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 and AMD64/macOS 10.13.
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 * ==================== TOOL CHANGES ====================
35 - The command option --collect-systime has been enhanced to specify
36 the unit used to record the elapsed time spent during system calls.
37 The command option now accepts the values no|yes|msec|usec|nsec,
38 where yes is a synonym of msec. When giving the value nsec, the
39 system cpu time of system calls is also recorded.
45 - Several memcheck options are now dynamically changeable.
46 Use valgrind --help-dyn-options to list them.
48 * ==================== OTHER CHANGES ====================
50 * New and modified GDB server monitor features:
52 - The gdbserver monitor commands that require an address and an optional
53 length argument now accepts the alternate 'C like' syntax "address[length]".
54 For example, the memcheck command "monitor who_points_at 0x12345678 120"
55 can now also be given as "monitor who_points_at 0x12345678[120]".
57 * ==================== FIXED BUGS ====================
59 The following bugs have been fixed or resolved. Note that "n-i-bz"
60 stands for "not in bugzilla" -- that is, a bug that was reported to us
61 but never got a bugzilla entry. We encourage you to file bugs in
62 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
63 than mailing the developers (or mailing lists) directly -- bugs that
64 are not entered into bugzilla tend to get forgotten about or ignored.
66 To see details of a given bug, visit
67 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
68 where XXXXXX is the bug number as listed below.
70 400593 In Coregrind, use statx for some internal syscalls if [f]stat[64] fail
71 404406 s390x: z14 miscellaneous instructions not implemented
72 405201 Incorrect size of struct vki_siginfo on 64-bit Linux architectures
73 406561 mcinfcallWSRU gdbserver_test fails on ppc64
74 406824 Unsupported baseline
75 407218 Add support for the copy_file_range syscall
76 407307 Intercept stpcpy also in ld.so for arm64
77 407764 drd cond_post_wait gets wrong (?) condition on s390x z13 system
78 408009 Expose rdrand and f16c even on avx if host cpu supports them
79 408091 Missing pkey syscalls
80 408414 Add support for missing for preadv2 and pwritev2 syscalls
81 409141 Valgrind hangs when SIGKILLed
82 409367 exit_group() after signal to thread waiting in futex() causes hangs
83 409780 References to non-existent configure.in
84 410599 Non-deterministic behaviour of pth_self_kill_15_other test
85 411134 Allow the user to change a set of command line options during execution
86 412344 Problem setting mips flags with specific paths
87 413330 avx-1 test fails on AMD EPYC 7401P 24-Core Processor
88 413603 callgrind_annotate/cg_annotate truncate function names at '#'
89 414565 Specific use case bug found in SysRes VG_(do_sys_sigprocmask)
91 n-i-bz Fix minor one time leaks in dhat.
92 n-i-bz Add --run-cxx-freeres=no in outer args to avoid inner crashes.
93 n-i-bz Add support for the Linux io_uring system calls
95 Release 3.15.0 (12 April 2019)
96 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
98 3.15.0 is a feature release with many improvements and the usual collection of
101 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
102 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
103 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
104 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
105 support for X86/macOS 10.13 and AMD64/macOS 10.13.
107 * ==================== CORE CHANGES ===================
109 * The XTree Massif output format now makes use of the information obtained
110 when specifying --read-inline-info=yes.
112 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
114 * ==================== TOOL CHANGES ====================
118 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
119 it has been promoted from an experimental tool to a regular tool. Run it
120 with --tool=dhat instead of --tool=exp-dhat.
122 - DHAT now prints only minimal data when the program ends, instead writing
123 the bulk of the profiling data to a file. As a result, the --show-top-n
124 and --sort-by options have been removed.
126 - Profile results can be viewed with the new viewer, dh_view.html. When
127 a run ends, a short message is printed, explaining how to view the result.
129 - See the documentation for more details.
133 - cg_annotate has a new option, --show-percs, which prints percentages next
138 - callgrind_annotate has a new option, --show-percs, which prints percentages
139 next to all event counts.
141 - callgrind_annotate now inserts commas in call counts, and
142 sort the caller/callee lists in the call tree.
146 - The default value for --read-inline-info is now "yes" on
147 Linux/Android/Solaris. It is still "no" on other OS.
151 - The option --xtree-leak=yes (to output leak result in xtree format)
152 automatically activates the option --show-leak-kinds=all, as xtree
153 visualisation tools such as kcachegrind can in any case select what kind
154 of leak to visualise.
156 - There has been further work to avoid false positives. In particular,
157 integer equality on partially defined inputs (C == and !=) is now handled
160 * ==================== OTHER CHANGES ====================
162 * The new option --show-error-list=no|yes displays, at the end of the run, the
163 list of detected errors and the used suppressions. Prior to this change,
164 showing this information could only be done by specifying "-v -v", but that
165 also produced a lot of other possibly-non-useful messages. The option -s is
166 equivalent to --show-error-list=yes.
168 * ==================== FIXED BUGS ====================
170 The following bugs have been fixed or resolved. Note that "n-i-bz"
171 stands for "not in bugzilla" -- that is, a bug that was reported to us
172 but never got a bugzilla entry. We encourage you to file bugs in
173 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
174 than mailing the developers (or mailing lists) directly -- bugs that
175 are not entered into bugzilla tend to get forgotten about or ignored.
177 To see details of a given bug, visit
178 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
179 where XXXXXX is the bug number as listed below.
181 385411 s390x: z13 vector floating-point instructions not implemented
182 397187 z13 vector register support for vgdb gdbserver
183 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
184 398870 Please add support for instruction vcvtps2ph
185 399287 amd64 front end: Illegal Instruction vcmptrueps
186 399301 Use inlined frames in Massif XTree output.
187 399322 Improve callgrind_annotate output
188 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
189 400164 helgrind test encounters mips x-compiler warnings and assembler error
190 400490 s390x: VRs allocated as if separate from FPRs
191 400491 s390x: Operand of LOCH treated as unsigned integer
192 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
193 options, which specify a mips64 processor
194 401112 LLVM 5.0 generates comparison against partially initialized data
195 401277 More bugs in z13 support
196 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
197 401578 drd: crashes sometimes on fork()
198 401627 memcheck errors with glibc avx2 optimized wcsncmp
199 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
200 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
201 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
203 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
204 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
205 402123 invalid assembler opcodes for mips32r2
206 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
207 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
208 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
209 402351 mips64 libvexmultiarch_test fails on s390x
211 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
212 402480 Do not use %rsp in clobber list
213 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
214 402515 Implement new option --show-error-list=no|yes / -s
215 402519 POWER 3.0 addex instruction incorrectly implemented
216 402781 Redo the cache used to process indirect branch targets
217 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
218 403552 s390x: wrong facility bit checked for vector facility
219 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
220 404638 Add VG_(replaceIndexXA)
221 404843 s390x: backtrace sometimes ends prematurely
222 404888 autotools cleanup series
223 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
224 405182 Valgrind fails to build with Clang
225 405205 filter_libc: remove the line holding the futex syscall error entirely
226 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
227 the upper and lower 32-bits of the 64-bit result
228 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
229 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
230 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
232 405403 s390x disassembler cannot be used on x86
233 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
234 405458 MIPS mkFormVEC arguments swapped?
