2 Release 3.16.0 (?? ?????? 2019)
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 ====================
39 - The command option --collect-systime has been enhanced to specify
40 the unit used to record the elapsed time spent during system calls.
41 The command option now accepts the values no|yes|msec|usec|nsec,
42 where yes is a synonym of msec. When giving the value nsec, the
43 system cpu time of system calls is also recorded.
49 - Several memcheck options are now dynamically changeable.
50 Use valgrind --help-dyn-options to list them.
52 * ==================== OTHER CHANGES ====================
54 * New and modified GDB server monitor features:
56 - The gdbserver monitor commands that require an address and an optional
57 length argument now accepts the alternate 'C like' syntax "address[length]".
58 For example, the memcheck command "monitor who_points_at 0x12345678 120"
59 can now also be given as "monitor who_points_at 0x12345678[120]".
61 * ==================== FIXED BUGS ====================
63 The following bugs have been fixed or resolved. Note that "n-i-bz"
64 stands for "not in bugzilla" -- that is, a bug that was reported to us
65 but never got a bugzilla entry. We encourage you to file bugs in
66 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
67 than mailing the developers (or mailing lists) directly -- bugs that
68 are not entered into bugzilla tend to get forgotten about or ignored.
70 To see details of a given bug, visit
71 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
72 where XXXXXX is the bug number as listed below.
74 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
75 404406 s390x: z14 miscellaneous instructions not implemented
76 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
77 406561 mcinfcallWSRU gdbserver_test fails on ppc64
78 406824 Unsupported baseline
79 407218 Add support for the copy_file_range syscall
80 407307 Intercept stpcpy also in ld.so for arm64
81 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
82 408009 Expose rdrand and f16c even on avx if host cpu supports them
83 408091 Missing pkey syscalls
84 408414 Add support for missing for preadv2 and pwritev2 syscalls
85 409141 Valgrind hangs when SIGKILLed
86 409206 Support for Linux PPS and PTP ioctls
87 409367 exit_group() after signal to thread waiting in futex() causes hangs
88 409780 References to non-existent configure.in
89 410556 Add support for BLKIO{MIN,OPT} and BLKALIGNOFF ioctls
90 410599 Non-deterministic behaviour of pth_self_kill_15_other test
91 410757 discrepancy for preadv2/pwritev2 syscalls across different versions
92 411134 Allow the user to change a set of command line options during execution
93 411451 amd64->IR of bt/btc/bts/btr with immediate clears zero flag
94 412344 Problem setting mips flags with specific paths
95 413119 Ioctl wrapper for DRM_IOCTL_I915_GEM_MMAP
96 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
97 413603 callgrind_annotate/cg_annotate truncate function names at '#'
98 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
100 n-i-bz Fix minor one time leaks in dhat.
101 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
102 n-i-bz Add support for the Linux io_uring system calls
104 Release 3.15.0 (12 April 2019)
105 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
107 3.15.0 is a feature release with many improvements and the usual collection of
110 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
111 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
112 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
113 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
114 support for X86/macOS 10.13 and AMD64/macOS 10.13.
116 * ==================== CORE CHANGES ===================
118 * The XTree Massif output format now makes use of the information obtained
119 when specifying --read-inline-info=yes.
121 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
123 * ==================== TOOL CHANGES ====================
127 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
128 it has been promoted from an experimental tool to a regular tool. Run it
129 with --tool=dhat instead of --tool=exp-dhat.
131 - DHAT now prints only minimal data when the program ends, instead writing
132 the bulk of the profiling data to a file. As a result, the --show-top-n
133 and --sort-by options have been removed.
135 - Profile results can be viewed with the new viewer, dh_view.html. When
136 a run ends, a short message is printed, explaining how to view the result.
138 - See the documentation for more details.
142 - cg_annotate has a new option, --show-percs, which prints percentages next
147 - callgrind_annotate has a new option, --show-percs, which prints percentages
148 next to all event counts.
150 - callgrind_annotate now inserts commas in call counts, and
151 sort the caller/callee lists in the call tree.
155 - The default value for --read-inline-info is now "yes" on
156 Linux/Android/Solaris. It is still "no" on other OS.
160 - The option --xtree-leak=yes (to output leak result in xtree format)
161 automatically activates the option --show-leak-kinds=all, as xtree
162 visualisation tools such as kcachegrind can in any case select what kind
163 of leak to visualise.
165 - There has been further work to avoid false positives. In particular,
166 integer equality on partially defined inputs (C == and !=) is now handled
169 * ==================== OTHER CHANGES ====================
171 * The new option --show-error-list=no|yes displays, at the end of the run, the
172 list of detected errors and the used suppressions. Prior to this change,
173 showing this information could only be done by specifying "-v -v", but that
174 also produced a lot of other possibly-non-useful messages. The option -s is
175 equivalent to --show-error-list=yes.
177 * ==================== FIXED BUGS ====================
179 The following bugs have been fixed or resolved. Note that "n-i-bz"
180 stands for "not in bugzilla" -- that is, a bug that was reported to us
181 but never got a bugzilla entry. We encourage you to file bugs in
182 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
183 than mailing the developers (or mailing lists) directly -- bugs that
184 are not entered into bugzilla tend to get forgotten about or ignored.
186 To see details of a given bug, visit
187 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
188 where XXXXXX is the bug number as listed below.
190 385411 s390x: z13 vector floating-point instructions not implemented
191 397187 z13 vector register support for vgdb gdbserver
192 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
193 398870 Please add support for instruction vcvtps2ph
194 399287 amd64 front end: Illegal Instruction vcmptrueps
195 399301 Use inlined frames in Massif XTree output.
196 399322 Improve callgrind_annotate output
197 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
198 400164 helgrind test encounters mips x-compiler warnings and assembler error
199 400490 s390x: VRs allocated as if separate from FPRs
200 400491 s390x: Operand of LOCH treated as unsigned integer
201 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
202 options, which specify a mips64 processor
203 401112 LLVM 5.0 generates comparison against partially initialized data
204 401277 More bugs in z13 support
205 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
206 401578 drd: crashes sometimes on fork()
207 401627 memcheck errors with glibc avx2 optimized wcsncmp
208 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
209 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
210 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
212 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
213 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
214 402123 invalid assembler opcodes for mips32r2
215 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
216 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
217 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
218 402351 mips64 libvexmultiarch_test fails on s390x
220 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
221 402480 Do not use %rsp in clobber list
222 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
223 402515 Implement new option --show-error-list=no|yes / -s
224 402519 POWER 3.0 addex instruction incorrectly implemented
225 402781 Redo the cache used to process indirect branch targets
226 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
227 403552 s390x: wrong facility bit checked for vector facility
228 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
229 404638 Add VG_(replaceIndexXA)
230 404843 s390x: backtrace sometimes ends prematurely
231 404888 autotools cleanup series
232 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
233 405182 Valgrind fails to build with Clang
234 405205 filter_libc: remove the line holding the futex syscall error entirely
235 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
236 the upper and lower 32-bits of the 64-bit result
237 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
238 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
239 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
241 405403 s390x disassembler cannot be used on x86
242 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
243 405458 MIPS mkFormVEC arguments swapped?
244 405716 drd: Fix an integer overflow in the stack margin calculation
245 405722 Support arm64 core dump
246 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
247 of the 64-bit destination field.
248 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
249 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
250 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
252 406256 PPC64, vector floating point instructions don't handle subnormal
253 according to VSCR[NJ] bit setting.
254 406352 cachegrind/callgrind fails ann tests because of missing a.c
255 406354 dhat is broken on x86 (32bit)
256 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
257 406357 gdbserver_tests fails because of gdb output change
258 406360 memcheck/tests/libstdc++.supp needs more supression variants
259 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
260 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
261 407340 PPC64, does not support the vlogefp, vexptefp instructions.
262 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
263 n-i-bz Fix callgrind_annotate non deterministic order for equal total
264 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
265 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
266 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
268 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
269 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
270 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
274 Release 3.14.0 (9 October 2018)
275 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
277 3.14.0 is a feature release with many improvements and the usual collection of
280 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
281 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
282 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
283 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
284 support for X86/macOS 10.13, AMD64/macOS 10.13.
286 * ==================== CORE CHANGES ===================
288 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
289 debug info for unloaded code. This allows saved stack traces (e.g. for
290 memory leaks) to include file/line info for code that has been dlclose'd (or
291 similar). See the user manual for more information and known limitations.
293 * Ability to specify suppressions based on source file name and line number.
295 * Majorly overhauled register allocator. No end-user changes, but the JIT
296 generates code a bit more quickly now.
298 * ================== PLATFORM CHANGES =================
300 * Preliminary support for macOS 10.13 has been added.
302 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
304 * mips: support for MIPS SIMD architecture (MSA) has been added.
306 * mips: support for MIPS N32 ABI has been added.
308 * s390: partial support for vector instructions (integer and string) has been
311 * ==================== TOOL CHANGES ====================
313 * Helgrind: Addition of a flag
314 --delta-stacktrace=no|yes [yes on linux amd64/x86]
315 which specifies how full history stack traces should be computed.
316 Setting this to =yes can speed up Helgrind by 25% when using
317 --history-level=full.
319 * Memcheck: reduced false positive rate for optimised code created by Clang 6
320 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
321 blocks more carefully to determine where it can avoid expensive definedness
322 checks without loss of precision. This is controlled by the flag
323 --expensive-definedness-checks=no|auto|yes [auto].
325 * ==================== OTHER CHANGES ====================
327 * Valgrind is now buildable with link-time optimisation (LTO). A new
328 configure option --enable-lto=yes allows building Valgrind with LTO. If the
329 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
330 Note that if you are doing Valgrind development, --enable-lto=yes massively
331 slows down the build process.
333 * ==================== FIXED BUGS ====================
335 The following bugs have been fixed or resolved. Note that "n-i-bz"
336 stands for "not in bugzilla" -- that is, a bug that was reported to us
337 but never got a bugzilla entry. We encourage you to file bugs in
338 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
339 than mailing the developers (or mailing lists) directly -- bugs that
340 are not entered into bugzilla tend to get forgotten about or ignored.
342 To see details of a given bug, visit
343 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
344 where XXXXXX is the bug number as listed below.
346 79362 Debug info is lost for .so files when they are dlclose'd
347 208052 strlcpy error when n = 0
348 255603 exp-sgcheck Assertion '!already_present' failed
349 338252 building valgrind with -flto (link time optimisation) fails
350 345763 MIPS N32 ABI support
351 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
352 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
353 372347 Replacement problem of the additional c++14/c++17 new/delete operators
354 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
355 376257 helgrind history full speed up using a cached stack
356 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
358 379748 Fix missing pselect syscall (OS X 10.11)
359 379754 Fix missing syscall ulock_wait (OS X 10.12)
360 380397 s390x: __GI_strcspn() replacemenet needed
361 381162 possible array overrun in VEX register allocator
362 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
363 381274 powerpc too chatty even with --sigill-diagnostics=no
364 381289 epoll_pwait can have a NULL sigmask
365 381553 VEX register allocator v3
366 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
367 381769 Use ucontext_t instead of struct ucontext
368 381805 arm32 needs ld.so index hardwire for new glibc security fixes
369 382256 gz compiler flag test doesn't work for gold
370 382407 vg_perf needs "--terse" command line option
371 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
372 382563 MIPS MSA ASE support
373 382998 xml-socket doesn't work
374 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
375 383723 Fix missing kevent_qos syscall (macOS 10.11)
376 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
377 384096 Mention AddrCheck at Memcheck's command line option [..]
378 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
379 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
380 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
381 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
382 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
383 384337 performance improvements to VEX register allocator v2 and v3
384 384526 reduce number of spill insns generated by VEX register allocator v3
385 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
386 384631 Sanitise client args as printed with -v
387 384633 Add a simple progress-reporting facility
388 384987 VEX regalloc: allocate caller-save registers for short lived vregs
389 385055 PPC VEX temporary storage exhausted
390 385182 PPC64 is missing support for the DSCR
391 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
392 385207 PPC64, generate_store_FPRF() generates too many Iops
393 385208 PPC64, xxperm instruction exhausts temporary memory
394 385210 PPC64, vpermr instruction could exhaust temporary memory
395 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
396 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
397 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
398 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
399 385408 s390x: z13 vector "support" instructions not implemented
400 385409 s390x: z13 vector integer instructions not implemented
401 385410 s390x: z13 vector string instructions not implemented
402 385412 s390x: new non-vector z13 instructions not implemented
403 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
404 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
405 385939 Optionally exit on the first error
406 386318 valgrind.org/info/tools.html is missing SGCheck
407 386425 running valgrind + wine on armv7l gives illegal opcode
408 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
409 387410 MIPSr6 support
410 387664 Memcheck: make expensive-definedness-checks be the default
411 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
412 387766 asm shifts cause false positive "Conditional jump or move depends
413 on uninitialised value"
414 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
415 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
416 388786 Support bpf syscall in amd64 Linux
417 388862 Add replacements for wmemchr and wcsnlen on Linux
418 389065 valgrind meets gcc flag -Wlogical-op
419 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
420 390471 suppression by specification of source-file line number
421 390723 make xtree dump files world wide readable, similar to log files
422 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
423 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
424 392118 unhandled amd64-linux syscall: 332 (statx)
425 392449 callgrind not clearing the number of calls properly
426 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
427 lxssp, stxssp and stxvl instructions.
428 393023 callgrind_control risks using the wrong vgdb
429 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
430 393099 posix_memalign() invalid write if alignment == 0
431 393146 failing assert "is_DebugInfo_active(di)"
432 395709 PPC64 is missing support for the xvnegsp instruction
433 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
435 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
436 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
437 396839 s390x: Trap instructions not implemented
438 396887 arch_prctl should return EINVAL on unknown option
439 == 397286 crash before launching binary (Unsupported arch_prctl option)
440 == 397393 valgrind: the 'impossible' happened: (Archlinux)
441 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
442 396906 compile tests failure on mips32-linux: broken inline asm in tests on
444 397012 glibc ld.so uses arch_prctl on i386
445 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
446 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
447 397424 glibc 2.27 and gdb_server tests
448 398028 Assertion `cfsi_fits` failing in simple C program
449 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
451 n-i-bz Fix missing workq_ops operations (macOS)
452 n-i-bz fix bug in strspn replacement
453 n-i-bz Add support for the Linux BLKFLSBUF ioctl
454 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
455 n-i-bz Fix possible stack trashing by semctl syscall wrapping
456 n-i-bz Add support for the Linux membarrier() system call
457 n-i-bz x86 front end: recognise and handle UD2 correctly
458 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
459 correctly aligned before entering the handler.
461 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
462 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
463 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
467 Release 3.13.0 (15 June 2017)
468 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
470 3.13.0 is a feature release with many improvements and the usual collection of
473 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
474 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
475 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
476 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
478 * ==================== CORE CHANGES ===================
480 * The translation cache size has been increased to keep up with the demands of
481 large applications. The maximum number of sectors has increased from 24 to
482 48. The default number of sectors has increased from 16 to 32 on all
483 targets except Android, where the increase is from 6 to 12.
485 * The amount of memory that Valgrind can use has been increased from 64GB to
486 128GB. In particular this means your application can allocate up to about
487 60GB when running on Memcheck.
489 * Valgrind's default load address has been changed from 0x3800'0000 to
490 0x5800'0000, so as to make it possible to load larger executables. This
491 should make it possible to load executables of size at least 1200MB.
493 * A massive spaceleak caused by reading compressed debuginfo files has been
494 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
497 * The C++ demangler has been updated.
499 * Support for demangling Rust symbols has been added.
501 * A new representation of stack traces, the "XTree", has been added. An XTree
502 is a tree of stacktraces with data associated with the stacktraces. This is
503 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
504 consumption of your program. Reporting is controlled by the new options
505 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
507 A report can also be produced on demand using the gdbserver monitor command
508 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
509 format' and 'massif format. The existing visualisers for these formats (e.g.
510 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
511 analyse these reports.
513 Memcheck can also produce XTree leak reports using the Callgrind file
514 format. For more details, see the user manual.
516 * ================== PLATFORM CHANGES =================
518 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
520 * amd64: fixes for JIT failure problems on long AVX2 code blocks
522 * amd64 and x86: support for CET prefixes has been added
524 * arm32: a few missing ARMv8 instructions have been implemented
526 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
527 Store-Conditional instructions has been added. This is to deal with
528 processor implementations that implement the LL/SC specifications strictly
529 and as a result cause Valgrind to hang in certain situations. The
530 alternative implementation is automatically enabled at startup, as required.
531 You can use the option --sim-hints=fallback-llsc to force-enable it if you
534 * Support for OSX 10.12 has been improved.
536 * On Linux, clone handling has been improved to honour CLONE_VFORK that
537 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
538 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
539 CLONE_VM exact semantics will (still) not work.
541 * The TileGX/Linux port has been removed because it appears to be both unused
544 * ==================== TOOL CHANGES ====================
548 - Memcheck should give fewer false positives when running optimised
549 Clang/LLVM generated code.
551 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
553 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
554 to produce the end of execution leak report in a xtree callgrind format
557 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
558 the leak report in an xtree file.
562 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
564 - For some workloads (typically, for big applications), Massif memory
565 consumption and CPU consumption has decreased significantly.
569 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
571 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
572 for Ada gnat compiled applications.
574 * ==================== OTHER CHANGES ====================
576 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
577 append the inner guest stacktrace to the inner host stacktrace. This helps
578 to investigate the errors reported by the outer, when they are caused by the
579 inner guest program (such as an inner regtest). See README_DEVELOPERS for
582 * To allow fast detection of callgrind files by desktop environments and file
583 managers, the format was extended to have an optional first line that
584 uniquely identifies the format ("# callgrind format"). Callgrind creates
585 this line now, as does the new xtree functionality.
