba's "Instant Hijack" program installed,
3371 Valgrind will fail with a SIGTRAP at start-up. See
3372 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
3377 - You will likely find --dsymutil=yes a useful option, as error
3378 messages may be imprecise without it.
3380 - Mac OS X support is new and therefore will be less robust than the
3381 Linux support. Please report any bugs you find.
3383 - Threaded programs may run more slowly than on Linux.
3385 Many thanks to Greg Parker for developing this port over several years.
3388 * Memcheck's leak checker has been improved.
3390 - The results for --leak-check=summary now match the summary results
3391 for --leak-check=full. Previously they could differ because
3392 --leak-check=summary counted "indirectly lost" blocks and
3393 "suppressed" blocks as "definitely lost".
3395 - Blocks that are only reachable via at least one interior-pointer,
3396 but are directly pointed to by a start-pointer, were previously
3397 marked as "still reachable". They are now correctly marked as
3400 - The default value for the --leak-resolution option has been
3401 changed from "low" to "high". In general, this means that more
3402 leak reports will be produced, but each leak report will describe
3403 fewer leaked blocks.
3405 - With --leak-check=full, "definitely lost" and "possibly lost"
3406 leaks are now considered as proper errors, ie. they are counted
3407 for the "ERROR SUMMARY" and affect the behaviour of
3408 --error-exitcode. These leaks are not counted as errors if
3409 --leak-check=summary is specified, however.
3411 - Documentation for the leak checker has been improved.
3414 * Various aspects of Valgrind's text output have changed.
3416 - Valgrind's start-up message has changed. It is shorter but also
3417 includes the command being run, which makes it easier to use
3418 --trace-children=yes. An example:
3420 - Valgrind's shut-down messages have also changed. This is most
3421 noticeable with Memcheck, where the leak summary now occurs before
3422 the error summary. This change was necessary to allow leaks to be
3423 counted as proper errors (see the description of the leak checker
3424 changes above for more details). This was also necessary to fix a
3425 longstanding bug in which uses of suppressions against leaks were
3426 not "counted", leading to difficulties in maintaining suppression
3427 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3429 - Behavior of -v has changed. In previous versions, -v printed out
3430 a mixture of marginally-user-useful information, and tool/core
3431 statistics. The statistics printing has now been moved to its own
3432 flag, --stats=yes. This means -v is less verbose and more likely
3433 to convey useful end-user information.
3435 - The format of some (non-XML) stack trace entries has changed a
3436 little. Previously there were six possible forms:
3438 0x80483BF: really (a.c:20)
3439 0x80483BF: really (in /foo/a.out)
3441 0x80483BF: (within /foo/a.out)
3442 0x80483BF: ??? (a.c:20)
3445 The third and fourth of these forms have been made more consistent
3446 with the others. The six possible forms are now:
3448 0x80483BF: really (a.c:20)
3449 0x80483BF: really (in /foo/a.out)
3450 0x80483BF: really (in ???)
3451 0x80483BF: ??? (in /foo/a.out)
3452 0x80483BF: ??? (a.c:20)
3455 Stack traces produced when --xml=yes is specified are different
3459 * Helgrind and Ptrcheck now support XML output, so they can be used
3460 from GUI tools. Also, the XML output mechanism has been
3463 - The XML format has been overhauled and generalised, so it is more
3464 suitable for error reporting tools in general. The Memcheck
3465 specific aspects of it have been removed. The new format, which
3466 is an evolution of the old format, is described in
3467 docs/internals/xml-output-protocol4.txt.
3469 - Memcheck has been updated to use the new format.
3471 - Helgrind and Ptrcheck are now able to emit output in this format.
3473 - The XML output mechanism has been overhauled. XML is now output
3474 to its own file descriptor, which means that:
3476 * Valgrind can output text and XML independently.
3478 * The longstanding problem of XML output being corrupted by
3479 unexpected un-tagged text messages is solved.