235 405716 drd: Fix an integer overflow in the stack margin calculation
236 405722 Support arm64 core dump
237 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
238 of the 64-bit destination field.
239 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
240 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
241 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
243 406256 PPC64, vector floating point instructions don't handle subnormal
244 according to VSCR[NJ] bit setting.
245 406352 cachegrind/callgrind fails ann tests because of missing a.c
246 406354 dhat is broken on x86 (32bit)
247 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
248 406357 gdbserver_tests fails because of gdb output change
249 406360 memcheck/tests/libstdc++.supp needs more supression variants
250 406422 none/tests/amd64-linux/map_32bits.vgtest fails too easily
251 406465 arm64 insn selector fails on "t0 = <expr>" where <expr> has type Ity_F16
252 407340 PPC64, does not support the vlogefp, vexptefp instructions.
253 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
254 n-i-bz Fix callgrind_annotate non deterministic order for equal total
255 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
256 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
257 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
259 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
260 (3.15.0.RC2: 11 April 2019, git 0c8be9bbede189ec580ec270521811766429595f)
261 (3.15.0: 14 April 2019, git 270037da8b508954f0f7d703a0bebf5364eec548)
265 Release 3.14.0 (9 October 2018)
266 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
268 3.14.0 is a feature release with many improvements and the usual collection of
271 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
272 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
273 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
274 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
275 support for X86/macOS 10.13, AMD64/macOS 10.13.
277 * ==================== CORE CHANGES ===================
279 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
280 debug info for unloaded code. This allows saved stack traces (e.g. for
281 memory leaks) to include file/line info for code that has been dlclose'd (or
282 similar). See the user manual for more information and known limitations.
284 * Ability to specify suppressions based on source file name and line number.
286 * Majorly overhauled register allocator. No end-user changes, but the JIT
287 generates code a bit more quickly now.
289 * ================== PLATFORM CHANGES =================
291 * Preliminary support for macOS 10.13 has been added.
293 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
295 * mips: support for MIPS SIMD architecture (MSA) has been added.
297 * mips: support for MIPS N32 ABI has been added.
299 * s390: partial support for vector instructions (integer and string) has been
302 * ==================== TOOL CHANGES ====================
304 * Helgrind: Addition of a flag
305 --delta-stacktrace=no|yes [yes on linux amd64/x86]
306 which specifies how full history stack traces should be computed.
307 Setting this to =yes can speed up Helgrind by 25% when using
308 --history-level=full.
310 * Memcheck: reduced false positive rate for optimised code created by Clang 6
311 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
312 blocks more carefully to determine where it can avoid expensive definedness
313 checks without loss of precision. This is controlled by the flag
314 --expensive-definedness-checks=no|auto|yes [auto].
316 * ==================== OTHER CHANGES ====================
318 * Valgrind is now buildable with link-time optimisation (LTO). A new
319 configure option --enable-lto=yes allows building Valgrind with LTO. If the
320 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
321 Note that if you are doing Valgrind development, --enable-lto=yes massively
322 slows down the build process.
324 * ==================== FIXED BUGS ====================
326 The following bugs have been fixed or resolved. Note that "n-i-bz"
327 stands for "not in bugzilla" -- that is, a bug that was reported to us
328 but never got a bugzilla entry. We encourage you to file bugs in
329 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
330 than mailing the developers (or mailing lists) directly -- bugs that
331 are not entered into bugzilla tend to get forgotten about or ignored.
333 To see details of a given bug, visit
334 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
335 where XXXXXX is the bug number as listed below.
337 79362 Debug info is lost for .so files when they are dlclose'd
338 208052 strlcpy error when n = 0
339 255603 exp-sgcheck Assertion '!already_present' failed
340 338252 building valgrind with -flto (link time optimisation) fails
341 345763 MIPS N32 ABI support
342 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
343 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
344 372347 Replacement problem of the additional c++14/c++17 new/delete operators
345 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
346 376257 helgrind history full speed up using a cached stack
347 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
349 379748 Fix missing pselect syscall (OS X 10.11)
350 379754 Fix missing syscall ulock_wait (OS X 10.12)
351 380397 s390x: __GI_strcspn() replacemenet needed
352 381162 possible array overrun in VEX register allocator
353 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
354 381274 powerpc too chatty even with --sigill-diagnostics=no
355 381289 epoll_pwait can have a NULL sigmask
356 381553 VEX register allocator v3
357 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
358 381769 Use ucontext_t instead of struct ucontext
359 381805 arm32 needs ld.so index hardwire for new glibc security fixes
360 382256 gz compiler flag test doesn't work for gold
361 382407 vg_perf needs "--terse" command line option
362 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
363 382563 MIPS MSA ASE support
364 382998 xml-socket doesn't work
365 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
366 383723 Fix missing kevent_qos syscall (macOS 10.11)
367 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
368 384096 Mention AddrCheck at Memcheck's command line option [..]
369 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
370 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
371 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
372 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
373 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
374 384337 performance improvements to VEX register allocator v2 and v3
375 384526 reduce number of spill insns generated by VEX register allocator v3
376 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
377 384631 Sanitise client args as printed with -v
378 384633 Add a simple progress-reporting facility
379 384987 VEX regalloc: allocate caller-save registers for short lived vregs
380 385055 PPC VEX temporary storage exhausted
381 385182 PPC64 is missing support for the DSCR
382 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
383 385207 PPC64, generate_store_FPRF() generates too many Iops
384 385208 PPC64, xxperm instruction exhausts temporary memory
385 385210 PPC64, vpermr instruction could exhaust temporary memory
386 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
387 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
388 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
389 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
390 385408 s390x: z13 vector "support" instructions not implemented
391 385409 s390x: z13 vector integer instructions not implemented
392 385410 s390x: z13 vector string instructions not implemented
393 385412 s390x: new non-vector z13 instructions not implemented
394 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
395 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
396 385939 Optionally exit on the first error
397 386318 valgrind.org/info/tools.html is missing SGCheck
398 386425 running valgrind + wine on armv7l gives illegal opcode
399 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
400 387410 MIPSr6 support
401 387664 Memcheck: make expensive-definedness-checks be the default
402 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
403 387766 asm shifts cause false positive "Conditional jump or move depends
404 on uninitialised value"
405 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
406 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
407 388786 Support bpf syscall in amd64 Linux
408 388862 Add replacements for wmemchr and wcsnlen on Linux
409 389065 valgrind meets gcc flag -Wlogical-op
410 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
411 390471 suppression by specification of source-file line number
412 390723 make xtree dump files world wide readable, similar to log files
413 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
414 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
415 392118 unhandled amd64-linux syscall: 332 (statx)
416 392449 callgrind not clearing the number of calls properly
417 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
418 lxssp, stxssp and stxvl instructions.