587 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
588 have a new %n format letter that is replaced by a sequence number.
590 * "--version -v" now shows the SVN revision numbers from which Valgrind was
593 * ==================== FIXED BUGS ====================
595 The following bugs have been fixed or resolved. Note that "n-i-bz"
596 stands for "not in bugzilla" -- that is, a bug that was reported to us
597 but never got a bugzilla entry. We encourage you to file bugs in
598 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
599 than mailing the developers (or mailing lists) directly -- bugs that
600 are not entered into bugzilla tend to get forgotten about or ignored.
602 To see details of a given bug, visit
603 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
604 where XXXXXX is the bug number as listed below.
606 162848 --log-file output isn't split when a program forks
607 340777 Illegal instruction on mips (ar71xx)
608 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
609 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
610 to a different stack.
611 344139 x86 stack-seg overrides, needed by the Wine people
612 344524 store conditional of guest applications always fail - observed on
614 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
615 352395 Please provide SVN revision info in --version -v
616 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
617 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
618 INVALID_POSIX_THREADID' failed
619 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
620 barrier implementation
621 358697 valgrind.h: Some code remains even when defining NVALGRIND
622 359202 Add musl libc configure/compile
623 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
624 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
625 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
626 362223 assertion failed when .valgrindrc is a directory instead of a file
627 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
628 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
629 368507 can't malloc chunks larger than about 34GB
630 368529 Android arm target link error, missing atexit and pthread_atfork
631 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
632 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
633 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
634 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
635 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
636 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
637 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
638 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
639 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
640 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
641 370028 Reduce the number of compiler warnings on MIPS platforms
642 370635 arm64 missing syscall getcpu
643 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
644 371227 Clean AArch64 syscall table
645 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
646 371471 Valgrind complains about non legit memory leaks on placement new (C++)
647 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
648 371503 disInstr(arm64): unhandled instruction 0xF89F0000
649 371869 support '%' in symbol Z-encoding
650 371916 execution tree xtree concept
651 372120 c++ demangler demangles symbols which are not c++
652 372185 Support of valgrind on ARMv8 with 32 bit executable
653 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
654 372195 Power PC, xxsel instruction is not always recognized.
655 372504 Hanging on exit_group
656 372600 process loops forever when fatal signals are arriving quickly
657 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
658 373046 Stacks registered by core are never deregistered
659 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
660 373086 Implement additional Xen hypercalls
661 373192 Calling posix_spawn in glibc 2.24 completely broken
662 373488 Support for fanotify API on ARM64 architecture
663 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
664 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
665 373938 const IRExpr arguments for matchIRExpr()
666 374719 some spelling fixes
667 374963 increase valgrind's load address to prevent mmap failure
668 375514 valgrind_get_tls_addr() does not work in case of static TLS
669 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
670 for ML_(find_rx_mapping)()
671 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
672 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
673 == 377159 "vex: the `impossible' happened" still present
674 == 375150 Assertion 'tres.status == VexTransOK' failed
675 == 378068 valgrind crashes on AVX2 function in FFmpeg
676 376142 Segfaults on MIPS Cavium Octeon boards
677 376279 disInstr(arm64): unhandled instruction 0xD50320FF
678 376455 Solaris: unhandled syscall lgrpsys(180)
679 376518 Solaris: unhandled fast trap getlgrp(6)
680 376611 ppc64 and arm64 don't know about prlimit64 syscall
681 376729 PPC64, remove R2 from the clobber list
683 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
684 to be wrongly marked as addressable
685 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
686 PIE enabled by default
687 377376 memcheck/tests/linux/getregset fails with glibc2.24
688 377427 PPC64, lxv instruction failing on odd destination register
689 377478 PPC64: ISA 3.0 setup fixes
690 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
691 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
692 and 2 args for FUTEX_TRYLOCK_PI
693 377717 Fix massive space leak when reading compressed debuginfo sections
694 377891 Update Xen 4.6 domctl wrappers
695 377930 fcntl syscall wrapper is missing flock structure check
696 378524 libvexmultiarch_test regression on s390x and ppc64
697 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
698 378673 Update libiberty demangler
699 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
700 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
701 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
702 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
703 (task_register_dyld_image_infos)
704 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
705 (task_register_dyld_shared_cache_image_info)
706 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
707 379473 MIPS: add support for rdhwr cycle counter register
708 379504 remove TileGX/Linux port
709 379525 Support more x86 nop opcodes
710 379838 disAMode(x86): not an addr!
711 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
712 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
713 379895 clock_gettime does not execute POST syscall wrapper
714 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
715 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
716 380200 xtree generated callgrind files refer to files without directory name
717 380202 Assertion failure for cache line size (cls == 64) on aarch64.
718 380397 s390x: __GI_strcspn() replacement needed
719 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
721 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
722 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
723 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
727 Release 3.12.0 (20 October 2016)
728 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
730 3.12.0 is a feature release with many improvements and the usual
731 collection of bug fixes.
733 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
734 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
735 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
736 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
737 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
738 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
740 * ================== PLATFORM CHANGES =================
742 * POWER: Support for ISA 3.0 has been added
744 * mips: support for O32 FPXX ABI has been added.
745 * mips: improved recognition of different processors
746 * mips: determination of page size now done at run time
748 * amd64: Partial support for AMD FMA4 instructions.
750 * arm, arm64: Support for v8 crypto and CRC instructions.
752 * Improvements and robustification of the Solaris port.
754 * Preliminary support for MacOS 10.12 (Sierra) has been added.
756 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
757 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
758 where possible. Valgrind's support for 32-bit x86 has stagnated in
759 recent years and has fallen far behind that for 64-bit x86
760 instructions. By contrast 64-bit x86 is well supported, up to and
763 * ==================== TOOL CHANGES ====================
767 - Added meta mempool support for describing a custom allocator which:
768 - Auto-frees all chunks assuming that destroying a pool destroys all
770 - Uses itself to allocate other memory blocks
772 - New flag --ignore-range-below-sp to ignore memory accesses below
773 the stack pointer, if you really have to. The related flag
774 --workaround-gcc296-bugs=yes is now deprecated. Use
775 --ignore-range-below-sp=1024-1 as a replacement.
779 - Improved thread startup time significantly on non-Linux platforms.
783 - Added collection of the metric "tot-blocks-allocd"
785 * ==================== OTHER CHANGES ====================
787 * Replacement/wrapping of malloc/new related functions is now done not just
788 for system libraries by default, but for any globally defined malloc/new
789 related function (both in shared libraries and statically linked alternative
790 malloc implementations). The dynamic (runtime) linker is excluded, though.
791 To only intercept malloc/new related functions in
792 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
793 "nouserintercepts" can be any non-existing library name).
794 This new functionality is not implemented for MacOS X.
796 * The maximum number of callers in a suppression entry is now equal to
797 the maximum size for --num-callers (500).
798 Note that --gen-suppressions=yes|all similarly generates suppressions
799 containing up to --num-callers frames.
801 * New and modified GDB server monitor features:
803 - Valgrind's gdbserver now accepts the command 'catch syscall'.
804 Note that you must have GDB >= 7.11 to use 'catch syscall' with
807 * New option --run-cxx-freeres=<yes|no> can be used to change whether
808 __gnu_cxx::__freeres() cleanup function is called or not. Default is
811 * Valgrind is able to read compressed debuginfo sections in two formats:
812 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
813 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
815 * Modest JIT-cost improvements: the cost of instrumenting code blocks
816 for the most common use case (x86_64-linux, Memcheck) has been
819 * Improved performance for programs that do a lot of discarding of
820 instruction address ranges of 8KB or less.
822 * The C++ symbol demangler has been updated.
824 * More robustness against invalid syscall parameters on Linux.
826 * ==================== FIXED BUGS ====================
828 The following bugs have been fixed or resolved. Note that "n-i-bz"
829 stands for "not in bugzilla" -- that is, a bug that was reported to us
830 but never got a bugzilla entry. We encourage you to file bugs in
831 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
832 than mailing the developers (or mailing lists) directly -- bugs that
833 are not entered into bugzilla tend to get forgotten about or ignored.
835 To see details of a given bug, visit
836 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
837 where XXXXXX is the bug number as listed below.
839 191069 Exiting due to signal not reported in XML output
840 199468 Suppressions: stack size limited to 25
841 while --num-callers allows more frames
842 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
843 278744 cvtps2pd with redundant RexW
844 303877 valgrind doesn't support compressed debuginfo sections.
845 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
846 348345 Assertion fails for negative lineno
847 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
848 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
849 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
850 351804 Crash on generating suppressions for "printf" call on OS X 10.10
851 352197 mips: mmap2() not wrapped correctly for page size > 4096
852 353083 arm64 doesn't implement various xattr system calls
853 353084 arm64 doesn't support sigpending system call
854 353137 www: update info for Supported Platforms
855 353138 www: update "The Valgrind Developers" page
856 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
859 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
860 353398 WARNING: unhandled amd64-solaris syscall: 207
861 353660 XML in auxwhat tag not escaping reserved symbols properly
862 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
863 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
864 353802 ELF debug info reader confused with multiple .rodata sections
865 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
866 353917 unhandled amd64-solaris syscall fchdir(120)
867 353920 unhandled amd64-solaris syscall: 170
868 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
869 354392 unhandled amd64-solaris syscall: 171
870 354797 Vbit test does not include Iops for Power 8 instruction support
871 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
875 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
876 355188 valgrind should intercept all malloc related global functions
877 355454 do not intercept malloc related symbols from the runtime linker
878 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
879 356044 Dwarf line info reader misinterprets is_stmt register
880 356112 mips: replace addi with addiu
881 356393 valgrind (vex) crashes because isZeroU happened
884 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
885 356678 arm64-linux: unhandled syscall 232 (mincore)
886 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
887 356823 Unsupported ARM instruction: stlex
888 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
889 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
890 357673 crash if I try to run valgrind with a binary link with libcurl
891 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
892 357871 pthread_spin_destroy not properly wrapped
893 357887 Calls to VG_(fclose) do not close the file descriptor
894 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
895 358030 support direct socket calls on x86 32bit (new in linux 4.3)
896 358478 drd/tests/std_thread.cpp doesn't build with GCC6
897 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
898 359181 Buffer Overflow during Demangling
899 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
900 359289 s390x: popcnt (B9E1) not implemented
901 359472 The Power PC vsubuqm instruction doesn't always give the correct result
902 359503 Add missing syscalls for aarch64 (arm64)
903 359645 "You need libc6-dbg" help message could be more helpful
904 359703 s390: wire up separate socketcalls system calls
905 359724 getsockname might crash - deref_UInt should call safe_to_deref
906 359733 amd64 implement ld.so strchr/index override like x86
907 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
908 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
910 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
911 359871 Incorrect mask handling in ppoll
912 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
913 360008 Contents of Power vr registers contents is not printed correctly when
914 the --vgdb-shadow-registers=yes option is used
915 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
917 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
918 360425 arm64 unsupported instruction ldpsw
920 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
921 360571 Error about the Android Runtime reading below the stack pointer on ARM
922 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
923 360749 kludge for multiple .rodata sections on Solaris no longer needed
924 360752 raise the number of reserved fds in m_main.c from 10 to 12
925 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
926 361226 s390x: risbgn (EC59) not implemented
927 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
928 361354 ppc64[le]: wire up separate socketcalls system calls
929 361615 Inconsistent termination for multithreaded process terminated by signal
930 361926 Unhandled Solaris syscall: sysfs(84)
931 362009 V dumps core on unimplemented functionality before threads are created
932 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
933 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
934 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
935 362953 Request for an update to the Valgrind Developers page
936 363680 add renameat2() support
937 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
938 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
939 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
940 364058 clarify in manual limitations of array overruns detections
941 364413 pselect sycallwrapper mishandles NULL sigmask
942 364728 Power PC, missing support for several HW registers in
943 get_otrack_shadow_offset_wrk()
944 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
945 365273 Invalid write to stack location reported after signal handler runs
946 365912 ppc64BE segfault during jm-insns test (RELRO)
947 366079 FPXX Support for MIPS32 Valgrind
948 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
949 366344 Multiple unhandled instruction for Aarch64
950 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
951 367995 Integration of memcheck with custom memory allocator
952 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
953 368412 False positive result for altivec capability check
954 368416 Add tc06_two_races_xml.exp output for ppc64
955 368419 Perf Events ioctls not implemented
956 368461 mmapunmap test fails on ppc64
957 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
958 369000 AMD64 fma4 instructions unsupported.
959 369169 ppc64 fails jm_int_isa_2_07 test
960 369175 jm_vec_isa_2_07 test crashes on ppc64
961 369209 valgrind loops and eats up all memory if cwd doesn't exist.
962 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
963 369359 msghdr_foreachfield can crash when handling bad iovec
964 369360 Bad sigprocmask old or new sets can crash valgrind
965 369361 vmsplice syscall wrapper crashes on bad iovec
966 369362 Bad sigaction arguments crash valgrind
967 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
968 369402 Bad set/get_thread_area pointer crashes valgrind
969 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
970 369446 valgrind crashes on unknown fcntl command
971 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
972 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
973 370265 ISA 3.0 HW cap stuff needs updating
974 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
975 372195 Power PC, xxsel instruction is not always recognized
977 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
978 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
979 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
980 n-i-bz Fix ppoll_alarm exclusion on OS X
981 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
982 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
983 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
984 n-i-bz amd64: memcheck false positive with shr %edx
985 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
986 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
987 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
988 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
990 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
991 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
992 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
996 Release 3.11.0 (22 September 2015)
997 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
999 3.11.0 is a feature release with many improvements and the usual
1000 collection of bug fixes.
1002 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
1003 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
1004 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
1005 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
1006 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
1007 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1009 * ================== PLATFORM CHANGES =================
1011 * Support for Solaris/x86 and Solaris/amd64 has been added.
1013 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1015 * Preliminary support for the Tilera TileGX architecture has been added.
1017 * s390x: It is now required for the host to have the "long displacement"
1018 facility. The oldest supported machine model is z990.
1020 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1021 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1023 * The JIT's register allocator is significantly faster, making the JIT
1024 as a whole somewhat faster, so JIT-intensive activities, for example
1025 program startup, are modestly faster, around 5%.
1027 * There have been changes to the default settings of several command
1028 line flags, as detailed below.
1030 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1031 capable hosts, the simulated CPUID will now indicate AVX2 support.
1033 * ==================== TOOL CHANGES ====================
1037 - The default value for --leak-check-heuristics has been changed from
1038 "none" to "all". This helps to reduce the number of possibly
1039 lost blocks, in particular for C++ applications.
1041 - The default value for --keep-stacktraces has been changed from
1042 "malloc-then-free" to "malloc-and-free". This has a small cost in
1043 memory (one word per malloc-ed block) but allows Memcheck to show the
1044 3 stacktraces of a dangling reference: where the block was allocated,
1045 where it was freed, and where it is acccessed after being freed.
1047 - The default value for --partial-loads-ok has been changed from "no" to
1048 "yes", so as to avoid false positive errors resulting from some kinds
1049 of vectorised loops.
1051 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1052 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1053 than get_vbits when you need to associate byte data value with
1054 their corresponding validity bits.
1056 - The 'block_list' monitor command has been enhanced:
1057 o it can print a range of loss records
1058 o it now accepts an optional argument 'limited <max_blocks>'
1059 to control the number of blocks printed.
1060 o if a block has been found using a heuristic, then
1061 'block_list' now shows the heuristic after the block size.
1062 o the loss records/blocks to print can be limited to the blocks
1063 found via specified heuristics.
1065 - The C helper functions used to instrument loads on
1066 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1067 replaced by handwritten assembly sequences. This gives speedups
1068 in the region of 0% to 7% for those targets only.
1070 - A new command line option, --expensive-definedness-checks=yes|no,
1071 has been added. This is useful for avoiding occasional invalid
1072 uninitialised-value errors in optimised code. Watch out for
1073 runtime degradation, as this can be up to 25%. As always, though,
1074 the slowdown is highly application specific. The default setting
1079 - A new monitor command 'all_snapshots <filename>' dumps all
1080 snapshots taken so far.
1084 - Significant memory reduction and moderate speedups for
1085 --history-level=full for applications accessing a lot of memory
1086 with many different stacktraces.
1088 - The default value for --conflict-cache-size=N has been doubled to
1089 2000000. Users that were not using the default value should
1090 preferably also double the value they give.
1092 The default was changed due to the changes in the "full history"
1093 implementation. Doubling the value gives on average a slightly more
1094 complete history and uses similar memory (or significantly less memory
1095 in the worst case) than the previous implementation.
1097 - The Helgrind monitor command 'info locks' now accepts an optional
1098 argument 'lock_addr', which shows information about the lock at the
1101 - When using --history-level=full, the new Helgrind monitor command
1102 'accesshistory <addr> [<len>]' will show the recorded accesses for
1103 <len> (or 1) bytes at <addr>.
1105 * ==================== OTHER CHANGES ====================
1107 * The default value for the --smc-check option has been changed from
1108 "stack" to "all-non-file" on targets that provide automatic D-I
1109 cache coherence (x86, amd64 and s390x). The result is to provide,
1110 by default, transparent support for JIT generated and self-modifying
1111 code on all targets.
1113 * Mac OS X only: the default value for the --dsymutil option has been
1114 changed from "no" to "yes", since any serious usage on Mac OS X
1115 always required it to be "yes".
1117 * The command line options --db-attach and --db-command have been removed.
1118 They were deprecated in 3.10.0.
1120 * When a process dies due to a signal, Valgrind now shows the signal
1121 and the stacktrace at default verbosity (i.e. verbosity 1).
1123 * The address description logic used by Memcheck and Helgrind now
1124 describes addresses in anonymous segments, file mmap-ed segments,
1125 shared memory segments and the brk data segment.