3481 As before, the destination for text output is specified using
3482 --log-file=, --log-fd= or --log-socket=.
3484 As before, XML output for a tool is enabled using --xml=yes.
3486 Because there's a new XML output channel, the XML output
3487 destination is now specified by --xml-file=, --xml-fd= or
3490 Initial feedback has shown this causes some confusion. To
3491 clarify, the two envisaged usage scenarios are:
3493 (1) Normal text output. In this case, do not specify --xml=yes
3494 nor any of --xml-file=, --xml-fd= or --xml-socket=.
3496 (2) XML output. In this case, specify --xml=yes, and one of
3497 --xml-file=, --xml-fd= or --xml-socket= to select the XML
3498 destination, one of --log-file=, --log-fd= or --log-socket=
3499 to select the destination for any remaining text messages,
3500 and, importantly, -q.
3502 -q makes Valgrind completely silent on the text channel,
3503 except in the case of critical failures, such as Valgrind
3504 itself segfaulting, or failing to read debugging information.
3505 Hence, in this scenario, it suffices to check whether or not
3506 any output appeared on the text channel. If yes, then it is
3507 likely to be a critical error which should be brought to the
3508 attention of the user. If no (the text channel produced no
3509 output) then it can be assumed that the run was successful.
3511 This allows GUIs to make the critical distinction they need to
3512 make (did the run fail or not?) without having to search or
3513 filter the text output channel in any way.
3515 It is also recommended to use --child-silent-after-fork=yes in
3519 * Improvements and changes in Helgrind:
3521 - XML output, as described above
3523 - Checks for consistent association between pthread condition
3524 variables and their associated mutexes are now performed.
3526 - pthread_spinlock functions are supported.
3528 - Modest performance improvements.
3530 - Initial (skeletal) support for describing the behaviour of
3531 non-POSIX synchronisation objects through ThreadSanitizer
3532 compatible ANNOTATE_* macros.
3534 - More controllable tradeoffs between performance and the level of
3535 detail of "previous" accesses in a race. There are now three
3538 * --history-level=full. This is the default, and was also the
3539 default in 3.4.x. It shows both stacks involved in a race, but
3540 requires a lot of memory and can be very slow in programs that
3541 do many inter-thread synchronisation events.
3543 * --history-level=none. This only shows the later stack involved
3544 in a race. This can be much faster than --history-level=full,
3545 but makes it much more difficult to find the other access
3546 involved in the race.
3548 The new intermediate setting is
3550 * --history-level=approx
3552 For the earlier (other) access, two stacks are presented. The
3553 earlier access is guaranteed to be somewhere in between the two
3554 program points denoted by those stacks. This is not as useful
3555 as showing the exact stack for the previous access (as per
3556 --history-level=full), but it is better than nothing, and it's
3557 almost as fast as --history-level=none.
3560 * New features and improvements in DRD:
3562 - The error messages printed by DRD are now easier to interpret.
3563 Instead of using two different numbers to identify each thread
3564 (Valgrind thread ID and DRD thread ID), DRD does now identify
3565 threads via a single number (the DRD thread ID). Furthermore
3566 "first observed at" information is now printed for all error
3567 messages related to synchronization objects.
3569 - Added support for named semaphores (sem_open() and sem_close()).
3571 - Race conditions between pthread_barrier_wait() and
3572 pthread_barrier_destroy() calls are now reported.
3574 - Added support for custom allocators through the macros
3575 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
3576 in <valgrind/valgrind.h>). An alternative for these two macros is
3577 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
3580 - Added support for annotating non-POSIX synchronization objects
3581 through several new ANNOTATE_*() macros.
3583 - OpenMP: added support for the OpenMP runtime (libgomp) included
3584 with gcc versions 4.3.0 and 4.4.0.
3588 - Added two new command-line options (--first-race-only and
3589 --segment-merging-interval).