419 393023 callgrind_control risks using the wrong vgdb
420 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
421 393099 posix_memalign() invalid write if alignment == 0
422 393146 failing assert "is_DebugInfo_active(di)"
423 395709 PPC64 is missing support for the xvnegsp instruction
424 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
426 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
427 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
428 396839 s390x: Trap instructions not implemented
429 396887 arch_prctl should return EINVAL on unknown option
430 == 397286 crash before launching binary (Unsupported arch_prctl option)
431 == 397393 valgrind: the 'impossible' happened: (Archlinux)
432 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
433 396906 compile tests failure on mips32-linux: broken inline asm in tests on
435 397012 glibc ld.so uses arch_prctl on i386
436 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
437 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
438 397424 glibc 2.27 and gdb_server tests
439 398028 Assertion `cfsi_fits` failing in simple C program
440 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
442 n-i-bz Fix missing workq_ops operations (macOS)
443 n-i-bz fix bug in strspn replacement
444 n-i-bz Add support for the Linux BLKFLSBUF ioctl
445 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
446 n-i-bz Fix possible stack trashing by semctl syscall wrapping
447 n-i-bz Add support for the Linux membarrier() system call
448 n-i-bz x86 front end: recognise and handle UD2 correctly
449 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
450 correctly aligned before entering the handler.
452 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
453 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
454 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
458 Release 3.13.0 (15 June 2017)
459 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
461 3.13.0 is a feature release with many improvements and the usual collection of
464 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
465 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
466 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
467 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
469 * ==================== CORE CHANGES ===================
471 * The translation cache size has been increased to keep up with the demands of
472 large applications. The maximum number of sectors has increased from 24 to
473 48. The default number of sectors has increased from 16 to 32 on all
474 targets except Android, where the increase is from 6 to 12.
476 * The amount of memory that Valgrind can use has been increased from 64GB to
477 128GB. In particular this means your application can allocate up to about
478 60GB when running on Memcheck.
480 * Valgrind's default load address has been changed from 0x3800'0000 to
481 0x5800'0000, so as to make it possible to load larger executables. This
482 should make it possible to load executables of size at least 1200MB.
484 * A massive spaceleak caused by reading compressed debuginfo files has been
485 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
488 * The C++ demangler has been updated.
490 * Support for demangling Rust symbols has been added.
492 * A new representation of stack traces, the "XTree", has been added. An XTree
493 is a tree of stacktraces with data associated with the stacktraces. This is
494 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
495 consumption of your program. Reporting is controlled by the new options
496 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
498 A report can also be produced on demand using the gdbserver monitor command
499 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
500 format' and 'massif format. The existing visualisers for these formats (e.g.
501 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
502 analyse these reports.
504 Memcheck can also produce XTree leak reports using the Callgrind file
505 format. For more details, see the user manual.
507 * ================== PLATFORM CHANGES =================
509 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
511 * amd64: fixes for JIT failure problems on long AVX2 code blocks
513 * amd64 and x86: support for CET prefixes has been added
515 * arm32: a few missing ARMv8 instructions have been implemented
517 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
518 Store-Conditional instructions has been added. This is to deal with
519 processor implementations that implement the LL/SC specifications strictly
520 and as a result cause Valgrind to hang in certain situations. The
521 alternative implementation is automatically enabled at startup, as required.
522 You can use the option --sim-hints=fallback-llsc to force-enable it if you
525 * Support for OSX 10.12 has been improved.
527 * On Linux, clone handling has been improved to honour CLONE_VFORK that
528 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
529 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
530 CLONE_VM exact semantics will (still) not work.
532 * The TileGX/Linux port has been removed because it appears to be both unused
535 * ==================== TOOL CHANGES ====================
539 - Memcheck should give fewer false positives when running optimised
540 Clang/LLVM generated code.
542 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
544 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
545 to produce the end of execution leak report in a xtree callgrind format
548 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
549 the leak report in an xtree file.
553 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
555 - For some workloads (typically, for big applications), Massif memory
556 consumption and CPU consumption has decreased significantly.
560 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
562 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
563 for Ada gnat compiled applications.
565 * ==================== OTHER CHANGES ====================
567 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
568 append the inner guest stacktrace to the inner host stacktrace. This helps
569 to investigate the errors reported by the outer, when they are caused by the
570 inner guest program (such as an inner regtest). See README_DEVELOPERS for
573 * To allow fast detection of callgrind files by desktop environments and file
574 managers, the format was extended to have an optional first line that
575 uniquely identifies the format ("# callgrind format"). Callgrind creates
576 this line now, as does the new xtree functionality.
578 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
579 have a new %n format letter that is replaced by a sequence number.
581 * "--version -v" now shows the SVN revision numbers from which Valgrind was
584 * ==================== FIXED BUGS ====================
586 The following bugs have been fixed or resolved. Note that "n-i-bz"
587 stands for "not in bugzilla" -- that is, a bug that was reported to us
588 but never got a bugzilla entry. We encourage you to file bugs in
589 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
590 than mailing the developers (or mailing lists) directly -- bugs that
591 are not entered into bugzilla tend to get forgotten about or ignored.
593 To see details of a given bug, visit
594 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
595 where XXXXXX is the bug number as listed below.
597 162848 --log-file output isn't split when a program forks
598 340777 Illegal instruction on mips (ar71xx)
599 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
600 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
601 to a different stack.
602 344139 x86 stack-seg overrides, needed by the Wine people
603 344524 store conditional of guest applications always fail - observed on
605 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
606 352395 Please provide SVN revision info in --version -v
607 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
608 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
609 INVALID_POSIX_THREADID' failed
610 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
611 barrier implementation
612 358697 valgrind.h: Some code remains even when defining NVALGRIND
613 359202 Add musl libc configure/compile
614 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
615 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
616 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
617 362223 assertion failed when .valgrindrc is a directory instead of a file
618 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
619 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
620 368507 can't malloc chunks larger than about 34GB
621 368529 Android arm target link error, missing atexit and pthread_atfork
622 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
623 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
624 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
625 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
626 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
627 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
628 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
629 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
630 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
631 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
632 370028 Reduce the number of compiler warnings on MIPS platforms
633 370635 arm64 missing syscall getcpu
634 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
635 371227 Clean AArch64 syscall table
636 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
637 371471 Valgrind complains about non legit memory leaks on placement new (C++)
638 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
639 371503 disInstr(arm64): unhandled instruction 0xF89F0000
640 371869 support '%' in symbol Z-encoding
641 371916 execution tree xtree concept
642 372120 c++ demangler demangles symbols which are not c++
643 372185 Support of valgrind on ARMv8 with 32 bit executable
644 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
645 372195 Power PC, xxsel instruction is not always recognized.