1127 * The new option --error-markers=<begin>,<end> can be used to mark the
1128 begin/end of errors in textual output mode, to facilitate
1129 searching/extracting errors in output files that mix valgrind errors
1130 with program output.
1132 * The new option --max-threads=<number> can be used to change the number
1133 of threads valgrind can handle. The default is 500 threads which
1134 should be more than enough for most applications.
1136 * The new option --valgrind-stacksize=<number> can be used to change the
1137 size of the private thread stacks used by Valgrind. This is useful
1138 for reducing memory use or increasing the stack size if Valgrind
1139 segfaults due to stack overflow.
1141 * The new option --avg-transtab-entry-size=<number> can be used to specify
1142 the expected instrumented block size, either to reduce memory use or
1143 to avoid excessive retranslation.
1145 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1147 * New and modified GDB server monitor features:
1149 - When a signal is reported in GDB, you can now use the GDB convenience
1150 variable $_siginfo to examine detailed signal information.
1152 - Valgrind's gdbserver now allows the user to change the signal
1153 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1154 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1155 continue without passing the signal to the process.
1157 - With GDB >= 7.10, the command 'target remote'
1158 will automatically load the executable file of the process running
1159 under Valgrind. This means you do not need to specify the executable
1160 file yourself, GDB will discover it itself. See GDB documentation about
1161 'qXfer:exec-file:read' packet for more info.
1163 * ==================== FIXED BUGS ====================
1165 The following bugs have been fixed or resolved. Note that "n-i-bz"
1166 stands for "not in bugzilla" -- that is, a bug that was reported to us
1167 but never got a bugzilla entry. We encourage you to file bugs in
1168 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1169 than mailing the developers (or mailing lists) directly -- bugs that
1170 are not entered into bugzilla tend to get forgotten about or ignored.
1172 To see details of a given bug, visit
1173 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1174 where XXXXXX is the bug number as listed below.
1176 116002 VG_(printf): Problems with justification of strings and integers
1177 155125 avoid cutting away file:lineno after long function name
1178 197259 Unsupported arch_prtctl PR_SET_GS option
1179 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
1180 201216 Fix Valgrind does not support pthread_sigmask() on OS X
1181 201435 Fix Darwin: -v does not show kernel version
1182 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
1183 211256 Fixed an outdated comment regarding the default platform.
1184 211529 Incomplete call stacks for code compiled by newer versions of MSVC
1185 211926 Avoid compilation warnings in valgrind.h with -pedantic
1186 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
1188 226609 Crediting upstream authors in man page
1189 231257 Valgrind omits path when executing script from shebang line
1190 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
1191 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
1192 269360 s390x: Fix addressing mode selection for compare-and-swap
1193 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
1195 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
1196 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
1197 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
1198 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
1199 330147 libmpiwrap PMPI_Get_count returns undefined value
1200 333051 mmap of huge pages fails due to incorrect alignment
1202 334802 valgrind does not always explain why a given option is bad
1203 335618 mov.w rN, pc/sp (ARM32)
1204 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
1209 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
1210 338602 AVX2 bit in CPUID missing
1211 338606 Strange message for scripts with invalid interpreter
1212 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
1213 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
1214 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
1216 339156 gdbsrv not called for fatal signal
1217 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
1218 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
1219 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
1220 339442 Fix testsuite build failure on OS X 10.9
1221 339542 Enable compilation with Intel's ICC compiler
1222 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
1223 339688 Mac-specific ASM does not support .version directive (cpuid,
1224 tronical and pushfpopf tests)
1225 339745 Valgrind crash when check Marmalade app (partial fix)
1226 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
1227 339778 Linux/TileGx platform support for Valgrind
1228 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
1229 339789 Fix none/tests/execve test on Mac OS X 10.9
1230 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
1231 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
1232 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
1233 340392 Allow user to select more accurate definedness checking in memcheck
1234 to avoid invalid complaints on optimised code
1235 340430 Fix some grammatical weirdness in the manual.
1236 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
1237 341419 Signal handler ucontext_t not filled out correctly on OS X
1238 341539 VG_(describe_addr) should not describe address as belonging to client
1239 segment if it is past the heap end
1240 341613 Enable building of manythreads and thread-exits tests on Mac OS X
1241 341615 Fix none/tests/darwin/access_extended test on Mac OS X
1242 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
1243 341789 aarch64: shmat fails with valgrind on ARMv8
1244 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
1245 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
1246 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
1247 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
1248 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
1249 342221 socket connect false positive uninit memory for unknown af family
1250 342353 Allow dumping full massif output while valgrind is still running
1251 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
1255 342603 Add I2C_SMBUS ioctl support
1256 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
1257 342683 Mark memory past the initial brk limit as unaddressable
1258 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
1259 342795 Internal glibc __GI_mempcpy call should be intercepted
1260 342841 s390x: Support instructions fiebr(a) and fidbr(a)
1261 343012 Unhandled syscall 319 (memfd_create)
1262 343069 Patch updating v4l2 API support
1263 343173 helgrind crash during stack unwind
1264 343219 fix GET_STARTREGS for arm
1265 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
1266 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
1267 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
1268 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
1269 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
1270 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
1271 343597 ppc64le: incorrect use of offseof macro
1272 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
1273 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
1274 343732 Unhandled syscall 144 (setgid) on aarch64
1275 343733 Unhandled syscall 187 (msgctl and related) on aarch64
1276 343802 s390x: False positive "conditional jump or move depends on [..]
1277 343902 --vgdb=yes doesn't break when --xml=yes is used
1278 343967 Don't warn about setuid/setgid/setcap executable for directories
1279 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
1280 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
1281 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
1282 344054 www - update info for Solaris/illumos
1283 344416 'make regtest' does not work cleanly on OS X
1284 344235 Remove duplicate include of pub_core_aspacemgr.h
1285 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
1286 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
1287 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
1288 344314 callgrind_annotate ... warnings about commands containing newlines
1289 344318 socketcall should wrap recvmmsg and sendmmsg
1290 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
1291 344416 Fix 'make regtest' does not work cleanly on OS X
1292 344499 Fix compilation for Linux kernel >= 4.0.0
1293 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
1294 unix:349 (__pthread_fchdir)
1295 344559 Garbage collection of unused segment names in address space manager
1296 344560 Fix stack traces missing penultimate frame on OS X
1297 344621 Fix memcheck/tests/err_disable4 test on OS X
1298 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
1299 344702 Fix missing libobjc suppressions on OS X 10.10
1301 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
1302 344939 Fix memcheck/tests/xml1 on OS X 10.10
1303 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
1304 345079 Fix build problems in VEX/useful/test_main.c
1305 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
1306 345177 arm64: prfm (reg) not implemented
1307 345215 Performance improvements for the register allocator
1308 345248 add support for Solaris OS in valgrind
1309 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
1310 345394 Fix memcheck/tests/strchr on OS X
1311 345637 Fix memcheck/tests/sendmsg on OS X
1312 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
1313 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
1314 345887 Fix an assertion in the address space manager
1315 345928 amd64: callstack only contains current function for small stacks
1316 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1317 345987 MIPS64: Implement cavium LHX instruction
1318 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1319 346185 Fix typo saving altivec register v24
1320 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1321 and LibVEX_GuestPPC64_get_CR()
1322 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1323 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1324 346307 fuse filesystem syscall deadlocks
1325 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1326 346411 MIPS: SysRes::_valEx handling is incorrect
1327 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1328 346474 PPC64 Power 8, spr TEXASRU register not supported
1329 346487 Compiler generates "note" about a future ABI change for PPC64
1330 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1331 and causing spurious "invalid read of size 8" warnings
1332 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1333 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1334 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1335 347322 Power PC regression test cleanup
1336 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1338 347389 unhandled syscall: 373 (Linux ARM syncfs)
1339 347686 Patch set to cleanup PPC64 regtests
1340 347978 Remove bash dependencies where not needed
1341 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1342 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1344 348102 Patch updating v4l2 API support
1345 348247 amd64 front end: jno jumps wrongly when overflow is not set
1346 348269 Improve mmap MAP_HUGETLB support.
1347 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1348 348345 Assertion fails for negative lineno
1349 348377 Unsupported ARM instruction: yield
1350 348565 Fix detection of command line option availability for clang
1351 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1352 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1353 348748 Fix redundant condition
1354 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1355 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1356 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1357 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1358 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1359 349626 Implemented additional Xen hypercalls
1360 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1361 349790 Clean up of the hardware capability checking utilities.
1362 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1363 349874 Fix typos in source code
1364 349879 memcheck: add handwritten assembly for helperc_LOADV*
1365 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1366 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1367 350202 Add limited param to 'monitor block_list'
1368 350290 s390x: Support instructions fixbr(a)
1369 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1370 350809 Fix none/tests/async-sigs for Solaris
1371 350811 Remove reference to --db-attach which has been removed.
1372 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1373 350854 hard-to-understand code in VG_(load_ELF)()
1374 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1375 351386 Solaris: Cannot run ld.so.1 under Valgrind
1376 351474 Fix VG_(iseqsigset) as obvious
1377 351531 Typo in /include/vki/vki-xen-physdev.h header guard
1378 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
1379 351858 ldsoexec support on Solaris
1380 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
1381 352130 helgrind reports false races for printfs using mempcpy on FILE* state
1382 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
1383 352320 arm64 crash on none/tests/nestedfs
1384 352765 Vbit test fails on Power 6
1385 352768 The mbar instruction is missing from the Power PC support
1386 352769 Power PC program priority register (PPR) is not supported
1387 n-i-bz Provide implementations of certain compiler builtins to support
1388 compilers that may not provide those
1389 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
1390 n-i-bz Fix compilation on distros with glibc < 2.5
1391 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
1392 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
1393 n-i-bz mremap did not work properly on shared memory
1394 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
1395 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
1397 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
1398 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
1399 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
1403 Release 3.10.1 (25 November 2014)
1404 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1405 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
1406 and backports fixes for all reported missing AArch64 ARMv8 instructions
1407 and syscalls from the trunk. If you package or deliver 3.10.0 for others
1408 to use, you might want to consider upgrading to 3.10.1 instead.
1410 The following bugs have been fixed or resolved. Note that "n-i-bz"
1411 stands for "not in bugzilla" -- that is, a bug that was reported to us
1412 but never got a bugzilla entry. We encourage you to file bugs in
1413 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1414 than mailing the developers (or mailing lists) directly -- bugs that
1415 are not entered into bugzilla tend to get forgotten about or ignored.
1417 To see details of a given bug, visit
1418 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1419 where XXXXXX is the bug number as listed below.
1421 335440 arm64: ld1 (single structure) is not implemented
1422 335713 arm64: unhanded instruction: prfm (immediate)
1423 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
1424 339182 ppc64: AvSplat ought to load destination vector register with [..]
1425 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
1426 339433 ppc64 lxvw4x instruction uses four 32-byte loads
1427 339645 Use correct tag names in sys_getdents/64 wrappers
1428 339706 Fix false positive for ioctl(TIOCSIG) on linux
1429 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
1430 339853 arm64 times syscall unknown
1431 339855 arm64 unhandled getsid/setsid syscalls
1432 339858 arm64 dmb sy not implemented
1433 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
1434 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
1435 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
1437 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
1438 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
1439 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
1440 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
1441 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
1442 340509 arm64: unhandled instruction fcvtas
1443 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
1444 340632 arm64: unhandled instruction fcvtas
1445 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
1446 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
1447 340788 warning: unhandled syscall: 318 (getrandom)
1448 340807 disInstr(arm): unhandled instruction: 0xEE989B20
1449 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
1450 340922 arm64: unhandled getgroups/setgroups syscalls
1451 350251 Fix typo in VEX utility program (test_main.c).
1452 350407 arm64: unhandled instruction ucvtf (vector, integer)
1453 350809 none/tests/async-sigs breaks when run under cron on Solaris
1454 350811 update README.solaris after r15445
1455 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
1456 350854 strange code in VG_(load_ELF)()
1457 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1458 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
1459 n-i-bz Add missing ]] to terminate CDATA.
1460 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
1461 n-i-bz Enable sys_fadvise64_64 on arm32.
1462 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
1463 n-i-bz Add test cases for all known arm64 load/store instructions.
1464 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
1465 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
1466 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
1467 n-i-bz arm64: Implement "fcvtpu w, s".
1468 n-i-bz arm64: implement ADDP and various others
1469 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
1470 n-i-bz arm64: enable FCVT{A,N}S X,S.
1472 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
1476 Release 3.10.0 (10 September 2014)
1477 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1479 3.10.0 is a feature release with many improvements and the usual
1480 collection of bug fixes.
1482 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1483 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1484 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
1485 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
1486 significantly improved relative to the 3.9.0 release.
1488 * ================== PLATFORM CHANGES =================
1490 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
1491 is mostly complete, and is usable, but some SIMD instructions are as
1494 * Support for little-endian variant of the 64-bit POWER architecture.
1496 * Support for Android on MIPS32.
1498 * Support for 64bit FPU on MIPS32 platforms.
1500 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
1502 * Configuration for and running on Android targets has changed.
1503 See README.android in the source tree for details.
1505 * ================== DEPRECATED FEATURES =================
1507 * --db-attach is now deprecated and will be removed in the next
1508 valgrind feature release. The built-in GDB server capabilities are
1509 superior and should be used instead. Learn more here:
1510 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
1512 * ==================== TOOL CHANGES ====================
1516 - Client code can now selectively disable and re-enable reporting of
1517 invalid address errors in specific ranges using the new client
1518 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
1519 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
1521 - Leak checker: there is a new leak check heuristic called
1522 "length64". This is used to detect interior pointers pointing 8
1523 bytes inside a block, on the assumption that the first 8 bytes
1524 holds the value "block size - 8". This is used by
1525 sqlite3MemMalloc, for example.
1527 - Checking of system call parameters: if a syscall parameter
1528 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
1529 several fields not initialised, an error is now reported for each
1530 field. Previously, an error was reported only for the first
1531 uninitialised field.
1533 - Mismatched alloc/free checking: a new flag
1534 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
1535 such checks if necessary.
1539 - Improvements to error messages:
1541 o Race condition error message involving heap allocated blocks also
1542 show the thread number that allocated the raced-on block.
1544 o All locks referenced by an error message are now announced.
1545 Previously, some error messages only showed the lock addresses.
1547 o The message indicating where a lock was first observed now also
1548 describes the address/location of the lock.
1550 - Helgrind now understands the Ada task termination rules and
1551 creates a happens-before relationship between a terminated task
1552 and its master. This avoids some false positives and avoids a big
1553 memory leak when a lot of Ada tasks are created and terminated.
1554 The interceptions are only activated with forthcoming releases of
1555 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
1557 - A new GDB server monitor command "info locks" giving the list of
1558 locks, their location, and their status.
1562 - callgrind_control now supports the --vgdb-prefix argument,
1563 which is needed if valgrind was started with this same argument.
1565 * ==================== OTHER CHANGES ====================
1567 * Unwinding through inlined function calls. Stack unwinding can now
1568 make use of Dwarf3 inlined-unwind information if it is available.
1569 The practical effect is that inlined calls become visible in stack
1570 traces. The suppression matching machinery has been adjusted
1571 accordingly. This is controlled by the new option
1572 --read-inline-info=yes|no. Currently this is enabled by default
1573 only on Linux and Android targets and only for the tools Memcheck,
1576 * Valgrind can now read EXIDX unwind information on 32-bit ARM
1577 targets. If an object contains both CFI and EXIDX unwind
1578 information, Valgrind will prefer the CFI over the EXIDX. This
1579 facilitates unwinding through system libraries on arm-android
1582 * Address description logic has been improved and is now common
1583 between Memcheck and Helgrind, resulting in better address
1584 descriptions for some kinds of error messages.
1586 * Error messages about dubious arguments (eg, to malloc or calloc) are
1587 output like other errors. This means that they can be suppressed
1588 and they have a stack trace.
1590 * The C++ demangler has been updated for better C++11 support.
1592 * New and modified GDB server monitor features:
1594 - Thread local variables/storage (__thread) can now be displayed.
1596 - The GDB server monitor command "v.info location <address>"
1597 displays information about an address. The information produced
1598 depends on the tool and on the options given to valgrind.
1599 Possibly, the following are described: global variables, local
1600 (stack) variables, allocated or freed blocks, ...
1602 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
1603 ask the GDB server to stop at the start of program execution, at
1604 the end of the program execution and on Valgrind internal errors.
1606 - A new monitor command "v.info stats" shows various Valgrind core
1607 and tool statistics.
1609 - A new monitor command "v.set hostvisibility" allows the GDB server
1610 to provide access to Valgrind internal host status/memory.
1612 * A new option "--aspace-minaddr=<address>" can in some situations
1613 allow the use of more memory by decreasing the address above which
1614 Valgrind maps memory. It can also be used to solve address
1615 conflicts with system libraries by increasing the default value.
1616 See user manual for details.
1618 * The amount of memory used by Valgrind to store debug info (unwind
1619 info, line number information and symbol data) has been
1620 significantly reduced, even though Valgrind now reads more
1621 information in order to support unwinding of inlined function calls.
1623 * Dwarf3 handling with --read-var-info=yes has been improved:
1625 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
1627 - Code compiled with
1628 -ffunction-sections -fdata-sections -Wl,--gc-sections
1629 no longer causes assertion failures.
1631 * Improved checking for the --sim-hints= and --kernel-variant=
1632 options. Unknown strings are now detected and reported to the user
1635 * The semantics of stack start/end boundaries in the valgrind.h
1636 VALGRIND_STACK_REGISTER client request has been clarified and
1637 documented. The convention is that start and end are respectively
1638 the lowest and highest addressable bytes of the stack.