3592 * Genuinely atomic support for x86/amd64/ppc atomic instructions
3594 Valgrind will now preserve (memory-access) atomicity of LOCK-
3595 prefixed x86/amd64 instructions, and any others implying a global
3596 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
3598 This means that Valgrinded processes will "play nicely" in
3599 situations where communication with other processes, or the kernel,
3600 is done through shared memory and coordinated with such atomic
3601 instructions. Prior to this change, such arrangements usually
3602 resulted in hangs, races or other synchronisation failures, because
3603 Valgrind did not honour atomicity of such instructions.
3606 * A new experimental tool, BBV, has been added. BBV generates basic
3607 block vectors for use with the SimPoint analysis tool, which allows
3608 a program's overall behaviour to be approximated by running only a
3609 fraction of it. This is useful for computer architecture
3610 researchers. You can run BBV by specifying --tool=exp-bbv (the
3611 "exp-" prefix is short for "experimental"). BBV was written by
3615 * Modestly improved support for running Windows applications under
3616 Wine. In particular, initial support for reading Windows .PDB debug
3617 information has been added.
3620 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
3621 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
3625 * The Valgrind client requests VALGRIND_PRINTF and
3626 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
3627 the string was always printed immediately on its own line. Now, the
3628 string will be added to a buffer but not printed until a newline is
3629 encountered, or other Valgrind output is printed (note that for
3630 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
3631 "other Valgrind output"). This allows you to use multiple
3632 VALGRIND_PRINTF calls to build up a single output line, and also to
3633 print multiple output lines with a single request (by embedding
3634 multiple newlines in the string).
3637 * The graphs drawn by Massif's ms_print program have changed slightly:
3639 - The half-height chars '.' and ',' are no longer drawn, because
3640 they are confusing. The --y option can be used if the default
3641 y-resolution is not high enough.
3643 - Horizontal lines are now drawn after the top of a snapshot if
3644 there is a gap until the next snapshot. This makes it clear that
3645 the memory usage has not dropped to zero between snapshots.
3648 * Something that happened in 3.4.0, but wasn't clearly announced: the
3649 option --read-var-info=yes can be used by some tools (Memcheck,
3650 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
3651 variable type and location information. This makes those tools
3652 start up more slowly and increases memory consumption, but
3653 descriptions of data addresses in error messages become more
3657 * exp-Omega, an experimental instantaneous leak-detecting tool, was
3658 disabled in 3.4.0 due to a lack of interest and maintenance,
3659 although the source code was still in the distribution. The source
3660 code has now been removed from the distribution. For anyone
3661 interested, the removal occurred in SVN revision r10247.
3664 * Some changes have been made to the build system.
3666 - VEX/ is now integrated properly into the build system. This means
3667 that dependency tracking within VEX/ now works properly, "make
3668 install" will work without requiring "make" before it, and
3669 parallel builds (ie. 'make -j') now work (previously a
3670 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
3671 was effectively ignored).
3673 - The --with-vex configure option has been removed. It was of
3674 little use and removing it simplified the build system.
3676 - The location of some install files has changed. This should not
3677 affect most users. Those who might be affected:
3679 * For people who use Valgrind with MPI programs, the installed
3680 libmpiwrap.so library has moved from
3681 $(INSTALL)/<platform>/libmpiwrap.so to
3682 $(INSTALL)/libmpiwrap-<platform>.so.
3684 * For people who distribute standalone Valgrind tools, the
3685 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
3686 have moved to $(INSTALL)/libcoregrind-<platform>.a.
3688 These changes simplify the build system.
3690 - Previously, all the distributed suppression (*.supp) files were
3691 installed. Now, only default.supp is installed. This should not
3692 affect users as the other installed suppression files were not
3693 read; the fact that they were installed was a mistake.