646 372504 Hanging on exit_group
647 372600 process loops forever when fatal signals are arriving quickly
648 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
649 373046 Stacks registered by core are never deregistered
650 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
651 373086 Implement additional Xen hypercalls
652 373192 Calling posix_spawn in glibc 2.24 completely broken
653 373488 Support for fanotify API on ARM64 architecture
654 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
655 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
656 373938 const IRExpr arguments for matchIRExpr()
657 374719 some spelling fixes
658 374963 increase valgrind's load address to prevent mmap failure
659 375514 valgrind_get_tls_addr() does not work in case of static TLS
660 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
661 for ML_(find_rx_mapping)()
662 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
663 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
664 == 377159 "vex: the `impossible' happened" still present
665 == 375150 Assertion 'tres.status == VexTransOK' failed
666 == 378068 valgrind crashes on AVX2 function in FFmpeg
667 376142 Segfaults on MIPS Cavium Octeon boards
668 376279 disInstr(arm64): unhandled instruction 0xD50320FF
669 376455 Solaris: unhandled syscall lgrpsys(180)
670 376518 Solaris: unhandled fast trap getlgrp(6)
671 376611 ppc64 and arm64 don't know about prlimit64 syscall
672 376729 PPC64, remove R2 from the clobber list
674 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
675 to be wrongly marked as addressable
676 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
677 PIE enabled by default
678 377376 memcheck/tests/linux/getregset fails with glibc2.24
679 377427 PPC64, lxv instruction failing on odd destination register
680 377478 PPC64: ISA 3.0 setup fixes
681 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
682 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
683 and 2 args for FUTEX_TRYLOCK_PI
684 377717 Fix massive space leak when reading compressed debuginfo sections
685 377891 Update Xen 4.6 domctl wrappers
686 377930 fcntl syscall wrapper is missing flock structure check
687 378524 libvexmultiarch_test regression on s390x and ppc64
688 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
689 378673 Update libiberty demangler
690 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
691 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
692 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
693 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
694 (task_register_dyld_image_infos)
695 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
696 (task_register_dyld_shared_cache_image_info)
697 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
698 379473 MIPS: add support for rdhwr cycle counter register
699 379504 remove TileGX/Linux port
700 379525 Support more x86 nop opcodes
701 379838 disAMode(x86): not an addr!
702 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
703 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
704 379895 clock_gettime does not execute POST syscall wrapper
705 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
706 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
707 380200 xtree generated callgrind files refer to files without directory name
708 380202 Assertion failure for cache line size (cls == 64) on aarch64.
709 380397 s390x: __GI_strcspn() replacement needed
710 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
712 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
713 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
714 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
718 Release 3.12.0 (20 October 2016)
719 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
721 3.12.0 is a feature release with many improvements and the usual
722 collection of bug fixes.
724 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
725 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
726 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
727 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
728 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
729 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
731 * ================== PLATFORM CHANGES =================
733 * POWER: Support for ISA 3.0 has been added
735 * mips: support for O32 FPXX ABI has been added.
736 * mips: improved recognition of different processors
737 * mips: determination of page size now done at run time
739 * amd64: Partial support for AMD FMA4 instructions.
741 * arm, arm64: Support for v8 crypto and CRC instructions.
743 * Improvements and robustification of the Solaris port.
745 * Preliminary support for MacOS 10.12 (Sierra) has been added.
747 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
748 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
749 where possible. Valgrind's support for 32-bit x86 has stagnated in
750 recent years and has fallen far behind that for 64-bit x86
751 instructions. By contrast 64-bit x86 is well supported, up to and
754 * ==================== TOOL CHANGES ====================
758 - Added meta mempool support for describing a custom allocator which:
759 - Auto-frees all chunks assuming that destroying a pool destroys all
761 - Uses itself to allocate other memory blocks
763 - New flag --ignore-range-below-sp to ignore memory accesses below
764 the stack pointer, if you really have to. The related flag
765 --workaround-gcc296-bugs=yes is now deprecated. Use
766 --ignore-range-below-sp=1024-1 as a replacement.
770 - Improved thread startup time significantly on non-Linux platforms.
774 - Added collection of the metric "tot-blocks-allocd"
776 * ==================== OTHER CHANGES ====================
778 * Replacement/wrapping of malloc/new related functions is now done not just
779 for system libraries by default, but for any globally defined malloc/new
780 related function (both in shared libraries and statically linked alternative
781 malloc implementations). The dynamic (runtime) linker is excluded, though.
782 To only intercept malloc/new related functions in
783 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
784 "nouserintercepts" can be any non-existing library name).
785 This new functionality is not implemented for MacOS X.
787 * The maximum number of callers in a suppression entry is now equal to
788 the maximum size for --num-callers (500).
789 Note that --gen-suppressions=yes|all similarly generates suppressions
790 containing up to --num-callers frames.
792 * New and modified GDB server monitor features:
794 - Valgrind's gdbserver now accepts the command 'catch syscall'.
795 Note that you must have GDB >= 7.11 to use 'catch syscall' with
798 * New option --run-cxx-freeres=<yes|no> can be used to change whether
799 __gnu_cxx::__freeres() cleanup function is called or not. Default is
802 * Valgrind is able to read compressed debuginfo sections in two formats:
803 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
804 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
806 * Modest JIT-cost improvements: the cost of instrumenting code blocks
807 for the most common use case (x86_64-linux, Memcheck) has been
810 * Improved performance for programs that do a lot of discarding of
811 instruction address ranges of 8KB or less.
813 * The C++ symbol demangler has been updated.
815 * More robustness against invalid syscall parameters on Linux.
817 * ==================== FIXED BUGS ====================
819 The following bugs have been fixed or resolved. Note that "n-i-bz"
820 stands for "not in bugzilla" -- that is, a bug that was reported to us
821 but never got a bugzilla entry. We encourage you to file bugs in
822 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
823 than mailing the developers (or mailing lists) directly -- bugs that
824 are not entered into bugzilla tend to get forgotten about or ignored.
826 To see details of a given bug, visit
827 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
828 where XXXXXX is the bug number as listed below.
830 191069 Exiting due to signal not reported in XML output
831 199468 Suppressions: stack size limited to 25
832 while --num-callers allows more frames
833 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
834 278744 cvtps2pd with redundant RexW
835 303877 valgrind doesn't support compressed debuginfo sections.