1640 * ==================== FIXED BUGS ====================
1642 The following bugs have been fixed or resolved. Note that "n-i-bz"
1643 stands for "not in bugzilla" -- that is, a bug that was reported to us
1644 but never got a bugzilla entry. We encourage you to file bugs in
1645 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1646 than mailing the developers (or mailing lists) directly -- bugs that
1647 are not entered into bugzilla tend to get forgotten about or ignored.
1649 To see details of a given bug, visit
1650 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1651 where XXXXXX is the bug number as listed below.
1653 175819 Support for ipv6 socket reporting with --track-fds
1654 232510 make distcheck fails
1655 249435 Analyzing wine programs with callgrind triggers a crash
1656 278972 support for inlined function calls in stacktraces and suppression
1658 291310 FXSAVE instruction marks memory as undefined on amd64
1659 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
1660 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
1661 315199 vgcore file for threaded app does not show which thread crashed
1662 315952 tun/tap ioctls are not supported
1663 323178 Unhandled instruction: PLDW register (ARM)
1664 323179 Unhandled instruction: PLDW immediate (ARM)
1665 324050 Helgrind: SEGV because of unaligned stack when using movdqa
1666 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
1667 325124 [MIPSEL] Compilation error
1668 325477 Phase 4 support for IBM Power ISA 2.07
1669 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
1670 325628 Phase 5 support for IBM Power ISA 2.07
1671 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
1672 325751 Missing the two privileged Power PC Transactional Memory Instructions
1673 325816 Phase 6 support for IBM Power ISA 2.07
1674 325856 Make SGCheck fail gracefully on unsupported platforms
1675 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
1676 326436 DRD: False positive in libstdc++ std::list::push_back
1677 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
1678 326462 Refactor vgdb to isolate invoker stuff into separate module
1679 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
1680 326623 DRD: false positive conflict report in a field assignment
1681 326724 Valgrind does not compile on OSX 1.9 Mavericks
1682 326816 Intercept for __strncpy_sse2_unaligned missing?
1683 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
1684 326983 Clear direction flag after tests on amd64.
1685 327212 Do not prepend the current directory to absolute path names.
1686 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
1687 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
1688 327284 s390x: Fix translation of the risbg instruction
1689 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
1690 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
1691 327916 DW_TAG_typedef may have no name
1692 327943 s390x: add a redirection for the 'index' function
1693 328100 XABORT not implemented
1694 328205 Implement additional Xen hypercalls
1695 328454 add support Backtraces with ARM unwind tables (EXIDX)
1696 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
1697 328711 valgrind.1 manpage "memcheck options" section is badly generated
1698 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
1699 329612 Incorrect handling of AT_BASE for image execution
1700 329694 clang warns about using uninitialized variable
1701 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
1702 330228 mmap must align to VKI_SHMLBA on mips32
1703 330257 LLVM does not support `-mno-dynamic-no-pic` option
1704 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
1705 330459 --track-fds=yes doesn't track eventfds
1706 330469 Add clock_adjtime syscall support
1707 330594 Missing sysalls on PowerPC / uClibc
1708 330622 Add test to regression suite for POWER instruction: dcbzl
1709 330939 Support for AMD's syscall instruction on x86
1711 330941 Typo in PRE(poll) syscall wrapper
1712 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
1713 331254 Fix expected output for memcheck/tests/dw4
1714 331255 Fix race condition in test none/tests/coolo_sigaction
1715 331257 Fix type of jump buffer in test none/tests/faultstatus
1716 331305 configure uses bash specific syntax
1717 331337 s390x WARNING: unhandled syscall: 326 (dup3)
1718 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
1719 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
1720 331829 Unexpected ioctl opcode sign extension
1721 331830 ppc64: WARNING: unhandled syscall: 96/97
1722 331839 drd/tests/sem_open specifies invalid semaphore name
1723 331847 outcome of drd/tests/thread_name is nondeterministic
1724 332037 Valgrind cannot handle Thumb "add pc, reg"
1725 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
1726 consistency checks enabled
1727 332263 intercepts for pthread_rwlock_timedrdlock and
1728 pthread_rwlock_timedwrlock are incorrect
1729 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
1731 332276 Implement additional Xen hypercalls
1732 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
1733 332765 Fix ms_print to create temporary files in a proper directory
1734 333072 drd: Add semaphore annotations
1735 333145 Tests for missaligned PC+#imm access for arm
1736 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
1737 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
1738 333248 WARNING: unhandled syscall: unix:443
1739 333428 ldr.w pc [rD, #imm] instruction leads to assertion
1740 333501 cachegrind: assertion: Cache set count is not a power of two.
1743 333666 Recognize MPX instructions and bnd prefix.
1744 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
1745 333817 Valgrind reports the memory areas written to by the SG_IO
1747 334049 lzcnt fails silently (x86_32)
1748 334384 Valgrind does not have support Little Endian support for
1750 334585 recvmmsg unhandled (+patch) (arm)
1751 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
1752 334727 Build fails with -Werror=format-security
1753 334788 clarify doc about --log-file initial program directory
1754 334834 PPC64 Little Endian support, patch 2
1755 334836 PPC64 Little Endian support, patch 3 testcase fixes
1756 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
1757 335034 Unhandled ioctl: HCIGETDEVLIST
1758 335155 vgdb, fix error print statement.
1759 335262 arm64: movi 8bit version is not supported
1760 335263 arm64: dmb instruction is not implemented
1761 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
1762 335496 arm64: sbc/abc instructions are not implemented
1763 335554 arm64: unhandled instruction: abs
1764 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
1765 335735 arm64: unhandled instruction: cnt
1766 335736 arm64: unhandled instruction: uaddlv
1767 335848 arm64: unhandled instruction: {s,u}cvtf
1768 335902 arm64: unhandled instruction: sli
1769 335903 arm64: unhandled instruction: umull (vector)
1770 336055 arm64: unhandled instruction: mov (element)
1771 336062 arm64: unhandled instruction: shrn{,2}
1772 336139 mip64: [...] valgrind hangs and spins on a single core [...]
1773 336189 arm64: unhandled Instruction: mvn
1774 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
1775 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
1776 336772 Make moans about unknown ioctls more informative
1777 336957 Add a section about the Solaris/illumos port on the webpage
1778 337094 ifunc wrapper is broken on ppc64
1779 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
1780 337528 leak check heuristic for block prefixed by length as 64bit number
1781 337740 Implement additional Xen hypercalls
1782 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
1783 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
1784 337871 deprecate --db-attach
1785 338023 Add support for all V4L2/media ioctls
1786 338024 inlined functions are not shown if DW_AT_ranges is used
1787 338106 Add support for 'kcmp' syscall
1788 338115 DRD: computed conflict set differs from actual after fork
1789 338160 implement display of thread local storage in gdbsrv
1790 338205 configure.ac and check for -Wno-tautological-compare
1791 338300 coredumps are missing one byte of every segment
1792 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
1793 338499 --sim-hints parsing broken due to wrong order in tokens
1794 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
1795 338681 Unable to unwind through clone thread created on i386-linux
1796 338698 race condition between gdbsrv and vgdb on startup
1797 338703 helgrind on arm-linux gets false positives in dynamic loader
1798 338791 alt dwz files can be relative of debug/main file
1799 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
1800 338932 build V-trunk with gcc-trunk
1801 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
1802 345079 Fix build problems in VEX/useful/test_main.c
1803 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
1804 n-i-bz s390x: Fix memory corruption for multithreaded applications
1805 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
1806 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
1807 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
1808 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
1809 n-i-bz Add command-line option --default-suppressions
1810 n-i-bz Add support for BLKDISCARDZEROES ioctl
1811 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
1812 n-i-bz Add support for sys_pivot_root and sys_unshare
1814 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
1815 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
1816 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
1820 Release 3.9.0 (31 October 2013)
1821 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1822 3.9.0 is a feature release with many improvements and the usual
1823 collection of bug fixes.
1825 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
1826 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
1827 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
1828 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
1830 * ================== PLATFORM CHANGES =================
1832 * Support for MIPS64 LE and BE running Linux. Valgrind has been
1833 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
1835 * Support for MIPS DSP ASE on MIPS32 platforms.
1837 * Support for s390x Decimal Floating Point instructions on hosts that
1838 have the DFP facility installed.
1840 * Support for POWER8 (Power ISA 2.07) instructions
1842 * Support for Intel AVX2 instructions. This is available only on 64
1845 * Initial support for Intel Transactional Synchronization Extensions,
1848 * Initial support for Hardware Transactional Memory on POWER.
1850 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
1851 run large GUI apps tolerably well.
1853 * ==================== TOOL CHANGES ====================
1857 - Improvements in handling of vectorised code, leading to
1858 significantly fewer false error reports. You need to use the flag
1859 --partial-loads-ok=yes to get the benefits of these changes.
1861 - Better control over the leak checker. It is now possible to
1862 specify which leak kinds (definite/indirect/possible/reachable)
1863 should be displayed, which should be regarded as errors, and which
1864 should be suppressed by a given leak suppression. This is done
1865 using the options --show-leak-kinds=kind1,kind2,..,
1866 --errors-for-leak-kinds=kind1,kind2,.. and an optional
1867 "match-leak-kinds:" line in suppression entries, respectively.
1869 Note that generated leak suppressions contain this new line and
1870 are therefore more specific than in previous releases. To get the
1871 same behaviour as previous releases, remove the "match-leak-kinds:"
1872 line from generated suppressions before using them.
1874 - Reduced "possible leak" reports from the leak checker by the use
1875 of better heuristics. The available heuristics provide detection
1876 of valid interior pointers to std::stdstring, to new[] allocated
1877 arrays with elements having destructors and to interior pointers
1878 pointing to an inner part of a C++ object using multiple
1879 inheritance. They can be selected individually using the
1880 option --leak-check-heuristics=heur1,heur2,...
1882 - Better control of stacktrace acquisition for heap-allocated
1883 blocks. Using the --keep-stacktraces option, it is possible to
1884 control independently whether a stack trace is acquired for each
1885 allocation and deallocation. This can be used to create better
1886 "use after free" errors or to decrease Valgrind's resource
1887 consumption by recording less information.
1889 - Better reporting of leak suppression usage. The list of used
1890 suppressions (shown when the -v option is given) now shows, for
1891 each leak suppressions, how many blocks and bytes it suppressed
1892 during the last leak search.
1896 - False errors resulting from the use of statically initialised
1897 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
1900 - False errors resulting from the use of pthread_cond_waits that
1901 timeout, have been removed.
1903 * ==================== OTHER CHANGES ====================
1905 * Some attempt to tune Valgrind's space requirements to the expected
1906 capabilities of the target:
1908 - The default size of the translation cache has been reduced from 8
1909 sectors to 6 on Android platforms, since each sector occupies
1910 about 40MB when using Memcheck.
1912 - The default size of the translation cache has been increased to 16
1913 sectors on all other platforms, reflecting the fact that large
1914 applications require instrumentation and storage of huge amounts
1915 of code. For similar reasons, the number of memory mapped
1916 segments that can be tracked has been increased by a factor of 6.
1918 - In all cases, the maximum number of sectors in the translation
1919 cache can be controlled by the new flag --num-transtab-sectors.
1921 * Changes in how debug info (line numbers, etc) is read:
1923 - Valgrind no longer temporarily mmaps the entire object to read
1924 from it. Instead, reading is done through a small fixed sized
1925 buffer. This avoids virtual memory usage spikes when Valgrind
1926 reads debuginfo from large shared objects.
1928 - A new experimental remote debug info server. Valgrind can read
1929 debug info from a different machine (typically, a build host)
1930 where debuginfo objects are stored. This can save a lot of time
1931 and hassle when running Valgrind on resource-constrained targets
1932 (phones, tablets) when the full debuginfo objects are stored
1933 somewhere else. This is enabled by the --debuginfo-server=
1936 - Consistency checking between main and debug objects can be
1937 disabled using the --allow-mismatched-debuginfo option.
1939 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
1940 scanning can recover stack traces in some cases when the normal
1941 unwind mechanisms fail. Stack scanning is best described as "a
1942 nasty, dangerous and misleading hack" and so is disabled by default.
1943 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
1946 * Detection and merging of recursive stack frame cycles. When your
1947 program has recursive algorithms, this limits the memory used by
1948 Valgrind for recorded stack traces and avoids recording
1949 uninteresting repeated calls. This is controlled by the command
1950 line option --merge-recursive-frame and by the monitor command
1951 "v.set merge-recursive-frames".
1953 * File name and line numbers for used suppressions. The list of used
1954 suppressions (shown when the -v option is given) now shows, for each
1955 used suppression, the file name and line number where the suppression
1958 * New and modified GDB server monitor features:
1960 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
1961 that can be used to execute gdbserver monitor commands from the
1964 - A new monitor command, "v.info open_fds", that gives the list of
1965 open file descriptors and additional details.
1967 - An optional message in the "v.info n_errs_found" monitor command,
1968 for example "v.info n_errs_found test 1234 finished", allowing a
1969 comment string to be added to the process output, perhaps for the
1970 purpose of separating errors of different tests or test phases.
1972 - A new monitor command "v.info execontext" that shows information
1973 about the stack traces recorded by Valgrind.
1975 - A new monitor command "v.do expensive_sanity_check_general" to run
1976 some internal consistency checks.
1978 * New flag --sigill-diagnostics to control whether a diagnostic
1979 message is printed when the JIT encounters an instruction it can't
1980 translate. The actual behavior -- delivery of SIGILL to the
1981 application -- is unchanged.
1983 * The maximum amount of memory that Valgrind can use on 64 bit targets
1984 has been increased from 32GB to 64GB. This should make it possible
1985 to run applications on Memcheck that natively require up to about 35GB.
1987 * ==================== FIXED BUGS ====================
1989 The following bugs have been fixed or resolved. Note that "n-i-bz"
1990 stands for "not in bugzilla" -- that is, a bug that was reported to us
1991 but never got a bugzilla entry. We encourage you to file bugs in
1992 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1993 than mailing the developers (or mailing lists) directly -- bugs that
1994 are not entered into bugzilla tend to get forgotten about or ignored.
1996 To see details of a given bug, visit
1997 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1998 where XXXXXX is the bug number as listed below.
2000 123837 system call: 4th argument is optional, depending on cmd
2001 135425 memcheck should tell you where Freed blocks were Mallocd
2002 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
2003 207815 Adds some of the drm ioctls to syswrap-linux.c
2004 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
2005 252955 Impossible to compile with ccache
2006 253519 Memcheck reports auxv pointer accesses as invalid reads.
2007 263034 Crash when loading some PPC64 binaries
2008 269599 Increase deepest backtrace
2009 274695 s390x: Support "compare to/from logical" instructions (z196)
2010 275800 s390x: Autodetect cache info (part 2)
2011 280271 Valgrind reports possible memory leaks on still-reachable std::string
2012 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2013 289578 Backtraces with ARM unwind tables (stack scan flags)
2014 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2015 304832 ppc32: build failure
2016 305431 Use find_buildid shdr fallback for separate .debug files
2017 305728 Add support for AVX2 instructions
2018 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2019 306035 s390x: Fix IR generation for LAAG and friends
2020 306054 s390x: Condition code computation for convert-to-int/logical
2021 306098 s390x: alternate opcode form for convert to/from fixed
2022 306587 Fix cache line detection from auxiliary vector for PPC.
2023 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2024 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2025 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2026 307101 sys_capget second argument can be NULL
2027 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2028 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2029 307113 s390x: DFP support
2030 307141 valgrind does't work in mips-linux system
2031 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2032 307285 x86_amd64 feature test for avx in test suite is wrong
2033 307290 memcheck overlap testcase needs memcpy version filter
2034 307463 Please add "&limit=0" to the "all open bugs" link
2035 307465 --show-possibly-lost=no should reduce the error count / exit code
2036 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2037 307729 pkgconfig support broken valgrind.pc
2038 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2039 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2040 308089 Unhandled syscall on ppc64: prctl
2041 308135 PPC32 MPC8xx has 16 bytes cache size
2042 308321 testsuite memcheck filter interferes with gdb_filter
2044 308341 vgdb should report process exit (or fatal signal)
2045 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2046 308495 Remove build dependency on installed Xen headers
2047 308573 Internal error on 64-bit instruction executed in 32-bit mode
2049 308627 pmovmskb validity bit propagation is imprecise
2050 308644 vgdb command for having the info for the track-fds option
2051 308711 give more info about aspacemgr and arenas in out_of_memory
2052 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2053 308718 ARM implement SMLALBB family of instructions
2054 308886 Missing support for PTRACE_SET/GETREGSET
2055 308930 syscall name_to_handle_at (303 on amd64) not handled
2056 309229 V-bit tester does not report number of tests generated
2057 309323 print unrecognized instuction on MIPS
2058 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2059 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2060 309430 Self hosting ppc64 encounters a vassert error on operand type
2061 309600 valgrind is a bit confused about 0-sized sections
2062 309823 Generate errors for still reachable blocks
2063 309921 PCMPISTRI validity bit propagation is imprecise
2064 309922 none/tests/ppc64/test_dfp5 sometimes fails
2065 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2066 310424 --read-var-info does not properly describe static variables
2067 310792 search additional path for debug symbols
2068 310931 s390x: Message-security assist (MSA) instruction extension [..]
2069 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2070 311318 ARM: "128-bit constant is not implemented" error message
2071 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2072 311690 V crashes because it redirects branches inside of a redirected function
2073 311880 x86_64: make regtest hangs at shell_valid1
2074 311922 WARNING: unhandled syscall: 170
2076 312171 ppc: insn selection for DFP
2077 312571 Rounding mode call wrong for the DFP Iops [..]