3696 * KNOWN LIMITATIONS:
3698 - Memcheck is unusable with the Intel compiler suite version 11.1,
3699 when it generates code for SSE2-and-above capable targets. This
3700 is because of icc's use of highly optimised inlined strlen
3701 implementations. It causes Memcheck to report huge numbers of
3702 false errors even in simple programs. Helgrind and DRD may also
3705 Versions 11.0 and earlier may be OK, but this has not been
3709 The following bugs have been fixed or resolved. Note that "n-i-bz"
3710 stands for "not in bugzilla" -- that is, a bug that was reported to us
3711 but never got a bugzilla entry. We encourage you to file bugs in
3712 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3713 mailing the developers (or mailing lists) directly -- bugs that are
3714 not entered into bugzilla tend to get forgotten about or ignored.
3716 To see details of a given bug, visit
3717 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3718 where XXXXXX is the bug number as listed below.
3720 84303 How about a LockCheck tool?
3721 91633 dereference of null ptr in vgPlain_st_basetype
3722 97452 Valgrind doesn't report any pthreads problems
3723 100628 leak-check gets assertion failure when using
3724 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
3725 108528 NPTL pthread cleanup handlers not called
3726 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
3727 110128 mallinfo is not implemented...
3728 110770 VEX: Generated files not always updated when making valgrind
3729 111102 Memcheck: problems with large (memory footprint) applications
3730 115673 Vex's decoder should never assert
3731 117564 False positive: Syscall param clone(child_tidptr) contains
3732 uninitialised byte(s)
3733 119404 executing ssh from inside valgrind fails
3734 133679 Callgrind does not write path names to sources with dwarf debug
3736 135847 configure.in problem with non gnu compilers (and possible fix)
3737 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
3738 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
3739 136230 memcheck reports "possibly lost", should be "still reachable"
3740 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
3741 137904 Valgrind reports a memory leak when using POSIX threads,
3743 139076 valgrind VT_GETSTATE error
3744 142228 complaint of elf_dynamic_do_rela in trivial usage
3745 145347 spurious warning with USBDEVFS_REAPURB
3746 148441 (wine) can't find memory leak in Wine, win32 binary
3748 148742 Leak-check fails assert on exit
3749 149878 add (proper) check for calloc integer overflow
3750 150606 Call graph is broken when using callgrind control
3751 152393 leak errors produce an exit code of 0. I need some way to
3752 cause leak errors to result in a nonzero exit code.
3753 157154 documentation (leak-resolution doc speaks about num-callers
3754 def=4) + what is a loss record
3755 159501 incorrect handling of ALSA ioctls
3756 162020 Valgrinding an empty/zero-byte file crashes valgrind
3757 162482 ppc: Valgrind crashes while reading stabs information
3758 162718 x86: avoid segment selector 0 in sys_set_thread_area()
3759 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
3760 163560 VEX/test_main.c is missing from valgrind-3.3.1
3761 164353 malloc_usable_size() doesn't return a usable size
3762 165468 Inconsistent formatting in memcheck manual -- please fix
3763 169505 main.c:286 (endOfInstr):
3764 Assertion 'ii->cost_offset == *cost_offset' failed
3765 177206 Generate default.supp during compile instead of configure
3766 177209 Configure valt_load_address based on arch+os
3767 177305 eventfd / syscall 323 patch lost
3768 179731 Tests fail to build because of inlining of non-local asm labels
3769 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
3770 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
3771 181594 Bogus warning for empty text segment
3772 181707 dwarf doesn't require enumerations to have name
3773 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3774 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3775 Assertion '!already_present' failed.
3776 185359 exp-ptrcheck: unhandled syscall getresuid()
3777 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
3778 185816 Valgrind is unable to handle debug info for files with split
3779 debug info that are prelinked afterwards
3780 185980 [darwin] unhandled syscall: sem_open
3781 186238 bbToIR_AMD64: disInstr miscalculated next %rip
3782 186507 exp-ptrcheck unhandled syscalls prctl, etc.