836 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
837 348345 Assertion fails for negative lineno
838 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
839 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
840 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
841 351804 Crash on generating suppressions for "printf" call on OS X 10.10
842 352197 mips: mmap2() not wrapped correctly for page size > 4096
843 353083 arm64 doesn't implement various xattr system calls
844 353084 arm64 doesn't support sigpending system call
845 353137 www: update info for Supported Platforms
846 353138 www: update "The Valgrind Developers" page
847 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
850 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
851 353398 WARNING: unhandled amd64-solaris syscall: 207
852 353660 XML in auxwhat tag not escaping reserved symbols properly
853 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
854 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
855 353802 ELF debug info reader confused with multiple .rodata sections
856 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
857 353917 unhandled amd64-solaris syscall fchdir(120)
858 353920 unhandled amd64-solaris syscall: 170
859 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
860 354392 unhandled amd64-solaris syscall: 171
861 354797 Vbit test does not include Iops for Power 8 instruction support
862 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
866 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
867 355188 valgrind should intercept all malloc related global functions
868 355454 do not intercept malloc related symbols from the runtime linker
869 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
870 356044 Dwarf line info reader misinterprets is_stmt register
871 356112 mips: replace addi with addiu
872 356393 valgrind (vex) crashes because isZeroU happened
875 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
876 356678 arm64-linux: unhandled syscall 232 (mincore)
877 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
878 356823 Unsupported ARM instruction: stlex
879 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
880 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
881 357673 crash if I try to run valgrind with a binary link with libcurl
882 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
883 357871 pthread_spin_destroy not properly wrapped
884 357887 Calls to VG_(fclose) do not close the file descriptor
885 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
886 358030 support direct socket calls on x86 32bit (new in linux 4.3)
887 358478 drd/tests/std_thread.cpp doesn't build with GCC6
888 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
889 359181 Buffer Overflow during Demangling
890 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
891 359289 s390x: popcnt (B9E1) not implemented
892 359472 The Power PC vsubuqm instruction doesn't always give the correct result
893 359503 Add missing syscalls for aarch64 (arm64)
894 359645 "You need libc6-dbg" help message could be more helpful
895 359703 s390: wire up separate socketcalls system calls
896 359724 getsockname might crash - deref_UInt should call safe_to_deref
897 359733 amd64 implement ld.so strchr/index override like x86
898 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
899 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
901 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
902 359871 Incorrect mask handling in ppoll
903 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
904 360008 Contents of Power vr registers contents is not printed correctly when
905 the --vgdb-shadow-registers=yes option is used
906 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
908 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
909 360425 arm64 unsupported instruction ldpsw
911 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
912 360571 Error about the Android Runtime reading below the stack pointer on ARM
913 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
914 360749 kludge for multiple .rodata sections on Solaris no longer needed
915 360752 raise the number of reserved fds in m_main.c from 10 to 12
916 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
917 361226 s390x: risbgn (EC59) not implemented
918 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
919 361354 ppc64[le]: wire up separate socketcalls system calls
920 361615 Inconsistent termination for multithreaded process terminated by signal
921 361926 Unhandled Solaris syscall: sysfs(84)
922 362009 V dumps core on unimplemented functionality before threads are created
923 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
924 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
925 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
926 362953 Request for an update to the Valgrind Developers page
927 363680 add renameat2() support
928 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
929 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
930 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
931 364058 clarify in manual limitations of array overruns detections
932 364413 pselect sycallwrapper mishandles NULL sigmask
933 364728 Power PC, missing support for several HW registers in
934 get_otrack_shadow_offset_wrk()
935 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
936 365273 Invalid write to stack location reported after signal handler runs
937 365912 ppc64BE segfault during jm-insns test (RELRO)
938 366079 FPXX Support for MIPS32 Valgrind
939 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
940 366344 Multiple unhandled instruction for Aarch64
941 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
942 367995 Integration of memcheck with custom memory allocator
943 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
944 368412 False positive result for altivec capability check
945 368416 Add tc06_two_races_xml.exp output for ppc64
946 368419 Perf Events ioctls not implemented
947 368461 mmapunmap test fails on ppc64
948 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
949 369000 AMD64 fma4 instructions unsupported.
950 369169 ppc64 fails jm_int_isa_2_07 test
951 369175 jm_vec_isa_2_07 test crashes on ppc64
952 369209 valgrind loops and eats up all memory if cwd doesn't exist.
953 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
954 369359 msghdr_foreachfield can crash when handling bad iovec
955 369360 Bad sigprocmask old or new sets can crash valgrind
956 369361 vmsplice syscall wrapper crashes on bad iovec
957 369362 Bad sigaction arguments crash valgrind
958 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
959 369402 Bad set/get_thread_area pointer crashes valgrind
960 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
961 369446 valgrind crashes on unknown fcntl command
962 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
963 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
964 370265 ISA 3.0 HW cap stuff needs updating
965 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
966 372195 Power PC, xxsel instruction is not always recognized
968 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
969 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
970 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
971 n-i-bz Fix ppoll_alarm exclusion on OS X
972 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
973 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
974 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
975 n-i-bz amd64: memcheck false positive with shr %edx
976 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
977 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
978 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
979 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
981 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
982 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
983 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
987 Release 3.11.0 (22 September 2015)
988 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
990 3.11.0 is a feature release with many improvements and the usual
991 collection of bug fixes.
993 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
994 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
995 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
996 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
997 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
998 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
1000 * ================== PLATFORM CHANGES =================
1002 * Support for Solaris/x86 and Solaris/amd64 has been added.
1004 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
1006 * Preliminary support for the Tilera TileGX architecture has been added.
1008 * s390x: It is now required for the host to have the "long displacement"
1009 facility. The oldest supported machine model is z990.
1011 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
1012 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
1014 * The JIT's register allocator is significantly faster, making the JIT
1015 as a whole somewhat faster, so JIT-intensive activities, for example
1016 program startup, are modestly faster, around 5%.
1018 * There have been changes to the default settings of several command
1019 line flags, as detailed below.
1021 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
1022 capable hosts, the simulated CPUID will now indicate AVX2 support.
1024 * ==================== TOOL CHANGES ====================
1028 - The default value for --leak-check-heuristics has been changed from
1029 "none" to "all". This helps to reduce the number of possibly
1030 lost blocks, in particular for C++ applications.
1032 - The default value for --keep-stacktraces has been changed from
1033 "malloc-then-free" to "malloc-and-free". This has a small cost in
1034 memory (one word per malloc-ed block) but allows Memcheck to show the
1035 3 stacktraces of a dangling reference: where the block was allocated,
1036 where it was freed, and where it is acccessed after being freed.
1038 - The default value for --partial-loads-ok has been changed from "no" to
1039 "yes", so as to avoid false positive errors resulting from some kinds
1040 of vectorised loops.