2078 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2079 312913 Dangling pointers error should also report the alloc stack trace
2080 312980 Building on Mountain Lion generates some compiler warnings
2081 313267 Adding MIPS64/Linux port to Valgrind
2084 313811 Buffer overflow in assert_fail
2085 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2086 314269 ppc: dead code in insn selection
2087 314718 ARM: implement integer divide instruction (sdiv and udiv)
2088 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2089 315441 sendmsg syscall should ignore unset msghdr msg_flags
2090 315534 msgrcv inside a thread causes valgrind to hang (block)
2091 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2092 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2093 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2094 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2095 316144 valgrind.1 manpage contains unknown ??? strings [..]
2096 316145 callgrind command line options in manpage reference (unknown) [..]
2097 316145 callgrind command line options in manpage reference [..]
2098 316181 drd: Fixed a 4x slowdown for certain applications
2099 316503 Valgrind does not support SSE4 "movntdqa" instruction
2100 316535 Use of |signed int| instead of |size_t| in valgrind messages
2101 316696 fluidanimate program of parsec 2.1 stuck
2102 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2103 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2104 317186 "Impossible happens" when occurs VCVT instruction on ARM
2105 317318 Support for Threading Building Blocks "scalable_malloc"
2106 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2107 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2108 317463 bmi testcase IR SANITY CHECK FAILURE
2109 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2110 318050 libmpiwrap fails to compile with out-of-source build
2111 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2112 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2113 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2114 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2115 318932 Add missing PPC64 and PPC32 system call support
2116 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2117 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2118 319494 VEX Makefile-gcc standalone build update after r2702
2119 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2120 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2121 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2122 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2123 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2124 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2125 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2126 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2127 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2128 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2129 320895 add fanotify support (patch included)
2130 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2131 321065 Valgrind updates for Xen 4.3
2132 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2133 321363 Unhandled instruction: SSAX (ARM + Thumb)
2134 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2135 321466 Unhandled instruction: SHASX (ARM + Thumb)
2136 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2137 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2138 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2139 321620 Unhandled instruction: UASX (ARM + Thumb)
2140 321621 Unhandled instruction: USAX (ARM + Thumb)
2141 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2142 321693 Unhandled instruction: LDRSBT (Thumb)
2143 321694 Unhandled instruction: UQASX (ARM + Thumb)
2144 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2145 321697 Unhandled instruction: UHASX (ARM + Thumb)
2146 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2147 321704 Unhandled instruction: REVSH (ARM + Thumb)
2148 321730 Add cg_diff and cg_merge man pages
2149 321738 Add vgdb and valgrind-listener man pages
2151 321891 Unhandled instruction: LDRHT (Thumb)
2152 321960 pthread_create() then alloca() causing invalid stack write errors
2153 321969 ppc32 and ppc64 don't support [lf]setxattr
2154 322254 Show threadname together with tid if set by application
2155 322294 Add initial support for IBM Power ISA 2.07
2156 322368 Assertion failure in wqthread_hijack under OS X 10.8
2157 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2158 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2159 322851 0bXXX binary literal syntax is not standard
2160 323035 Unhandled instruction: LDRSHT(Thumb)
2161 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2162 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2163 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2164 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2165 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2166 323437 Phase 2 support for IBM Power ISA 2.07
2167 323713 Support mmxext (integer sse) subset on i386 (athlon)
2168 323803 Transactional memory instructions are not supported for Power
2169 323893 SSE3 not available on amd cpus in valgrind
2170 323905 Probable false positive from Valgrind/drd on close()
2171 323912 valgrind.h header isn't compatible for mingw64
2172 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
2173 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
2174 324181 mmap does not handle MAP_32BIT
2175 324227 memcheck false positive leak when a thread calls exit+block [..]
2176 324421 Support for fanotify API on ARM architecture
2177 324514 gdbserver monitor cmd output behaviour consistency [..]
2178 324518 ppc64: Emulation of dcbt instructions does not handle [..]
2179 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
2180 324582 When access is made to freed memory, report both allocation [..]
2181 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
2182 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
2183 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
2184 324834 Unhandled instructions in Microsoft C run-time for x86_64
2185 324894 Phase 3 support for IBM Power ISA 2.07
2186 326091 drd: Avoid false race reports from optimized strlen() impls
2187 326113 valgrind libvex hwcaps error on AMD64
2188 n-i-bz Some wrong command line options could be ignored
2189 n-i-bz patch to allow fair-sched on android
2190 n-i-bz report error for vgdb snapshot requested before execution
2191 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
2193 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
2197 Release 3.8.1 (19 September 2012)
2198 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2199 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
2200 that occur moderately frequently in real use cases, adds support for
2201 some missing instructions on ARM, and fixes a deadlock condition on
2202 MacOSX. If you package or deliver 3.8.0 for others to use, you might
2203 want to consider upgrading to 3.8.1 instead.
2205 The following bugs have been fixed or resolved. Note that "n-i-bz"
2206 stands for "not in bugzilla" -- that is, a bug that was reported to us
2207 but never got a bugzilla entry. We encourage you to file bugs in
2208 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2209 than mailing the developers (or mailing lists) directly -- bugs that
2210 are not entered into bugzilla tend to get forgotten about or ignored.
2212 To see details of a given bug, visit
2213 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2214 where XXXXXX is the bug number as listed below.
2217 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
2218 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
2219 298281 wcslen causes false(?) uninitialised value warnings
2220 301281 valgrind hangs on OS X when the process calls system()
2221 304035 disInstr(arm): unhandled instruction 0xE1023053
2222 304867 implement MOVBE instruction in x86 mode
2223 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
2224 305042 amd64: implement 0F 7F encoding of movq between two registers
2225 305199 ARM: implement QDADD and QDSUB
2226 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
2227 305513 killed by fatal signal: SIGSEGV
2228 305690 DRD reporting invalid semaphore when sem_trywait fails
2229 305926 Invalid alignment checks for some AVX instructions
2230 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
2231 306310 3.8.0 release tarball missing some files
2232 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
2233 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
2234 n-i-bz shmat of a segment > 4Gb does not work
2235 n-i-bz simulate_control_c script wrong USR1 signal number on mips
2236 n-i-bz vgdb ptrace calls wrong on mips [...]
2237 n-i-bz Fixes for more MPI false positives
2238 n-i-bz exp-sgcheck's memcpy causes programs to segfault
2239 n-i-bz OSX build w/ clang: asserts at startup
2240 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
2241 n-i-bz fix a couple of union tag-vs-field mixups
2242 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
2244 The following bugs were fixed in 3.8.0 but not listed in this NEWS
2247 254088 Valgrind should know about UD2 instruction
2250 304754 NEWS blows TeX's little mind
2252 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
2256 Release 3.8.0 (10 August 2012)
2257 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2258 3.8.0 is a feature release with many improvements and the usual
2259 collection of bug fixes.
2261 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2262 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
2263 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
2264 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
2265 There is initial support for MacOSX 10.8, but it is not usable for
2266 serious work at present.
2268 * ================== PLATFORM CHANGES =================
2270 * Support for MIPS32 platforms running Linux. Valgrind has been
2271 tested on MIPS32 and MIPS32r2 platforms running different Debian
2272 Squeeze and MeeGo distributions. Both little-endian and big-endian
2273 cores are supported. The tools Memcheck, Massif and Lackey have
2274 been tested and are known to work. See README.mips for more details.
2276 * Preliminary support for Android running on x86.
2278 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
2280 * Support for Intel AVX instructions and for AES instructions. This
2281 support is available only for 64 bit code.
2283 * Support for POWER Decimal Floating Point instructions.
2285 * ==================== TOOL CHANGES ====================
2287 * Non-libc malloc implementations are now supported. This is useful
2288 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
2289 Using the new option --soname-synonyms, such tools can be informed
2290 that the malloc implementation is either linked statically into the
2291 executable, or is present in some other shared library different
2292 from libc.so. This makes it possible to process statically linked
2293 programs, and programs using other malloc libraries, for example
2294 TCMalloc or JEMalloc.
2296 * For tools that provide their own replacement for malloc et al, the
2297 option --redzone-size=<number> allows users to specify the size of
2298 the padding blocks (redzones) added before and after each client
2299 allocated block. Smaller redzones decrease the memory needed by
2300 Valgrind. Bigger redzones increase the chance to detect blocks
2301 overrun or underrun. Prior to this change, the redzone size was
2302 hardwired to 16 bytes in Memcheck.
2306 - The leak_check GDB server monitor command now can
2307 control the maximum nr of loss records to output.
2309 - Reduction of memory use for applications allocating
2310 many blocks and/or having many partially defined bytes.
2312 - Addition of GDB server monitor command 'block_list' that lists
2313 the addresses/sizes of the blocks of a leak search loss record.
2315 - Addition of GDB server monitor command 'who_points_at' that lists
2316 the locations pointing at a block.
2318 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2319 detect an invalid access of these redzones, by marking them
2320 noaccess. Similarly, if a redzone size is given for a memory
2321 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2322 This still allows to find some bugs if the user has forgotten to
2323 mark the pool superblock noaccess.
2325 - Performance of memory leak check has been improved, especially in
2326 cases where there are many leaked blocks and/or many suppression
2327 rules used to suppress leak reports.
2329 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2330 more precise analysis, which is important for LLVM/Clang
2331 generated code. This is at the cost of somewhat reduced
2332 performance. Note there is no change to analysis precision or
2333 costs on Linux targets.
2337 - Added even more facilities that can help finding the cause of a data
2338 race, namely the command-line option --ptrace-addr and the macro
2339 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2341 - Fixed a subtle bug that could cause false positive data race reports.
2343 * ==================== OTHER CHANGES ====================
2345 * The C++ demangler has been updated so as to work well with C++
2346 compiled by up to at least g++ 4.6.
2348 * Tool developers can make replacement/wrapping more flexible thanks
2349 to the new option --soname-synonyms. This was reported above, but
2350 in fact is very general and applies to all function
2351 replacement/wrapping, not just to malloc-family functions.
2353 * Round-robin scheduling of threads can be selected, using the new
2354 option --fair-sched= yes. Prior to this change, the pipe-based
2355 thread serialisation mechanism (which is still the default) could
2356 give very unfair scheduling. --fair-sched=yes improves
2357 responsiveness of interactive multithreaded applications, and
2358 improves repeatability of results from the thread checkers Helgrind
2361 * For tool developers: support to run Valgrind on Valgrind has been
2362 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2364 * gdbserver now shows the float shadow registers as integer
2365 rather than float values, as the shadow values are mostly
2366 used as bit patterns.
2368 * Increased limit for the --num-callers command line flag to 500.
2370 * Performance improvements for error matching when there are many
2371 suppression records in use.
2373 * Improved support for DWARF4 debugging information (bug 284184).
2375 * Initial support for DWZ compressed Dwarf debug info.
2377 * Improved control over the IR optimiser's handling of the tradeoff
2378 between performance and precision of exceptions. Specifically,
2379 --vex-iropt-precise-memory-exns has been removed and replaced by
2380 --vex-iropt-register-updates, with extended functionality. This
2381 allows the Valgrind gdbserver to always show up to date register
2384 * Modest performance gains through the use of translation chaining for
2387 * ==================== FIXED BUGS ====================
2389 The following bugs have been fixed or resolved. Note that "n-i-bz"
2390 stands for "not in bugzilla" -- that is, a bug that was reported to us
2391 but never got a bugzilla entry. We encourage you to file bugs in
2392 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2393 than mailing the developers (or mailing lists) directly -- bugs that
2394 are not entered into bugzilla tend to get forgotten about or ignored.
2396 To see details of a given bug, visit
2397 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2398 where XXXXXX is the bug number as listed below.
2400 197914 Building valgrind from svn now requires automake-1.10
2401 203877 increase to 16Mb maximum allowed alignment for memalign et al
2402 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
2403 247386 make perf does not run all performance tests
2404 270006 Valgrind scheduler unfair
2405 270777 Adding MIPS/Linux port to Valgrind
2406 270796 s390x: Removed broken support for the TS insn
2407 271438 Fix configure for proper SSE4.2 detection
2408 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
2409 273475 Add support for AVX instructions
2410 274078 improved configure logic for mpicc
2411 276993 fix mremap 'no thrash checks'
2412 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
2413 281482 memcheck incorrect byte allocation count in realloc() for silly argument
2414 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
2415 283413 Fix wrong sanity check
2416 283671 Robustize alignment computation in LibVEX_Alloc
2417 283961 Adding support for some HCI IOCTLs
2418 284124 parse_type_DIE: confused by: DWARF 4
2419 284864 == 273475 (Add support for AVX instructions)
2420 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
2421 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
2422 285725 == 273475 (Add support for AVX instructions)
2423 286261 add wrapper for linux I2C_RDWR ioctl
2424 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
2425 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
2426 286384 configure fails "checking for a supported version of gcc"
2427 286497 == 273475 (Add support for AVX instructions)
2428 286596 == 273475 (Add support for AVX instructions)
2429 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
2430 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
2431 287260 Incorrect conditional jump or move depends on uninitialised value(s)
2432 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
2433 287307 == 273475 (Add support for AVX instructions)
2434 287858 VG_(strerror): unknown error
2435 288298 (MacOSX) unhandled syscall shm_unlink
2436 288995 == 273475 (Add support for AVX instructions)
2437 289470 Loading of large Mach-O thin binaries fails.
2438 289656 == 273475 (Add support for AVX instructions)
2439 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
2440 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
2441 289839 s390x: Provide support for unicode conversion instructions
2442 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
2443 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
2444 290655 Add support for AESKEYGENASSIST instruction
2445 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
2446 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
2447 291253 ES register not initialised in valgrind simulation
2448 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
2449 291865 s390x: Support the "Compare Double and Swap" family of instructions
2450 292300 == 273475 (Add support for AVX instructions)
2451 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
2452 292493 == 273475 (Add support for AVX instructions)
2453 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
2454 292627 Missing support for some SCSI ioctls
2455 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
2456 292841 == 273475 (Add support for AVX instructions)
2457 292993 implement the getcpu syscall on amd64-linux
2458 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
2459 293088 Add some VEX sanity checks for ppc64 unhandled instructions
2460 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
2461 293754 PCMPxSTRx not implemented for 16-bit characters
2462 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
2463 293808 CLFLUSH not supported by latest VEX for amd64
2464 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
2465 294048 MPSADBW instruction not implemented
2466 294055 regtest none/tests/shell fails when locale is not set to C
2467 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
2468 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
2469 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
2470 294260 disInstr_AMD64: disInstr miscalculated next %rip
2471 294523 --partial-loads-ok=yes causes false negatives
2472 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
2473 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
2474 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
2475 295089 can not annotate source for both helgrind and drd
2476 295221 POWER Processor decimal floating point instruction support missing
2477 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
2478 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
2479 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
2480 295617 ARM - Add some missing syscalls
2481 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
2482 296229 Linux user input device ioctls missing wrappers
2483 296318 ELF Debug info improvements (more than one rx/rw mapping)
2484 296422 Add translation chaining support
2485 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
2486 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
2487 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
2488 297078 gdbserver signal handling problems [..]
2489 297147 drd false positives on newly allocated memory
2490 297329 disallow decoding of IBM Power DFP insns on some machines
2491 297497 POWER Processor decimal floating point instruction support missing
2492 297701 Another alias for strncasecmp_l in libc-2.13.so
2493 297911 'invalid write' not reported when using APIs for custom mem allocators.
2494 297976 s390x: revisit EX implementation
2495 297991 Valgrind interferes with mmap()+ftell()
2496 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
2497 297993 Fix compilation of valgrind with gcc -g3.
2498 298080 POWER Processor DFP support missing, part 3
2499 298227 == 273475 (Add support for AVX instructions)
2500 298335 == 273475 (Add support for AVX instructions)
2501 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
2502 298394 s390x: Don't bail out on an unknown machine model. [..]
2503 298421 accept4() syscall (366) support is missing for ARM
2504 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
2505 298732 valgrind installation problem in ubuntu with kernel version 3.x
2506 298862 POWER Processor DFP instruction support missing, part 4
2507 298864 DWARF reader mis-parses DW_FORM_ref_addr
2508 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
2509 299053 Support DWARF4 DW_AT_high_pc constant form
2510 299104 == 273475 (Add support for AVX instructions)
2511 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
2512 299629 dup3() syscall (358) support is missing for ARM
2513 299694 POWER Processor DFP instruction support missing, part 5
2514 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
2515 299803 == 273475 (Add support for AVX instructions)
2516 299804 == 273475 (Add support for AVX instructions)
2517 299805 == 273475 (Add support for AVX instructions)
2518 300140 ARM - Missing (T1) SMMUL
2519 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
2520 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
2521 300414 FCOM and FCOMP unimplemented for amd64 guest
2522 301204 infinite loop in canonicaliseSymtab with ifunc symbol
2523 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
2524 301265 add x86 support to Android build
2525 301984 configure script doesn't detect certain versions of clang
2526 302205 Fix compiler warnings for POWER VEX code and POWER test cases
2527 302287 Unhandled movbe instruction on Atom processors
2528 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
2529 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
2530 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
2531 302656 == 273475 (Add support for AVX instructions)
2532 302709 valgrind for ARM needs extra tls support for android emulator [..]
2533 302827 add wrapper for CDROM_GET_CAPABILITY
2534 302901 Valgrind crashes with dwz optimized debuginfo
2535 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
2536 303116 Add support for the POWER instruction popcntb
2537 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
2538 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
2539 303466 == 273475 (Add support for AVX instructions)
2540 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
2541 303963 strstr() function produces wrong results under valgrind callgrind
2542 304054 CALL_FN_xx macros need to enforce stack alignment
2543 304561 tee system call not supported
2544 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
2545 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
2546 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
2547 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
2548 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
2549 n-i-bz s390x: Shadow registers can now be examined using vgdb
2551 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
2552 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
2556 Release 3.7.0 (5 November 2011)
2557 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2558 3.7.0 is a feature release with many significant improvements and the
2559 usual collection of bug fixes.