3783 186790 Suppression pattern used for leaks are not reported
3784 186796 Symbols with length>200 in suppression files are ignored
3785 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
3786 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
3787 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
3788 188046 bashisms in the configure script
3789 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
3790 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
3791 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
3792 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
3793 assertion fail "!lock->heldBy"
3794 188427 Add support for epoll_create1 (with patch)
3795 188530 Support for SIOCGSTAMPNS
3796 188560 Include valgrind.spec in the tarball
3797 188572 Valgrind on Mac should suppress setenv() mem leak
3798 189054 Valgrind fails to build because of duplicate non-local asm labels
3799 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
3800 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
3801 189763 drd assertion failure: s_threadinfo[tid].is_recording
3802 190219 unhandled syscall: 328 (x86-linux)
3803 190391 dup of 181394; see above
3804 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
3805 190820 No debug information on powerpc-linux
3806 191095 PATCH: Improve usbdevfs ioctl handling
3807 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
3809 191189 --xml=yes should obey --gen-suppressions=all
3810 191192 syslog() needs a suppression on macosx
3811 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
3812 191761 getrlimit on MacOSX
3813 191992 multiple --fn-skip only works sometimes; dependent on order
3814 192634 V. reports "aspacem sync_check_mapping_callback:
3815 segment mismatch" on Darwin
3816 192954 __extension__ missing on 2 client requests
3817 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
3818 194474 "INSTALL" file has different build instructions than "README"
3819 194671 Unhandled syscall (sem_wait?) from mac valgrind
3820 195069 memcheck: reports leak (memory still reachable) for
3822 195169 drd: (vgDrd_barrier_post_wait):
3823 Assertion 'r->sg[p->post_iteration]' failed.
3824 195268 valgrind --log-file doesn't accept ~/...
3825 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
3826 195860 WARNING: unhandled syscall: unix:223
3827 196528 need a error suppression for pthread_rwlock_init under os x?
3828 197227 Support aio_* syscalls on Darwin
3829 197456 valgrind should reject --suppressions=(directory)
3830 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
3831 197591 unhandled syscall 27 (mincore)
3832 197793 Merge DCAS branch to the trunk == 85756, 142103
3833 197794 Avoid duplicate filenames in Vex
3834 197898 make check fails on current SVN
3835 197901 make check fails also under exp-ptrcheck in current SVN
3836 197929 Make --leak-resolution=high the default
3837 197930 Reduce spacing between leak reports
3838 197933 Print command line of client at start-up, and shorten preamble
3839 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
3840 198395 add BBV to the distribution as an experimental tool
3841 198624 Missing syscalls on Darwin: 82, 167, 281, 347
3842 198649 callgrind_annotate doesn't cumulate counters
3843 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
3844 199977 Valgrind complains about an unrecognized instruction in the
3845 atomic_incs test program
3846 200029 valgrind isn't able to read Fedora 12 debuginfo
3847 200760 darwin unhandled syscall: unix:284
3848 200827 DRD doesn't work on Mac OS X
3849 200990 VG_(read_millisecond_timer)() does not work correctly
3850 201016 Valgrind does not support pthread_kill() on Mac OS
3851 201169 Document --read-var-info
3852 201323 Pre-3.5.0 performance sanity checking
3853 201384 Review user manual for the 3.5.0 release
3854 201585 mfpvr not implemented on ppc
3855 201708 tests failing because x86 direction flag is left set
3856 201757 Valgrind doesn't handle any recent sys_futex additions
3857 204377 64-bit valgrind can not start a shell script
3858 (with #!/path/to/shell) if the shell is a 32-bit executable
3859 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
3860 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
3861 about memory allocation events with memory access tracing enabled
3862 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
3864 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
3868 Release 3.4.1 (28 February 2009)
3869 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3870 3.4.1 is a bug-fix release that fixes some regressions and assertion
3871 failures in debug info reading in 3.4.0, most notably incorrect stack
3872 traces on amd64-linux on older (glibc-2.3 based) systems. Various
3873 other debug info problems are also fixed. A number of bugs in the
3874 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
3876 In view of the fact that 3.4.0 contains user-visible regressions
3877 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
3878 encouraged to ship 3.4.1 in preference to 3.4.0.