1042 - A new monitor command 'xb <addr> <len>' shows the validity bits of
1043 <len> bytes at <addr>. The monitor command 'xb' is easier to use
1044 than get_vbits when you need to associate byte data value with
1045 their corresponding validity bits.
1047 - The 'block_list' monitor command has been enhanced:
1048 o it can print a range of loss records
1049 o it now accepts an optional argument 'limited <max_blocks>'
1050 to control the number of blocks printed.
1051 o if a block has been found using a heuristic, then
1052 'block_list' now shows the heuristic after the block size.
1053 o the loss records/blocks to print can be limited to the blocks
1054 found via specified heuristics.
1056 - The C helper functions used to instrument loads on
1057 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
1058 replaced by handwritten assembly sequences. This gives speedups
1059 in the region of 0% to 7% for those targets only.
1061 - A new command line option, --expensive-definedness-checks=yes|no,
1062 has been added. This is useful for avoiding occasional invalid
1063 uninitialised-value errors in optimised code. Watch out for
1064 runtime degradation, as this can be up to 25%. As always, though,
1065 the slowdown is highly application specific. The default setting
1070 - A new monitor command 'all_snapshots <filename>' dumps all
1071 snapshots taken so far.
1075 - Significant memory reduction and moderate speedups for
1076 --history-level=full for applications accessing a lot of memory
1077 with many different stacktraces.
1079 - The default value for --conflict-cache-size=N has been doubled to
1080 2000000. Users that were not using the default value should
1081 preferably also double the value they give.
1083 The default was changed due to the changes in the "full history"
1084 implementation. Doubling the value gives on average a slightly more
1085 complete history and uses similar memory (or significantly less memory
1086 in the worst case) than the previous implementation.
1088 - The Helgrind monitor command 'info locks' now accepts an optional
1089 argument 'lock_addr', which shows information about the lock at the
1092 - When using --history-level=full, the new Helgrind monitor command
1093 'accesshistory <addr> [<len>]' will show the recorded accesses for
1094 <len> (or 1) bytes at <addr>.
1096 * ==================== OTHER CHANGES ====================
1098 * The default value for the --smc-check option has been changed from
1099 "stack" to "all-non-file" on targets that provide automatic D-I
1100 cache coherence (x86, amd64 and s390x). The result is to provide,
1101 by default, transparent support for JIT generated and self-modifying
1102 code on all targets.
1104 * Mac OS X only: the default value for the --dsymutil option has been
1105 changed from "no" to "yes", since any serious usage on Mac OS X
1106 always required it to be "yes".
1108 * The command line options --db-attach and --db-command have been removed.
1109 They were deprecated in 3.10.0.
1111 * When a process dies due to a signal, Valgrind now shows the signal
1112 and the stacktrace at default verbosity (i.e. verbosity 1).
1114 * The address description logic used by Memcheck and Helgrind now
1115 describes addresses in anonymous segments, file mmap-ed segments,
1116 shared memory segments and the brk data segment.
1118 * The new option --error-markers=<begin>,<end> can be used to mark the
1119 begin/end of errors in textual output mode, to facilitate
1120 searching/extracting errors in output files that mix valgrind errors
1121 with program output.
1123 * The new option --max-threads=<number> can be used to change the number
1124 of threads valgrind can handle. The default is 500 threads which
1125 should be more than enough for most applications.
1127 * The new option --valgrind-stacksize=<number> can be used to change the
1128 size of the private thread stacks used by Valgrind. This is useful
1129 for reducing memory use or increasing the stack size if Valgrind
1130 segfaults due to stack overflow.
1132 * The new option --avg-transtab-entry-size=<number> can be used to specify
1133 the expected instrumented block size, either to reduce memory use or
1134 to avoid excessive retranslation.
1136 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1138 * New and modified GDB server monitor features:
1140 - When a signal is reported in GDB, you can now use the GDB convenience
1141 variable $_siginfo to examine detailed signal information.
1143 - Valgrind's gdbserver now allows the user to change the signal
1144 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1145 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1146 continue without passing the signal to the process.
1148 - With GDB >= 7.10, the command 'target remote'
1149 will automatically load the executable file of the process running
1150 under Valgrind. This means you do not need to specify the executable
1151 file yourself, GDB will discover it itself. See GDB documentation about
1152 'qXfer:exec-file:read' packet for more info.
1154 * ==================== FIXED BUGS ====================
1156 The following bugs have been fixed or resolved. Note that "n-i-bz"
1157 stands for "not in bugzilla" -- that is, a bug that was reported to us
1158 but never got a bugzilla entry. We encourage you to file bugs in
1159 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1160 than mailing the developers (or mailing lists) directly -- bugs that
1161 are not entered into bugzilla tend to get forgotten about or ignored.
1163 To see details of a given bug, visit
1164 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1165 where XXXXXX is the bug number as listed below.
1167 116002 VG_(printf): Problems with justification of strings and integers
1168 155125 avoid cutting away file:lineno after long function name
1169 197259 Unsupported arch_prtctl PR_SET_GS option
1170 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
1171 201216 Fix Valgrind does not support pthread_sigmask() on OS X
1172 201435 Fix Darwin: -v does not show kernel version
1173 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
1174 211256 Fixed an outdated comment regarding the default platform.
1175 211529 Incomplete call stacks for code compiled by newer versions of MSVC
1176 211926 Avoid compilation warnings in valgrind.h with -pedantic
1177 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
1179 226609 Crediting upstream authors in man page
1180 231257 Valgrind omits path when executing script from shebang line
1181 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
1182 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
1183 269360 s390x: Fix addressing mode selection for compare-and-swap
1184 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
1186 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
1187 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
1188 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
1189 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
1190 330147 libmpiwrap PMPI_Get_count returns undefined value
1191 333051 mmap of huge pages fails due to incorrect alignment
1193 334802 valgrind does not always explain why a given option is bad
1194 335618 mov.w rN, pc/sp (ARM32)
1195 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
1200 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
1201 338602 AVX2 bit in CPUID missing
1202 338606 Strange message for scripts with invalid interpreter
1203 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
1204 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
1205 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
1207 339156 gdbsrv not called for fatal signal
1208 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
1209 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
1210 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
1211 339442 Fix testsuite build failure on OS X 10.9
1212 339542 Enable compilation with Intel's ICC compiler
1213 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
1214 339688 Mac-specific ASM does not support .version directive (cpuid,
1215 tronical and pushfpopf tests)
1216 339745 Valgrind crash when check Marmalade app (partial fix)
1217 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
1218 339778 Linux/TileGx platform support for Valgrind
1219 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
1220 339789 Fix none/tests/execve test on Mac OS X 10.9
1221 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
1222 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
1223 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
1224 340392 Allow user to select more accurate definedness checking in memcheck
1225 to avoid invalid complaints on optimised code
1226 340430 Fix some grammatical weirdness in the manual.