2561 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2562 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
2563 Support for recent distros and toolchain components (glibc 2.14, gcc
2564 4.6, MacOSX 10.7) has been added.
2566 * ================== PLATFORM CHANGES =================
2568 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
2569 analyse 64-bit programs running on z/Architecture. Most user space
2570 instructions up to and including z10 are supported. Valgrind has
2571 been tested extensively on z9, z10, and z196 machines running SLES
2572 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
2573 known to work well. Callgrind, Helgrind, and DRD work reasonably
2574 well on z9 and later models. See README.s390 for more details.
2576 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
2577 64-bit processes are supported. Some complex threaded applications
2578 (Firefox) are observed to hang when run as 32 bit applications,
2579 whereas 64-bit versions run OK. The cause is unknown. Memcheck
2580 will likely report some false errors. In general, expect some rough
2581 spots. This release also supports MacOSX 10.6, but drops support
2584 * Preliminary support for Android (on ARM). Valgrind can now run
2585 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
2586 README.android for more details, plus instructions on how to get
2589 * Support for the IBM Power ISA 2.06 (Power7 instructions)
2591 * General correctness and performance improvements for ARM/Linux, and,
2592 by extension, ARM/Android.
2594 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
2595 instruction set support is under development but is not available in
2598 * Support for AIX5 has been removed.
2600 * ==================== TOOL CHANGES ====================
2602 * Memcheck: some incremental changes:
2604 - reduction of memory use in some circumstances
2606 - improved handling of freed memory, which in some circumstances
2607 can cause detection of use-after-free that would previously have
2610 - fix of a longstanding bug that could cause false negatives (missed
2611 errors) in programs doing vector saturated narrowing instructions.
2613 * Helgrind: performance improvements and major memory use reductions,
2614 particularly for large, long running applications which perform many
2615 synchronisation (lock, unlock, etc) events. Plus many smaller
2618 - display of locksets for both threads involved in a race
2620 - general improvements in formatting/clarity of error messages
2622 - addition of facilities and documentation regarding annotation
2623 of thread safe reference counted C++ classes
2625 - new flag --check-stack-refs=no|yes [yes], to disable race checking
2626 on thread stacks (a performance hack)
2628 - new flag --free-is-write=no|yes [no], to enable detection of races
2629 where one thread accesses heap memory but another one frees it,
2630 without any coordinating synchronisation event
2632 * DRD: enabled XML output; added support for delayed thread deletion
2633 in order to detect races that occur close to the end of a thread
2634 (--join-list-vol); fixed a memory leak triggered by repeated client
2635 memory allocatation and deallocation; improved Darwin support.
2637 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
2639 * exp-sgcheck: this tool has been reduced in scope so as to improve
2640 performance and remove checking that Memcheck does better.
2641 Specifically, the ability to check for overruns for stack and global
2642 arrays is unchanged, but the ability to check for overruns of heap
2643 blocks has been removed. The tool has accordingly been renamed to
2644 exp-sgcheck ("Stack and Global Array Checking").
2646 * ==================== OTHER CHANGES ====================
2648 * GDB server: Valgrind now has an embedded GDB server. That means it
2649 is possible to control a Valgrind run from GDB, doing all the usual
2650 things that GDB can do (single stepping, breakpoints, examining
2651 data, etc). Tool-specific functionality is also available. For
2652 example, it is possible to query the definedness state of variables
2653 or memory from within GDB when running Memcheck; arbitrarily large
2654 memory watchpoints are supported, etc. To use the GDB server, start
2655 Valgrind with the flag --vgdb-error=0 and follow the on-screen
2658 * Improved support for unfriendly self-modifying code: a new option
2659 --smc-check=all-non-file is available. This adds the relevant
2660 consistency checks only to code that originates in non-file-backed
2661 mappings. In effect this confines the consistency checking only to
2662 code that is or might be JIT generated, and avoids checks on code
2663 that must have been compiled ahead of time. This significantly
2664 improves performance on applications that generate code at run time.
2666 * It is now possible to build a working Valgrind using Clang-2.9 on
2669 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
2670 These enable and disable error reporting on a per-thread, and
2671 nestable, basis. This is useful for hiding errors in particularly
2672 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
2673 now uses this facility.
2675 * Added the --mod-funcname option to cg_diff.
2677 * ==================== FIXED BUGS ====================
2679 The following bugs have been fixed or resolved. Note that "n-i-bz"
2680 stands for "not in bugzilla" -- that is, a bug that was reported to us
2681 but never got a bugzilla entry. We encourage you to file bugs in
2682 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2683 mailing the developers (or mailing lists) directly -- bugs that are
2684 not entered into bugzilla tend to get forgotten about or ignored.
2686 To see details of a given bug, visit
2687 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2688 where XXXXXX is the bug number as listed below.
2690 79311 malloc silly arg warning does not give stack trace
2691 210935 port valgrind.h (not valgrind) to win32 to support client requests
2692 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
2693 243404 Port to zSeries
2694 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
2695 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
2696 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
2697 253206 Some fixes for the faultstatus testcase
2698 255223 capget testcase fails when running as root
2699 256703 xlc_dbl_u32.c testcase broken
2700 256726 Helgrind tests have broken inline asm
2701 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
2702 264800 testcase compile failure on zseries
2703 265762 make public VEX headers compilable by G++ 3.x
2704 265771 assertion in jumps.c (r11523) fails with glibc-2.3
2705 266753 configure script does not give the user the option to not use QtCore
2706 266931 gen_insn_test.pl is broken
2707 266961 ld-linux.so.2 i?86-linux strlen issues
2708 266990 setns instruction causes false positive
2709 267020 Make directory for temporary files configurable at run-time.
2710 267342 == 267997 (segmentation fault on Mac OS 10.6)
2711 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
2712 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
2713 267488 regtest: darwin support for 64-bit build
2714 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
2715 267630 Add support for IBM Power ISA 2.06 -- stage 1
2716 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
2717 267819 Add client request for informing the core about reallocation
2718 267925 laog data structure quadratic for a single sequence of lock
2719 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
2720 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
2721 268513 missed optimizations in fold_Expr
2722 268619 s390x: fpr - gpr transfer facility
2723 268620 s390x: reconsider "long displacement" requirement
2724 268621 s390x: improve IR generation for XC
2725 268715 s390x: FLOGR is not universally available
2726 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
2727 268930 s390x: MHY is not universally available
2728 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
2729 269079 Support ptrace system call on ARM
2730 269144 missing "Bad option" error message
2731 269209 conditional load and store facility (z196)
2732 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
2733 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
2734 269736 s390x: minor code generation tweaks
2735 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
2736 269863 s390x: remove unused function parameters
2737 269864 s390x: tweak s390_emit_load_cc
2738 269884 == 250101 (overhead for huge blocks exhausts space too soon)
2739 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
2740 270115 s390x: rewrite some testcases
2741 270309 == 267997 (valgrind crash on startup)
2742 270320 add support for Linux FIOQSIZE ioctl() call
2743 270326 segfault while trying to sanitize the environment passed to execle
2744 270794 IBM POWER7 support patch causes regression in none/tests
2745 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
2746 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
2747 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
2748 270959 s390x: invalid use of R0 as base register
2749 271042 VSX configure check fails when it should not
2750 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
2751 271259 s390x: fix code confusion
2752 271337 == 267997 (Valgrind segfaults on MacOS X)
2753 271385 s390x: Implement Ist_MBE
2754 271501 s390x: misc cleanups
2755 271504 s390x: promote likely and unlikely
2756 271579 ppc: using wrong enum type
2757 271615 unhandled instruction "popcnt" (arch=amd10h)
2758 271730 Fix bug when checking ioctls: duplicate check
2759 271776 s390x: provide STFLE instruction support
2760 271779 s390x: provide clock instructions like STCK
2761 271799 Darwin: ioctls without an arg report a memory error
2762 271820 arm: fix type confusion
2763 271917 pthread_cond_timedwait failure leads to not-locked false positive
2764 272067 s390x: fix DISP20 macro
2765 272615 A typo in debug output in mc_leakcheck.c
2766 272661 callgrind_annotate chokes when run from paths containing regex chars
2767 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
2768 272955 Unhandled syscall error for pwrite64 on ppc64 arch
2769 272967 make documentation build-system more robust
2770 272986 Fix gcc-4.6 warnings with valgrind.h
2771 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
2772 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
2773 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
2774 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
2775 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
2776 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
2777 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
2778 273778 exp-ptrcheck: unhandled sysno == 259
2779 274089 exp-ptrcheck: unhandled sysno == 208
2780 274378 s390x: Various dispatcher tweaks
2781 274447 WARNING: unhandled syscall: 340
2782 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
2783 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
2784 274926 valgrind does not build against linux-3
2785 275148 configure FAIL with glibc-2.14
2786 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
2787 275168 Make Valgrind work for MacOSX 10.7 Lion
2788 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
2789 275278 valgrind does not build on Linux kernel 3.0.* due to silly
2790 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
2791 275308 Fix implementation for ppc64 fres instruc
2792 275339 s390x: fix testcase compile warnings
2793 275517 s390x: Provide support for CKSM instruction
2794 275710 s390x: get rid of redundant address mode calculation
2795 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
2796 275852 == 250101 (valgrind uses all swap space and is killed)
2797 276784 Add support for IBM Power ISA 2.06 -- stage 3
2798 276987 gdbsrv: fix tests following recent commits
2799 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
2800 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
2801 277471 Unhandled syscall: 340
2802 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
2803 277653 ARM: support Thumb2 PLD instruction
2804 277663 ARM: NEON float VMUL by scalar incorrect
2805 277689 ARM: tests for VSTn with register post-index are broken
2806 277694 ARM: BLX LR instruction broken in ARM mode
2807 277780 ARM: VMOV.F32 (immediate) instruction is broken
2808 278057 fuse filesystem syscall deadlocks
2809 278078 Unimplemented syscall 280 on ppc32
2810 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
2811 278454 VALGRIND_STACK_DEREGISTER has wrong output type
2812 278502 == 275284 (Valgrind confuses memcpy() and memmove())
2813 278892 gdbsrv: factorize gdb version handling, fix doc and typos
2814 279027 Support for MVCL and CLCL instruction
2815 279027 s390x: Provide support for CLCL and MVCL instructions
2816 279062 Remove a redundant check in the insn selector for ppc.
2817 279071 JDK creates PTEST with redundant REX.W prefix
2818 279212 gdbsrv: add monitor cmd v.info scheduler.
2819 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
2820 279698 memcheck discards valid-bits for packuswb
2821 279795 memcheck reports uninitialised values for mincore on amd64
2822 279994 Add support for IBM Power ISA 2.06 -- stage 3
2823 280083 mempolicy syscall check errors
2824 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
2825 280710 s390x: config files for nightly builds
2826 280757 /tmp dir still used by valgrind even if TMPDIR is specified
2827 280965 Valgrind breaks fcntl locks when program does mmap
2828 281138 WARNING: unhandled syscall: 340
2829 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
2830 281304 == 275168 (Darwin: dyld "cannot load inserted library")
2831 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
2832 281468 s390x: handle do_clone and gcc clones in call traces
2833 281488 ARM: VFP register corruption
2834 281828 == 275284 (false memmove warning: "Source and destination overlap")
2835 281883 s390x: Fix system call wrapper for "clone".
2836 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
2837 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
2838 282238 SLES10: make check fails
2839 282979 strcasestr needs replacement with recent(>=2.12) glibc
2840 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
2841 283243 Regression in ppc64 memcheck tests
2842 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
2843 283427 re-connect epoll_pwait syscall on ARM linux
2844 283600 gdbsrv: android: port vgdb.c
2845 283709 none/tests/faultstatus needs to account for page size
2846 284305 filter_gdb needs enhancement to work on ppc64
2847 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
2848 284472 Thumb2 ROR.W encoding T2 not implemented
2849 284621 XML-escape process command line in XML output
2850 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
2851 that have non-power-of-2 sizes (also AMDs)
2852 n-i-bz don't be spooked by libraries mashed by elfhack
2853 n-i-bz don't be spooked by libxul.so linked with gold
2854 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
2856 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
2857 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
2858 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
2862 Release 3.6.1 (16 February 2011)
2863 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2864 3.6.1 is a bug fix release. It adds support for some SSE4
2865 instructions that were omitted in 3.6.0 due to lack of time. Initial
2866 support for glibc-2.13 has been added. A number of bugs causing
2867 crashing or assertion failures have been fixed.
2869 The following bugs have been fixed or resolved. Note that "n-i-bz"
2870 stands for "not in bugzilla" -- that is, a bug that was reported to us
2871 but never got a bugzilla entry. We encourage you to file bugs in
2872 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2873 mailing the developers (or mailing lists) directly -- bugs that are
2874 not entered into bugzilla tend to get forgotten about or ignored.
2876 To see details of a given bug, visit
2877 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2878 where XXXXXX is the bug number as listed below.
2880 188572 Valgrind on Mac should suppress setenv() mem leak
2881 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
2882 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
2883 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
2884 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
2885 254420 memory pool tracking broken
2886 254957 Test code failing to compile due to changes in memcheck.h
2887 255009 helgrind/drd: crash on chmod with invalid parameter
2888 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
2889 255355 helgrind/drd: crash on threaded programs doing fork
2891 255418 (SSE4.x) rint call compiled with ICC
2892 255822 --gen-suppressions can create invalid files: "too many callers [...]"
2893 255888 closing valgrindoutput tag outputted to log-stream on error
2894 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
2895 255966 Slowness when using mempool annotations
2896 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
2897 256600 super-optimized strcasecmp() false positive
2898 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
2899 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
2900 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
2901 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
2902 257276 Missing case in memcheck --track-origins=yes
2903 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
2904 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
2905 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
2906 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
2907 263099 callgrind_annotate counts Ir improperly [...]
2908 263877 undefined coprocessor instruction on ARMv7
2909 265964 configure FAIL with glibc-2.13
2910 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
2911 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
2912 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
2913 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
2914 n-i-bz DRD: disable --free-is-write due to implementation difficulties
2916 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
2920 Release 3.6.0 (21 October 2010)
2921 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2922 3.6.0 is a feature release with many significant improvements and the
2923 usual collection of bug fixes.
2925 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2926 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
2927 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
2929 -------------------------
2931 Here are some highlights. Details are shown further down:
2933 * Support for ARM/Linux.
2935 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
2937 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
2939 * Support for the SSE4.2 instruction set.
2941 * Enhancements to the Callgrind profiler, including the ability to
2942 handle CPUs with three levels of cache.
2944 * A new experimental heap profiler, DHAT.
2946 * A huge number of bug fixes and small enhancements.
2948 -------------------------
2950 Here are details of the above changes, together with descriptions of
2951 many other changes, and a list of fixed bugs.
2953 * ================== PLATFORM CHANGES =================
2955 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
2956 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
2957 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
2959 This requires a CPU capable of running the ARMv7-A instruction set
2960 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
2961 of the user space instruction set, including ARM and Thumb integer
2962 code, VFPv3, NEON and V6 media instructions. The Memcheck,
2963 Cachegrind and Massif tools work properly; other tools work to
2966 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
2967 with support for recent releases of the underlying toolchain
2968 components, notably gcc-4.5 and glibc-2.12.
2970 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
2971 support also works much better on OS X 10.5, and is as solid as
2974 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
2975 64-bit mode. In 32-bit mode, support is only available up to and
2976 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
2977 supported in 64-bit mode, and 32-bit mode does in fact support the
2978 bare minimum SSE4 instructions to needed to run programs on Mac OS X
2979 10.6 on 32-bit targets.
2981 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
2982 and including version 2.05 is supported.
2984 * ==================== TOOL CHANGES ====================
2986 * Cachegrind has a new processing script, cg_diff, which finds the
2987 difference between two profiles. It's very useful for evaluating
2988 the performance effects of a change in a program.
2990 Related to this change, the meaning of cg_annotate's (rarely-used)
2991 --threshold option has changed; this is unlikely to affect many
2992 people, if you do use it please see the user manual for details.
2994 * Callgrind now can do branch prediction simulation, similar to
2995 Cachegrind. In addition, it optionally can count the number of
2996 executed global bus events. Both can be used for a better
2997 approximation of a "Cycle Estimation" as derived event (you need to
2998 update the event formula in KCachegrind yourself).
3000 * Cachegrind and Callgrind now refer to the LL (last-level) cache
3001 rather than the L2 cache. This is to accommodate machines with
3002 three levels of caches -- if Cachegrind/Callgrind auto-detects the
3003 cache configuration of such a machine it will run the simulation as
3004 if the L2 cache isn't present. This means the results are less
3005 likely to match the true result for the machine, but
3006 Cachegrind/Callgrind's results are already only approximate, and
3007 should not be considered authoritative. The results are still
3008 useful for giving a general idea about a program's locality.
3010 * Massif has a new option, --pages-as-heap, which is disabled by
3011 default. When enabled, instead of tracking allocations at the level
3012 of heap blocks (as allocated with malloc/new/new[]), it instead
3013 tracks memory allocations at the level of memory pages (as mapped by
3014 mmap, brk, etc). Each mapped page is treated as its own block.
3015 Interpreting the page-level output is harder than the heap-level
3016 output, but this option is useful if you want to account for every
3017 byte of memory used by a program.
3019 * DRD has two new command-line options: --free-is-write and
3020 --trace-alloc. The former allows to detect reading from already freed
3021 memory, and the latter allows tracing of all memory allocations and
3024 * DRD has several new annotations. Custom barrier implementations can
3025 now be annotated, as well as benign races on static variables.