3880 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
3881 bugzilla" -- that is, a bug that was reported to us but never got a
3882 bugzilla entry. We encourage you to file bugs in bugzilla
3883 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
3884 developers (or mailing lists) directly -- bugs that are not entered
3885 into bugzilla tend to get forgotten about or ignored.
3887 n-i-bz Fix various bugs reading icc-11 generated debug info
3888 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
3889 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
3890 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
3891 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
3892 179618 exp-ptrcheck crashed / exit prematurely
3893 179624 helgrind: false positive races with pthread_create and
3894 recv/open/close/read
3895 134207 pkg-config output contains @VG_PLATFORM@
3896 176926 floating point exception at valgrind startup with PPC 440EPX
3897 181594 Bogus warning for empty text segment
3898 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
3899 181707 Dwarf3 doesn't require enumerations to have name
3900 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3901 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3902 Assertion '!already_present' failed.
3903 185359 exp-ptrcheck unhandled syscall getresuid()
3905 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
3906 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
3910 Release 3.4.0 (2 January 2009)
3911 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3912 3.4.0 is a feature release with many significant improvements and the
3913 usual collection of bug fixes. This release supports X86/Linux,
3914 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
3915 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
3917 3.4.0 brings some significant tool improvements. Memcheck can now
3918 report the origin of uninitialised values, the thread checkers
3919 Helgrind and DRD are much improved, and we have a new experimental
3920 tool, exp-Ptrcheck, which is able to detect overruns of stack and
3921 global arrays. In detail:
3923 * Memcheck is now able to track the origin of uninitialised values.
3924 When it reports an uninitialised value error, it will try to show
3925 the origin of the value, as either a heap or stack allocation.
3926 Origin tracking is expensive and so is not enabled by default. To
3927 use it, specify --track-origins=yes. Memcheck's speed will be
3928 essentially halved, and memory usage will be significantly
3929 increased. Nevertheless it can drastically reduce the effort
3930 required to identify the root cause of uninitialised value errors,
3931 and so is often a programmer productivity win, despite running more
3934 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
3935 3.4.0, will be released shortly.
3937 * Helgrind's race detection algorithm has been completely redesigned
3938 and reimplemented, to address usability and scalability concerns:
3940 - The new algorithm has a lower false-error rate: it is much less
3941 likely to report races that do not really exist.
3943 - Helgrind will display full call stacks for both accesses involved
3944 in a race. This makes it easier to identify the root causes of
3947 - Limitations on the size of program that can run have been removed.
3949 - Performance has been modestly improved, although that is very
3952 - Direct support for Qt4 threading has been added.
3954 - pthread_barriers are now directly supported.
3956 - Helgrind works well on all supported Linux targets.
3958 * The DRD thread debugging tool has seen major improvements:
3960 - Greatly improved performance and significantly reduced memory
3963 - Support for several major threading libraries (Boost.Thread, Qt4,
3964 glib, OpenMP) has been added.
3966 - Support for atomic instructions, POSIX semaphores, barriers and
3967 reader-writer locks has been added.
3969 - Works now on PowerPC CPUs too.
3971 - Added support for printing thread stack usage at thread exit time.
3973 - Added support for debugging lock contention.
3975 - Added a manual for Drd.
3977 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
3978 checks for misuses of pointers. In that sense it is a bit like
3979 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
3980 detect overruns of stack and global arrays, it can detect
3981 arbitrarily far out-of-bounds accesses to heap blocks, and it can
3982 detect accesses heap blocks that have been freed a very long time
3983 ago (millions of blocks in the past).