1227 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
1228 341419 Signal handler ucontext_t not filled out correctly on OS X
1229 341539 VG_(describe_addr) should not describe address as belonging to client
1230 segment if it is past the heap end
1231 341613 Enable building of manythreads and thread-exits tests on Mac OS X
1232 341615 Fix none/tests/darwin/access_extended test on Mac OS X
1233 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
1234 341789 aarch64: shmat fails with valgrind on ARMv8
1235 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
1236 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
1237 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
1238 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
1239 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
1240 342221 socket connect false positive uninit memory for unknown af family
1241 342353 Allow dumping full massif output while valgrind is still running
1242 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
1246 342603 Add I2C_SMBUS ioctl support
1247 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
1248 342683 Mark memory past the initial brk limit as unaddressable
1249 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
1250 342795 Internal glibc __GI_mempcpy call should be intercepted
1251 342841 s390x: Support instructions fiebr(a) and fidbr(a)
1252 343012 Unhandled syscall 319 (memfd_create)
1253 343069 Patch updating v4l2 API support
1254 343173 helgrind crash during stack unwind
1255 343219 fix GET_STARTREGS for arm
1256 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
1257 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
1258 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
1259 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
1260 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
1261 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
1262 343597 ppc64le: incorrect use of offseof macro
1263 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
1264 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
1265 343732 Unhandled syscall 144 (setgid) on aarch64
1266 343733 Unhandled syscall 187 (msgctl and related) on aarch64
1267 343802 s390x: False positive "conditional jump or move depends on [..]
1268 343902 --vgdb=yes doesn't break when --xml=yes is used
1269 343967 Don't warn about setuid/setgid/setcap executable for directories
1270 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
1271 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
1272 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
1273 344054 www - update info for Solaris/illumos
1274 344416 'make regtest' does not work cleanly on OS X
1275 344235 Remove duplicate include of pub_core_aspacemgr.h
1276 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
1277 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
1278 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
1279 344314 callgrind_annotate ... warnings about commands containing newlines
1280 344318 socketcall should wrap recvmmsg and sendmmsg
1281 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
1282 344416 Fix 'make regtest' does not work cleanly on OS X
1283 344499 Fix compilation for Linux kernel >= 4.0.0
1284 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
1285 unix:349 (__pthread_fchdir)
1286 344559 Garbage collection of unused segment names in address space manager
1287 344560 Fix stack traces missing penultimate frame on OS X
1288 344621 Fix memcheck/tests/err_disable4 test on OS X
1289 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
1290 344702 Fix missing libobjc suppressions on OS X 10.10
1292 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
1293 344939 Fix memcheck/tests/xml1 on OS X 10.10
1294 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
1295 345079 Fix build problems in VEX/useful/test_main.c
1296 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
1297 345177 arm64: prfm (reg) not implemented
1298 345215 Performance improvements for the register allocator
1299 345248 add support for Solaris OS in valgrind
1300 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
1301 345394 Fix memcheck/tests/strchr on OS X
1302 345637 Fix memcheck/tests/sendmsg on OS X
1303 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
1304 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
1305 345887 Fix an assertion in the address space manager
1306 345928 amd64: callstack only contains current function for small stacks
1307 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1308 345987 MIPS64: Implement cavium LHX instruction
1309 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1310 346185 Fix typo saving altivec register v24
1311 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1312 and LibVEX_GuestPPC64_get_CR()
1313 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1314 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1315 346307 fuse filesystem syscall deadlocks
1316 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1317 346411 MIPS: SysRes::_valEx handling is incorrect
1318 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1319 346474 PPC64 Power 8, spr TEXASRU register not supported
1320 346487 Compiler generates "note" about a future ABI change for PPC64
1321 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1322 and causing spurious "invalid read of size 8" warnings
1323 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1324 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1325 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1326 347322 Power PC regression test cleanup
1327 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1329 347389 unhandled syscall: 373 (Linux ARM syncfs)
1330 347686 Patch set to cleanup PPC64 regtests
1331 347978 Remove bash dependencies where not needed
1332 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1333 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1335 348102 Patch updating v4l2 API support
1336 348247 amd64 front end: jno jumps wrongly when overflow is not set
1337 348269 Improve mmap MAP_HUGETLB support.
1338 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1339 348345 Assertion fails for negative lineno
1340 348377 Unsupported ARM instruction: yield
1341 348565 Fix detection of command line option availability for clang
1342 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1343 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1344 348748 Fix redundant condition
1345 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1346 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1347 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1348 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1349 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1350 349626 Implemented additional Xen hypercalls
1351 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1352 349790 Clean up of the hardware capability checking utilities.
1353 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1354 349874 Fix typos in source code
1355 349879 memcheck: add handwritten assembly for helperc_LOADV*
1356 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1357 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1358 350202 Add limited param to 'monitor block_list'
1359 350290 s390x: Support instructions fixbr(a)
1360 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1361 350809 Fix none/tests/async-sigs for Solaris
1362 350811 Remove reference to --db-attach which has been removed.
1363 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1364 350854 hard-to-understand code in VG_(load_ELF)()
1365 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1366 351386 Solaris: Cannot run ld.so.1 under Valgrind
1367 351474 Fix VG_(iseqsigset) as obvious
1368 351531 Typo in /include/vki/vki-xen-physdev.h header guard
1369 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
1370 351858 ldsoexec support on Solaris
1371 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
1372 352130 helgrind reports false races for printfs using mempcpy on FILE* state
1373 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
1374 352320 arm64 crash on none/tests/nestedfs
1375 352765 Vbit test fails on Power 6
1376 352768 The mbar instruction is missing from the Power PC support
1377 352769 Power PC program priority register (PPR) is not supported
1378 n-i-bz Provide implementations of certain compiler builtins to support
1379 compilers that may not provide those
1380 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
1381 n-i-bz Fix compilation on distros with glibc < 2.5
1382 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
1383 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
1384 n-i-bz mremap did not work properly on shared memory
1385 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
1386 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
1388 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
1389 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
1390 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
1394 Release 3.10.1 (25 November 2014)
1395 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1396 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
1397 and backports fixes for all reported missing AArch64 ARMv8 instructions
1398 and syscalls from the trunk. If you package or deliver 3.10.0 for others
1399 to use, you might want to consider upgrading to 3.10.1 instead.
1401 The following bugs have been fixed or resolved. Note that "n-i-bz"
1402 stands for "not in bugzilla" -- that is, a bug that was reported to us
1403 but never got a bugzilla entry. We encourage you to file bugs in
1404 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1405 than mailing the developers (or mailing lists) directly -- bugs that
1406 are not entered into bugzilla tend to get forgotten about or ignored.