3027 * DRD's happens before / happens after annotations have been made more
3028 powerful, so that they can now also be used to annotate e.g. a smart
3029 pointer implementation.
3031 * Helgrind's annotation set has also been drastically improved, so as
3032 to provide to users a general set of annotations to describe locks,
3033 semaphores, barriers and condition variables. Annotations to
3034 describe thread-safe reference counted heap objects have also been
3037 * Memcheck has a new command-line option, --show-possibly-lost, which
3038 is enabled by default. When disabled, the leak detector will not
3039 show possibly-lost blocks.
3041 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3042 has been added. DHAT keeps track of allocated heap blocks, and also
3043 inspects every memory reference to see which block (if any) is being
3044 accessed. This gives a lot of insight into block lifetimes,
3045 utilisation, turnover, liveness, and the location of hot and cold
3046 fields. You can use DHAT to do hot-field profiling.
3048 * ==================== OTHER CHANGES ====================
3050 * Improved support for unfriendly self-modifying code: the extra
3051 overhead incurred by --smc-check=all has been reduced by
3052 approximately a factor of 5 as compared with 3.5.0.
3054 * Ability to show directory names for source files in error messages.
3055 This is combined with a flexible mechanism for specifying which
3056 parts of the paths should be shown. This is enabled by the new flag
3059 * A new flag, --require-text-symbol, which will stop the run if a
3060 specified symbol is not found it a given shared object when it is
3061 loaded into the process. This makes advanced working with function
3062 intercepting and wrapping safer and more reliable.
3064 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3065 and control of Valgrind is now available for the tools Memcheck and
3066 Helgrind. XML output from Valgrind is available for Memcheck,
3067 Helgrind and exp-Ptrcheck.
3069 * More reliable stack unwinding on amd64-linux, particularly in the
3070 presence of function wrappers, and with gcc-4.5 compiled code.
3072 * Modest scalability (performance improvements) for massive
3073 long-running applications, particularly for those with huge amounts
3076 * Support for analyzing programs running under Wine with has been
3077 improved. The header files <valgrind/valgrind.h>,
3078 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3079 Windows-programs compiled with MinGW or one of the Microsoft Visual
3082 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3083 The 32-bit simulator was not affected. This did not occur often,
3084 but when it did would usually crash the program under test.
3087 * A large number of bugs were fixed. These are shown below.
3089 * A number of bugs were investigated, and were candidates for fixing,
3090 but are not fixed in 3.6.0, due to lack of developer time. They may
3091 get fixed in later releases. They are:
3093 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3094 212419 false positive "lock order violated" (A+B vs A)
3095 213685 Undefined value propagates past dependency breaking instruction
3096 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3097 237920 valgrind segfault on fork failure
3098 242137 support for code compiled by LLVM-2.8
3099 242423 Another unknown Intel cache config value
3100 243232 Inconsistent Lock Orderings report with trylock
3101 243483 ppc: callgrind triggers VEX assertion failure
3102 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3103 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3105 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3106 249435 Analyzing wine programs with callgrind triggers a crash
3107 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3108 250065 Handling large allocations
3109 250101 huge "free" memory usage due to m_mallocfree.c
3110 "superblocks fragmentation"
3111 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3112 252091 Callgrind on ARM does not detect function returns correctly
3113 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3114 254420 memory pool tracking broken
3115 n-i-bz support for adding symbols for JIT generated code
3118 The following bugs have been fixed or resolved. Note that "n-i-bz"
3119 stands for "not in bugzilla" -- that is, a bug that was reported to us
3120 but never got a bugzilla entry. We encourage you to file bugs in
3121 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3122 mailing the developers (or mailing lists) directly -- bugs that are
3123 not entered into bugzilla tend to get forgotten about or ignored.
3125 To see details of a given bug, visit
3126 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3127 where XXXXXX is the bug number as listed below.
3129 135264 dcbzl instruction missing
3131 153699 Valgrind should report unaligned reads with movdqa
3133 190429 Valgrind reports lost of errors in ld.so
3134 with x86_64 2.9.90 glibc
3135 197266 valgrind appears to choke on the xmms instruction
3137 197988 Crash when demangling very large symbol names
3138 202315 unhandled syscall: 332 (inotify_init1)
3139 203256 Add page-level profiling to Massif
3140 205093 dsymutil=yes needs quotes, locking (partial fix)
3141 205241 Snow Leopard 10.6 support (partial fix)
3142 206600 Leak checker fails to upgrade indirect blocks when their
3143 parent becomes reachable
3144 210935 port valgrind.h (not valgrind) to win32 so apps run under
3145 wine can make client requests
3146 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3147 within Linux ip-stack checksum functions
3148 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3150 213685 Undefined value propagates past dependency breaking instruction
3152 215914 Valgrind inserts bogus empty environment variable
3154 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3155 222545 shmat fails under valgind on some arm targets
3156 222560 ARM NEON support
3159 232509 Docs build fails with formatting inside <title></title> elements
3161 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3162 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3163 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3164 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3165 237485 symlink (syscall 57) is not supported on Mac OS
3166 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3168 238208 is_just_below_ESP doesn't take into account red-zone
3169 238345 valgrind passes wrong $0 when executing a shell script
3170 238679 mq_timedreceive syscall doesn't flag the reception buffer
3172 238696 fcntl command F_DUPFD_CLOEXEC not supported
3173 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3174 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3175 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
3177 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
3183 242606 unhandled syscall: setegid (in Ptrcheck)
3184 242814 Helgrind "Impossible has happened" during
3185 QApplication::initInstance();
3186 243064 Valgrind attempting to read debug information from iso
3187 243270 Make stack unwinding in Valgrind wrappers more reliable
3188 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
3189 sysno = 277 (mq_open)
3190 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
3191 244493 ARM VFP d16-d31 registers support
3192 244670 add support for audit_session_self syscall on Mac OS 10.6
3193 244921 The xml report of helgrind tool is not well format
3194 244923 In the xml report file, the <preamble> not escape the
3195 xml char, eg '<','&','>'
3196 245535 print full path names in plain text reports
3197 245925 x86-64 red zone handling problem
3198 246258 Valgrind not catching integer underruns + new [] s
3199 246311 reg/reg cmpxchg doesn't work on amd64
3200 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
3201 246888 Improve Makefile.vex.am
3202 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
3203 to [f]chmod_extended
3204 247526 IBM POWER6 (ISA 2.05) support is incomplete
3205 247561 Some leak testcases fails due to reachable addresses in
3207 247875 sizeofIRType to handle Ity_I128
3208 247894 [PATCH] unhandled syscall sys_readahead
3209 247980 Doesn't honor CFLAGS passed to configure
3210 248373 darwin10.supp is empty in the trunk
3211 248822 Linux FIBMAP ioctl has int parameter instead of long
3212 248893 [PATCH] make readdwarf.c big endianess safe to enable
3213 unwinding on big endian systems
3214 249224 Syscall 336 not supported (SYS_proc_info)
3216 249775 Incorrect scheme for detecting NEON capabilities of host CPU
3217 249943 jni JVM init fails when using valgrind
3218 249991 Valgrind incorrectly declares AESKEYGENASSIST support
3220 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
3221 250799 frexp$fenv_access_off function generates SIGILL
3222 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
3223 251251 support pclmulqdq insn
3224 251362 valgrind: ARM: attach to debugger either fails or provokes
3226 251674 Unhandled syscall 294
3229 254257 Add support for debugfiles found by build-id
3230 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
3231 254646 Wrapped functions cause stack misalignment on OS X
3232 (and possibly Linux)
3233 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
3235 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
3239 Release 3.5.0 (19 August 2009)
3240 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3241 3.5.0 is a feature release with many significant improvements and the
3242 usual collection of bug fixes. The main improvement is that Valgrind
3243 now works on Mac OS X.
3245 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
3246 and X86/Darwin. Support for recent distros and toolchain components
3247 (glibc 2.10, gcc 4.5) has been added.
3249 -------------------------
3251 Here is a short summary of the changes. Details are shown further
3254 * Support for Mac OS X (10.5.x).
3256 * Improvements and simplifications to Memcheck's leak checker.
3258 * Clarification and simplifications in various aspects of Valgrind's
3261 * XML output for Helgrind and Ptrcheck.
3263 * Performance and stability improvements for Helgrind and DRD.
3265 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
3267 * A new experimental tool, BBV, useful for computer architecture
3270 * Improved Wine support, including ability to read Windows PDB
3273 -------------------------
3275 Here are details of the above changes, followed by descriptions of
3276 many other minor changes, and a list of fixed bugs.
3279 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
3280 called "Darwin" because that is the name of the OS core, which is the
3281 level that Valgrind works at.)
3285 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
3286 because it would require work and 10.4 is only becoming less common.
3288 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
3289 fairly well. For 10.5.x, 32-bit programs are the default even on
3290 64-bit machines, so it handles most current programs.
3292 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
3293 officially supported, but simple programs at least will probably work.
3294 However, start-up is slow.
3296 - PowerPC machines are not supported.
3298 Things that don't work:
3300 - The Ptrcheck tool.
3302 - Objective-C garbage collection.
3306 - If you have Rogue Amoeba's "Instant Hijack" program installed,
3307 Valgrind will fail with a SIGTRAP at start-up. See
3308 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
3313 - You will likely find --dsymutil=yes a useful option, as error
3314 messages may be imprecise without it.
3316 - Mac OS X support is new and therefore will be less robust than the
3317 Linux support. Please report any bugs you find.
3319 - Threaded programs may run more slowly than on Linux.
3321 Many thanks to Greg Parker for developing this port over several years.
3324 * Memcheck's leak checker has been improved.
3326 - The results for --leak-check=summary now match the summary results
3327 for --leak-check=full. Previously they could differ because
3328 --leak-check=summary counted "indirectly lost" blocks and
3329 "suppressed" blocks as "definitely lost".
3331 - Blocks that are only reachable via at least one interior-pointer,
3332 but are directly pointed to by a start-pointer, were previously
3333 marked as "still reachable". They are now correctly marked as
3336 - The default value for the --leak-resolution option has been
3337 changed from "low" to "high". In general, this means that more
3338 leak reports will be produced, but each leak report will describe
3339 fewer leaked blocks.
3341 - With --leak-check=full, "definitely lost" and "possibly lost"
3342 leaks are now considered as proper errors, ie. they are counted
3343 for the "ERROR SUMMARY" and affect the behaviour of
3344 --error-exitcode. These leaks are not counted as errors if
3345 --leak-check=summary is specified, however.
3347 - Documentation for the leak checker has been improved.
3350 * Various aspects of Valgrind's text output have changed.
3352 - Valgrind's start-up message has changed. It is shorter but also
3353 includes the command being run, which makes it easier to use
3354 --trace-children=yes. An example:
3356 - Valgrind's shut-down messages have also changed. This is most
3357 noticeable with Memcheck, where the leak summary now occurs before
3358 the error summary. This change was necessary to allow leaks to be
3359 counted as proper errors (see the description of the leak checker
3360 changes above for more details). This was also necessary to fix a
3361 longstanding bug in which uses of suppressions against leaks were
3362 not "counted", leading to difficulties in maintaining suppression
3363 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3365 - Behavior of -v has changed. In previous versions, -v printed out
3366 a mixture of marginally-user-useful information, and tool/core
3367 statistics. The statistics printing has now been moved to its own
3368 flag, --stats=yes. This means -v is less verbose and more likely
3369 to convey useful end-user information.
3371 - The format of some (non-XML) stack trace entries has changed a
3372 little. Previously there were six possible forms:
3374 0x80483BF: really (a.c:20)
3375 0x80483BF: really (in /foo/a.out)
3377 0x80483BF: (within /foo/a.out)
3378 0x80483BF: ??? (a.c:20)
3381 The third and fourth of these forms have been made more consistent
3382 with the others. The six possible forms are now:
3384 0x80483BF: really (a.c:20)
3385 0x80483BF: really (in /foo/a.out)
3386 0x80483BF: really (in ???)
3387 0x80483BF: ??? (in /foo/a.out)
3388 0x80483BF: ??? (a.c:20)
3391 Stack traces produced when --xml=yes is specified are different
3395 * Helgrind and Ptrcheck now support XML output, so they can be used
3396 from GUI tools. Also, the XML output mechanism has been
3399 - The XML format has been overhauled and generalised, so it is more
3400 suitable for error reporting tools in general. The Memcheck
3401 specific aspects of it have been removed. The new format, which
3402 is an evolution of the old format, is described in
3403 docs/internals/xml-output-protocol4.txt.
3405 - Memcheck has been updated to use the new format.
3407 - Helgrind and Ptrcheck are now able to emit output in this format.
3409 - The XML output mechanism has been overhauled. XML is now output
3410 to its own file descriptor, which means that:
3412 * Valgrind can output text and XML independently.
3414 * The longstanding problem of XML output being corrupted by
3415 unexpected un-tagged text messages is solved.
3417 As before, the destination for text output is specified using
3418 --log-file=, --log-fd= or --log-socket=.
3420 As before, XML output for a tool is enabled using --xml=yes.
3422 Because there's a new XML output channel, the XML output
3423 destination is now specified by --xml-file=, --xml-fd= or
3426 Initial feedback has shown this causes some confusion. To
3427 clarify, the two envisaged usage scenarios are:
3429 (1) Normal text output. In this case, do not specify --xml=yes
3430 nor any of --xml-file=, --xml-fd= or --xml-socket=.
3432 (2) XML output. In this case, specify --xml=yes, and one of
3433 --xml-file=, --xml-fd= or --xml-socket= to select the XML
3434 destination, one of --log-file=, --log-fd= or --log-socket=
3435 to select the destination for any remaining text messages,
3436 and, importantly, -q.
3438 -q makes Valgrind completely silent on the text channel,
3439 except in the case of critical failures, such as Valgrind
3440 itself segfaulting, or failing to read debugging information.
3441 Hence, in this scenario, it suffices to check whether or not
3442 any output appeared on the text channel. If yes, then it is
3443 likely to be a critical error which should be brought to the
3444 attention of the user. If no (the text channel produced no
3445 output) then it can be assumed that the run was successful.
3447 This allows GUIs to make the critical distinction they need to
3448 make (did the run fail or not?) without having to search or
3449 filter the text output channel in any way.
3451 It is also recommended to use --child-silent-after-fork=yes in
3455 * Improvements and changes in Helgrind:
3457 - XML output, as described above
3459 - Checks for consistent association between pthread condition
3460 variables and their associated mutexes are now performed.
3462 - pthread_spinlock functions are supported.
3464 - Modest performance improvements.
3466 - Initial (skeletal) support for describing the behaviour of
3467 non-POSIX synchronisation objects through ThreadSanitizer
3468 compatible ANNOTATE_* macros.
3470 - More controllable tradeoffs between performance and the level of
3471 detail of "previous" accesses in a race. There are now three
3474 * --history-level=full. This is the default, and was also the
3475 default in 3.4.x. It shows both stacks involved in a race, but
3476 requires a lot of memory and can be very slow in programs that
3477 do many inter-thread synchronisation events.
3479 * --history-level=none. This only shows the later stack involved
3480 in a race. This can be much faster than --history-level=full,
3481 but makes it much more difficult to find the other access
3482 involved in the race.
3484 The new intermediate setting is
3486 * --history-level=approx
3488 For the earlier (other) access, two stacks are presented. The
3489 earlier access is guaranteed to be somewhere in between the two
3490 program points denoted by those stacks. This is not as useful
3491 as showing the exact stack for the previous access (as per
3492 --history-level=full), but it is better than nothing, and it's
3493 almost as fast as --history-level=none.
3496 * New features and improvements in DRD:
3498 - The error messages printed by DRD are now easier to interpret.
3499 Instead of using two different numbers to identify each thread
3500 (Valgrind thread ID and DRD thread ID), DRD does now identify
3501 threads via a single number (the DRD thread ID). Furthermore
3502 "first observed at" information is now printed for all error
3503 messages related to synchronization objects.
3505 - Added support for named semaphores (sem_open() and sem_close()).
3507 - Race conditions between pthread_barrier_wait() and
3508 pthread_barrier_destroy() calls are now reported.
3510 - Added support for custom allocators through the macros
3511 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
3512 in <valgrind/valgrind.h>). An alternative for these two macros is
3513 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
3516 - Added support for annotating non-POSIX synchronization objects
3517 through several new ANNOTATE_*() macros.
3519 - OpenMP: added support for the OpenMP runtime (libgomp) included
3520 with gcc versions 4.3.0 and 4.4.0.
3524 - Added two new command-line options (--first-race-only and
3525 --segment-merging-interval).
3528 * Genuinely atomic support for x86/amd64/ppc atomic instructions
3530 Valgrind will now preserve (memory-access) atomicity of LOCK-
3531 prefixed x86/amd64 instructions, and any others implying a global
3532 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
3534 This means that Valgrinded processes will "play nicely" in
3535 situations where communication with other processes, or the kernel,
3536 is done through shared memory and coordinated with such atomic
3537 instructions. Prior to this change, such arrangements usually
3538 resulted in hangs, races or other synchronisation failures, because
3539 Valgrind did not honour atomicity of such instructions.
3542 * A new experimental tool, BBV, has been added. BBV generates basic
3543 block vectors for use with the SimPoint analysis tool, which allows
3544 a program's overall behaviour to be approximated by running only a
3545 fraction of it. This is useful for computer architecture
3546 researchers. You can run BBV by specifying --tool=exp-bbv (the
3547 "exp-" prefix is short for "experimental"). BBV was written by
3551 * Modestly improved support for running Windows applications under
3552 Wine. In particular, initial support for reading Windows .PDB debug
3553 information has been added.