3985 Ptrcheck currently works only on x86-linux and amd64-linux. To use
3986 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
3987 of the main Valgrind documentation. As this is an experimental
3988 tool, we would be particularly interested in hearing about your
3989 experiences with it.
3991 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
3992 longer built by default, although the code remains in the repository
3993 and the tarball. This is due to three factors: a perceived lack of
3994 users, a lack of maintenance, and concerns that it may not be
3995 possible to achieve reliable operation using the existing design.
3997 * As usual, support for the latest Linux distros and toolchain
3998 components has been added. It should work well on Fedora Core 10,
3999 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
4000 state) is supported, as is glibc-2.9. The C++ demangler has been
4001 updated so as to work well with C++ compiled by even the most recent
4004 * You can now use frame-level wildcards in suppressions. This was a
4005 frequently-requested enhancement. A line "..." in a suppression now
4006 matches zero or more frames. This makes it easier to write
4007 suppressions which are precise yet insensitive to changes in
4010 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
4012 * Very basic support for IBM Power6 has been added (64-bit processes only).
4014 * Valgrind is now cross-compilable. For example, it is possible to
4015 cross compile Valgrind on an x86/amd64-linux host, so that it runs
4016 on a ppc32/64-linux target.
4018 * You can set the main thread's stack size at startup using the
4019 new --main-stacksize= flag (subject of course to ulimit settings).
4020 This is useful for running apps that need a lot of stack space.
4022 * The limitation that you can't use --trace-children=yes together
4023 with --db-attach=yes has been removed.
4025 * The following bugs have been fixed. Note that "n-i-bz" stands for
4026 "not in bugzilla" -- that is, a bug that was reported to us but
4027 never got a bugzilla entry. We encourage you to file bugs in
4028 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
4029 mailing the developers (or mailing lists) directly.
4031 n-i-bz Make return types for some client requests 64-bit clean
4032 n-i-bz glibc 2.9 support
4033 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
4034 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
4035 n-i-bz Building in an env without gdb gives bogus gdb attach
4036 92456 Tracing the origin of uninitialised memory
4037 106497 Valgrind does not demangle some C++ template symbols
4039 151612 Suppression with "..." (frame-level wildcards in .supp files)
4040 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
4041 159285 unhandled syscall:25 (stime, on x86-linux)
4042 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
4043 160954 ppc build of valgrind crashes with illegal instruction (isel)
4044 160956 mallinfo implementation, w/ patch
4045 162092 Valgrind fails to start gnome-system-monitor
4046 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
4047 163794 assertion failure with "--track-origins=yes"
4048 163933 sigcontext.err and .trapno must be set together
4049 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
4050 164476 Missing kernel module loading system calls
4051 164669 SVN regression: mmap() drops posix file locks
4052 166581 Callgrind output corruption when program forks
4053 167288 Patch file for missing system calls on Cell BE
4054 168943 unsupported scas instruction pentium
4055 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
4056 172417 x86->IR: 0x82 ...
4057 172563 amd64->IR: 0xD9 0xF5 - fprem1
4058 173099 .lds linker script generation error
4059 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
4060 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
4062 174908 --log-file value not expanded correctly for core file
4063 175044 Add lookup_dcookie for amd64
4064 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4066 Developer-visible changes:
4068 * Valgrind's debug-info reading machinery has been majorly overhauled.
4069 It can now correctly establish the addresses for ELF data symbols,
4070 which is something that has never worked properly before now.
4072 Also, Valgrind can now read DWARF3 type and location information for
4073 stack and global variables. This makes it possible to use the
4074 framework to build tools that rely on knowing the type and locations
4075 of stack and global variables, for example exp-Ptrcheck.
4077 Reading of such information is disabled by default, because most
4078 tools don't need it, and because it is expensive in space and time.
4079 However, you can force Valgrind to read it, using the
4080 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4081 make use of such information, if present, to provide source-level
4082 descriptions of data addresses in the error messages they create.
4084 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4085 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).