3556 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
3557 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
3561 * The Valgrind client requests VALGRIND_PRINTF and
3562 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
3563 the string was always printed immediately on its own line. Now, the
3564 string will be added to a buffer but not printed until a newline is
3565 encountered, or other Valgrind output is printed (note that for
3566 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
3567 "other Valgrind output"). This allows you to use multiple
3568 VALGRIND_PRINTF calls to build up a single output line, and also to
3569 print multiple output lines with a single request (by embedding
3570 multiple newlines in the string).
3573 * The graphs drawn by Massif's ms_print program have changed slightly:
3575 - The half-height chars '.' and ',' are no longer drawn, because
3576 they are confusing. The --y option can be used if the default
3577 y-resolution is not high enough.
3579 - Horizontal lines are now drawn after the top of a snapshot if
3580 there is a gap until the next snapshot. This makes it clear that
3581 the memory usage has not dropped to zero between snapshots.
3584 * Something that happened in 3.4.0, but wasn't clearly announced: the
3585 option --read-var-info=yes can be used by some tools (Memcheck,
3586 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
3587 variable type and location information. This makes those tools
3588 start up more slowly and increases memory consumption, but
3589 descriptions of data addresses in error messages become more
3593 * exp-Omega, an experimental instantaneous leak-detecting tool, was
3594 disabled in 3.4.0 due to a lack of interest and maintenance,
3595 although the source code was still in the distribution. The source
3596 code has now been removed from the distribution. For anyone
3597 interested, the removal occurred in SVN revision r10247.
3600 * Some changes have been made to the build system.
3602 - VEX/ is now integrated properly into the build system. This means
3603 that dependency tracking within VEX/ now works properly, "make
3604 install" will work without requiring "make" before it, and
3605 parallel builds (ie. 'make -j') now work (previously a
3606 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
3607 was effectively ignored).
3609 - The --with-vex configure option has been removed. It was of
3610 little use and removing it simplified the build system.
3612 - The location of some install files has changed. This should not
3613 affect most users. Those who might be affected:
3615 * For people who use Valgrind with MPI programs, the installed
3616 libmpiwrap.so library has moved from
3617 $(INSTALL)/<platform>/libmpiwrap.so to
3618 $(INSTALL)/libmpiwrap-<platform>.so.
3620 * For people who distribute standalone Valgrind tools, the
3621 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
3622 have moved to $(INSTALL)/libcoregrind-<platform>.a.
3624 These changes simplify the build system.
3626 - Previously, all the distributed suppression (*.supp) files were
3627 installed. Now, only default.supp is installed. This should not
3628 affect users as the other installed suppression files were not
3629 read; the fact that they were installed was a mistake.
3632 * KNOWN LIMITATIONS:
3634 - Memcheck is unusable with the Intel compiler suite version 11.1,
3635 when it generates code for SSE2-and-above capable targets. This
3636 is because of icc's use of highly optimised inlined strlen
3637 implementations. It causes Memcheck to report huge numbers of
3638 false errors even in simple programs. Helgrind and DRD may also
3641 Versions 11.0 and earlier may be OK, but this has not been
3645 The following bugs have been fixed or resolved. Note that "n-i-bz"
3646 stands for "not in bugzilla" -- that is, a bug that was reported to us
3647 but never got a bugzilla entry. We encourage you to file bugs in
3648 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3649 mailing the developers (or mailing lists) directly -- bugs that are
3650 not entered into bugzilla tend to get forgotten about or ignored.
3652 To see details of a given bug, visit
3653 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3654 where XXXXXX is the bug number as listed below.
3656 84303 How about a LockCheck tool?
3657 91633 dereference of null ptr in vgPlain_st_basetype
3658 97452 Valgrind doesn't report any pthreads problems
3659 100628 leak-check gets assertion failure when using
3660 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
3661 108528 NPTL pthread cleanup handlers not called
3662 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
3663 110128 mallinfo is not implemented...
3664 110770 VEX: Generated files not always updated when making valgrind
3665 111102 Memcheck: problems with large (memory footprint) applications
3666 115673 Vex's decoder should never assert
3667 117564 False positive: Syscall param clone(child_tidptr) contains
3668 uninitialised byte(s)
3669 119404 executing ssh from inside valgrind fails
3670 133679 Callgrind does not write path names to sources with dwarf debug
3672 135847 configure.in problem with non gnu compilers (and possible fix)
3673 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
3674 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
3675 136230 memcheck reports "possibly lost", should be "still reachable"
3676 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
3677 137904 Valgrind reports a memory leak when using POSIX threads,
3679 139076 valgrind VT_GETSTATE error
3680 142228 complaint of elf_dynamic_do_rela in trivial usage
3681 145347 spurious warning with USBDEVFS_REAPURB
3682 148441 (wine) can't find memory leak in Wine, win32 binary
3684 148742 Leak-check fails assert on exit
3685 149878 add (proper) check for calloc integer overflow
3686 150606 Call graph is broken when using callgrind control
3687 152393 leak errors produce an exit code of 0. I need some way to
3688 cause leak errors to result in a nonzero exit code.
3689 157154 documentation (leak-resolution doc speaks about num-callers
3690 def=4) + what is a loss record
3691 159501 incorrect handling of ALSA ioctls
3692 162020 Valgrinding an empty/zero-byte file crashes valgrind
3693 162482 ppc: Valgrind crashes while reading stabs information
3694 162718 x86: avoid segment selector 0 in sys_set_thread_area()
3695 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
3696 163560 VEX/test_main.c is missing from valgrind-3.3.1
3697 164353 malloc_usable_size() doesn't return a usable size
3698 165468 Inconsistent formatting in memcheck manual -- please fix
3699 169505 main.c:286 (endOfInstr):
3700 Assertion 'ii->cost_offset == *cost_offset' failed
3701 177206 Generate default.supp during compile instead of configure
3702 177209 Configure valt_load_address based on arch+os
3703 177305 eventfd / syscall 323 patch lost
3704 179731 Tests fail to build because of inlining of non-local asm labels
3705 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
3706 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
3707 181594 Bogus warning for empty text segment
3708 181707 dwarf doesn't require enumerations to have name
3709 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3710 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3711 Assertion '!already_present' failed.
3712 185359 exp-ptrcheck: unhandled syscall getresuid()
3713 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
3714 185816 Valgrind is unable to handle debug info for files with split
3715 debug info that are prelinked afterwards
3716 185980 [darwin] unhandled syscall: sem_open
3717 186238 bbToIR_AMD64: disInstr miscalculated next %rip
3718 186507 exp-ptrcheck unhandled syscalls prctl, etc.
3719 186790 Suppression pattern used for leaks are not reported
3720 186796 Symbols with length>200 in suppression files are ignored
3721 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
3722 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
3723 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
3724 188046 bashisms in the configure script
3725 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
3726 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
3727 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
3728 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
3729 assertion fail "!lock->heldBy"
3730 188427 Add support for epoll_create1 (with patch)
3731 188530 Support for SIOCGSTAMPNS
3732 188560 Include valgrind.spec in the tarball
3733 188572 Valgrind on Mac should suppress setenv() mem leak
3734 189054 Valgrind fails to build because of duplicate non-local asm labels
3735 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
3736 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
3737 189763 drd assertion failure: s_threadinfo[tid].is_recording
3738 190219 unhandled syscall: 328 (x86-linux)
3739 190391 dup of 181394; see above
3740 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
3741 190820 No debug information on powerpc-linux
3742 191095 PATCH: Improve usbdevfs ioctl handling
3743 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
3745 191189 --xml=yes should obey --gen-suppressions=all
3746 191192 syslog() needs a suppression on macosx
3747 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
3748 191761 getrlimit on MacOSX
3749 191992 multiple --fn-skip only works sometimes; dependent on order
3750 192634 V. reports "aspacem sync_check_mapping_callback:
3751 segment mismatch" on Darwin
3752 192954 __extension__ missing on 2 client requests
3753 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
3754 194474 "INSTALL" file has different build instructions than "README"
3755 194671 Unhandled syscall (sem_wait?) from mac valgrind
3756 195069 memcheck: reports leak (memory still reachable) for
3758 195169 drd: (vgDrd_barrier_post_wait):
3759 Assertion 'r->sg[p->post_iteration]' failed.
3760 195268 valgrind --log-file doesn't accept ~/...
3761 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
3762 195860 WARNING: unhandled syscall: unix:223
3763 196528 need a error suppression for pthread_rwlock_init under os x?
3764 197227 Support aio_* syscalls on Darwin
3765 197456 valgrind should reject --suppressions=(directory)
3766 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
3767 197591 unhandled syscall 27 (mincore)
3768 197793 Merge DCAS branch to the trunk == 85756, 142103
3769 197794 Avoid duplicate filenames in Vex
3770 197898 make check fails on current SVN
3771 197901 make check fails also under exp-ptrcheck in current SVN
3772 197929 Make --leak-resolution=high the default
3773 197930 Reduce spacing between leak reports
3774 197933 Print command line of client at start-up, and shorten preamble
3775 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
3776 198395 add BBV to the distribution as an experimental tool
3777 198624 Missing syscalls on Darwin: 82, 167, 281, 347
3778 198649 callgrind_annotate doesn't cumulate counters
3779 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
3780 199977 Valgrind complains about an unrecognized instruction in the
3781 atomic_incs test program
3782 200029 valgrind isn't able to read Fedora 12 debuginfo
3783 200760 darwin unhandled syscall: unix:284
3784 200827 DRD doesn't work on Mac OS X
3785 200990 VG_(read_millisecond_timer)() does not work correctly
3786 201016 Valgrind does not support pthread_kill() on Mac OS
3787 201169 Document --read-var-info
3788 201323 Pre-3.5.0 performance sanity checking
3789 201384 Review user manual for the 3.5.0 release
3790 201585 mfpvr not implemented on ppc
3791 201708 tests failing because x86 direction flag is left set
3792 201757 Valgrind doesn't handle any recent sys_futex additions
3793 204377 64-bit valgrind can not start a shell script
3794 (with #!/path/to/shell) if the shell is a 32-bit executable
3795 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
3796 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
3797 about memory allocation events with memory access tracing enabled
3798 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
3800 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
3804 Release 3.4.1 (28 February 2009)
3805 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3806 3.4.1 is a bug-fix release that fixes some regressions and assertion
3807 failures in debug info reading in 3.4.0, most notably incorrect stack
3808 traces on amd64-linux on older (glibc-2.3 based) systems. Various
3809 other debug info problems are also fixed. A number of bugs in the
3810 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
3812 In view of the fact that 3.4.0 contains user-visible regressions
3813 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
3814 encouraged to ship 3.4.1 in preference to 3.4.0.
3816 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
3817 bugzilla" -- that is, a bug that was reported to us but never got a
3818 bugzilla entry. We encourage you to file bugs in bugzilla
3819 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
3820 developers (or mailing lists) directly -- bugs that are not entered
3821 into bugzilla tend to get forgotten about or ignored.
3823 n-i-bz Fix various bugs reading icc-11 generated debug info
3824 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
3825 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
3826 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
3827 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
3828 179618 exp-ptrcheck crashed / exit prematurely
3829 179624 helgrind: false positive races with pthread_create and
3830 recv/open/close/read
3831 134207 pkg-config output contains @VG_PLATFORM@
3832 176926 floating point exception at valgrind startup with PPC 440EPX
3833 181594 Bogus warning for empty text segment
3834 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
3835 181707 Dwarf3 doesn't require enumerations to have name
3836 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3837 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3838 Assertion '!already_present' failed.
3839 185359 exp-ptrcheck unhandled syscall getresuid()
3841 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
3842 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
3846 Release 3.4.0 (2 January 2009)
3847 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3848 3.4.0 is a feature release with many significant improvements and the
3849 usual collection of bug fixes. This release supports X86/Linux,
3850 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
3851 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
3853 3.4.0 brings some significant tool improvements. Memcheck can now
3854 report the origin of uninitialised values, the thread checkers
3855 Helgrind and DRD are much improved, and we have a new experimental
3856 tool, exp-Ptrcheck, which is able to detect overruns of stack and
3857 global arrays. In detail:
3859 * Memcheck is now able to track the origin of uninitialised values.
3860 When it reports an uninitialised value error, it will try to show
3861 the origin of the value, as either a heap or stack allocation.
3862 Origin tracking is expensive and so is not enabled by default. To
3863 use it, specify --track-origins=yes. Memcheck's speed will be
3864 essentially halved, and memory usage will be significantly
3865 increased. Nevertheless it can drastically reduce the effort
3866 required to identify the root cause of uninitialised value errors,
3867 and so is often a programmer productivity win, despite running more
3870 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
3871 3.4.0, will be released shortly.
3873 * Helgrind's race detection algorithm has been completely redesigned
3874 and reimplemented, to address usability and scalability concerns:
3876 - The new algorithm has a lower false-error rate: it is much less
3877 likely to report races that do not really exist.
3879 - Helgrind will display full call stacks for both accesses involved
3880 in a race. This makes it easier to identify the root causes of
3883 - Limitations on the size of program that can run have been removed.
3885 - Performance has been modestly improved, although that is very
3888 - Direct support for Qt4 threading has been added.
3890 - pthread_barriers are now directly supported.
3892 - Helgrind works well on all supported Linux targets.
3894 * The DRD thread debugging tool has seen major improvements:
3896 - Greatly improved performance and significantly reduced memory
3899 - Support for several major threading libraries (Boost.Thread, Qt4,
3900 glib, OpenMP) has been added.
3902 - Support for atomic instructions, POSIX semaphores, barriers and
3903 reader-writer locks has been added.
3905 - Works now on PowerPC CPUs too.
3907 - Added support for printing thread stack usage at thread exit time.
3909 - Added support for debugging lock contention.
3911 - Added a manual for Drd.
3913 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
3914 checks for misuses of pointers. In that sense it is a bit like
3915 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
3916 detect overruns of stack and global arrays, it can detect
3917 arbitrarily far out-of-bounds accesses to heap blocks, and it can
3918 detect accesses heap blocks that have been freed a very long time
3919 ago (millions of blocks in the past).
3921 Ptrcheck currently works only on x86-linux and amd64-linux. To use
3922 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
3923 of the main Valgrind documentation. As this is an experimental
3924 tool, we would be particularly interested in hearing about your
3925 experiences with it.
3927 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
3928 longer built by default, although the code remains in the repository
3929 and the tarball. This is due to three factors: a perceived lack of
3930 users, a lack of maintenance, and concerns that it may not be
3931 possible to achieve reliable operation using the existing design.
3933 * As usual, support for the latest Linux distros and toolchain
3934 components has been added. It should work well on Fedora Core 10,
3935 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
3936 state) is supported, as is glibc-2.9. The C++ demangler has been
3937 updated so as to work well with C++ compiled by even the most recent
3940 * You can now use frame-level wildcards in suppressions. This was a
3941 frequently-requested enhancement. A line "..." in a suppression now
3942 matches zero or more frames. This makes it easier to write
3943 suppressions which are precise yet insensitive to changes in
3946 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
3948 * Very basic support for IBM Power6 has been added (64-bit processes only).
3950 * Valgrind is now cross-compilable. For example, it is possible to
3951 cross compile Valgrind on an x86/amd64-linux host, so that it runs
3952 on a ppc32/64-linux target.
3954 * You can set the main thread's stack size at startup using the
3955 new --main-stacksize= flag (subject of course to ulimit settings).
3956 This is useful for running apps that need a lot of stack space.
3958 * The limitation that you can't use --trace-children=yes together
3959 with --db-attach=yes has been removed.
3961 * The following bugs have been fixed. Note that "n-i-bz" stands for
3962 "not in bugzilla" -- that is, a bug that was reported to us but
3963 never got a bugzilla entry. We encourage you to file bugs in
3964 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3965 mailing the developers (or mailing lists) directly.
3967 n-i-bz Make return types for some client requests 64-bit clean
3968 n-i-bz glibc 2.9 support
3969 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
3970 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
3971 n-i-bz Building in an env without gdb gives bogus gdb attach
3972 92456 Tracing the origin of uninitialised memory
3973 106497 Valgrind does not demangle some C++ template symbols
3975 151612 Suppression with "..." (frame-level wildcards in .supp files)
3976 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
3977 159285 unhandled syscall:25 (stime, on x86-linux)
3978 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
3979 160954 ppc build of valgrind crashes with illegal instruction (isel)
3980 160956 mallinfo implementation, w/ patch
3981 162092 Valgrind fails to start gnome-system-monitor
3982 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
3983 163794 assertion failure with "--track-origins=yes"
3984 163933 sigcontext.err and .trapno must be set together
3985 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
3986 164476 Missing kernel module loading system calls
3987 164669 SVN regression: mmap() drops posix file locks
3988 166581 Callgrind output corruption when program forks
3989 167288 Patch file for missing system calls on Cell BE
3990 168943 unsupported scas instruction pentium
3991 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
3992 172417 x86->IR: 0x82 ...
3993 172563 amd64->IR: 0xD9 0xF5 - fprem1
3994 173099 .lds linker script generation error
3995 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
3996 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
3998 174908 --log-file value not expanded correctly for core file
3999 175044 Add lookup_dcookie for amd64
4000 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
4002 Developer-visible changes:
4004 * Valgrind's debug-info reading machinery has been majorly overhauled.
4005 It can now correctly establish the addresses for ELF data symbols,
4006 which is something that has never worked properly before now.
4008 Also, Valgrind can now read DWARF3 type and location information for
4009 stack and global variables. This makes it possible to use the
4010 framework to build tools that rely on knowing the type and locations
4011 of stack and global variables, for example exp-Ptrcheck.
4013 Reading of such information is disabled by default, because most
4014 tools don't need it, and because it is expensive in space and time.
4015 However, you can force Valgrind to read it, using the
4016 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4017 make use of such information, if present, to provide source-level
4018 descriptions of data addresses in the error messages they create.
4020 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4021 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).