1408 To see details of a given bug, visit
1409 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1410 where XXXXXX is the bug number as listed below.
1412 335440 arm64: ld1 (single structure) is not implemented
1413 335713 arm64: unhanded instruction: prfm (immediate)
1414 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
1415 339182 ppc64: AvSplat ought to load destination vector register with [..]
1416 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
1417 339433 ppc64 lxvw4x instruction uses four 32-byte loads
1418 339645 Use correct tag names in sys_getdents/64 wrappers
1419 339706 Fix false positive for ioctl(TIOCSIG) on linux
1420 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
1421 339853 arm64 times syscall unknown
1422 339855 arm64 unhandled getsid/setsid syscalls
1423 339858 arm64 dmb sy not implemented
1424 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
1425 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
1426 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
1428 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
1429 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
1430 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
1431 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
1432 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
1433 340509 arm64: unhandled instruction fcvtas
1434 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
1435 340632 arm64: unhandled instruction fcvtas
1436 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
1437 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
1438 340788 warning: unhandled syscall: 318 (getrandom)
1439 340807 disInstr(arm): unhandled instruction: 0xEE989B20
1440 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
1441 340922 arm64: unhandled getgroups/setgroups syscalls
1442 350251 Fix typo in VEX utility program (test_main.c).
1443 350407 arm64: unhandled instruction ucvtf (vector, integer)
1444 350809 none/tests/async-sigs breaks when run under cron on Solaris
1445 350811 update README.solaris after r15445
1446 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
1447 350854 strange code in VG_(load_ELF)()
1448 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1449 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
1450 n-i-bz Add missing ]] to terminate CDATA.
1451 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
1452 n-i-bz Enable sys_fadvise64_64 on arm32.
1453 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
1454 n-i-bz Add test cases for all known arm64 load/store instructions.
1455 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
1456 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
1457 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
1458 n-i-bz arm64: Implement "fcvtpu w, s".
1459 n-i-bz arm64: implement ADDP and various others
1460 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
1461 n-i-bz arm64: enable FCVT{A,N}S X,S.
1463 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
1467 Release 3.10.0 (10 September 2014)
1468 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1470 3.10.0 is a feature release with many improvements and the usual
1471 collection of bug fixes.
1473 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1474 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1475 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
1476 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
1477 significantly improved relative to the 3.9.0 release.
1479 * ================== PLATFORM CHANGES =================
1481 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
1482 is mostly complete, and is usable, but some SIMD instructions are as
1485 * Support for little-endian variant of the 64-bit POWER architecture.
1487 * Support for Android on MIPS32.
1489 * Support for 64bit FPU on MIPS32 platforms.
1491 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
1493 * Configuration for and running on Android targets has changed.
1494 See README.android in the source tree for details.
1496 * ================== DEPRECATED FEATURES =================
1498 * --db-attach is now deprecated and will be removed in the next
1499 valgrind feature release. The built-in GDB server capabilities are
1500 superior and should be used instead. Learn more here:
1501 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
1503 * ==================== TOOL CHANGES ====================
1507 - Client code can now selectively disable and re-enable reporting of
1508 invalid address errors in specific ranges using the new client
1509 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
1510 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
1512 - Leak checker: there is a new leak check heuristic called
1513 "length64". This is used to detect interior pointers pointing 8
1514 bytes inside a block, on the assumption that the first 8 bytes
1515 holds the value "block size - 8". This is used by
1516 sqlite3MemMalloc, for example.
1518 - Checking of system call parameters: if a syscall parameter
1519 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
1520 several fields not initialised, an error is now reported for each
1521 field. Previously, an error was reported only for the first
1522 uninitialised field.
1524 - Mismatched alloc/free checking: a new flag
1525 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
1526 such checks if necessary.
1530 - Improvements to error messages:
1532 o Race condition error message involving heap allocated blocks also
1533 show the thread number that allocated the raced-on block.
1535 o All locks referenced by an error message are now announced.
1536 Previously, some error messages only showed the lock addresses.
1538 o The message indicating where a lock was first observed now also
1539 describes the address/location of the lock.
1541 - Helgrind now understands the Ada task termination rules and
1542 creates a happens-before relationship between a terminated task
1543 and its master. This avoids some false positives and avoids a big
1544 memory leak when a lot of Ada tasks are created and terminated.
1545 The interceptions are only activated with forthcoming releases of
1546 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
1548 - A new GDB server monitor command "info locks" giving the list of
1549 locks, their location, and their status.
1553 - callgrind_control now supports the --vgdb-prefix argument,
1554 which is needed if valgrind was started with this same argument.
1556 * ==================== OTHER CHANGES ====================
1558 * Unwinding through inlined function calls. Stack unwinding can now
1559 make use of Dwarf3 inlined-unwind information if it is available.
1560 The practical effect is that inlined calls become visible in stack
1561 traces. The suppression matching machinery has been adjusted
1562 accordingly. This is controlled by the new option
1563 --read-inline-info=yes|no. Currently this is enabled by default
1564 only on Linux and Android targets and only for the tools Memcheck,
1567 * Valgrind can now read EXIDX unwind information on 32-bit ARM
1568 targets. If an object contains both CFI and EXIDX unwind
1569 information, Valgrind will prefer the CFI over the EXIDX. This
1570 facilitates unwinding through system libraries on arm-android
1573 * Address description logic has been improved and is now common
1574 between Memcheck and Helgrind, resulting in better address
1575 descriptions for some kinds of error messages.
1577 * Error messages about dubious arguments (eg, to malloc or calloc) are
1578 output like other errors. This means that they can be suppressed
1579 and they have a stack trace.
1581 * The C++ demangler has been updated for better C++11 support.
1583 * New and modified GDB server monitor features:
1585 - Thread local variables/storage (__thread) can now be displayed.
1587 - The GDB server monitor command "v.info location <address>"
1588 displays information about an address. The information produced
1589 depends on the tool and on the options given to valgrind.
1590 Possibly, the following are described: global variables, local
1591 (stack) variables, allocated or freed blocks, ...
1593 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
1594 ask the GDB server to stop at the start of program execution, at
1595 the end of the program execution and on Valgrind internal errors.
1597 - A new monitor command "v.info stats" shows various Valgrind core
1598 and tool statistics.
1600 - A new monitor command "v.set hostvisibility" allows the GDB server
1601 to provide access to Valgrind internal host status/memory.
1603 * A new option "--aspace-minaddr=<address>" can in some situations
1604 allow the use of more memory by decreasing the address above which
1605 Valgrind maps memory. It can also be used to solve address
1606 conflicts with system libraries by increasing the default value.
1607 See user manual for details.
1609 * The amount of memory used by Valgrind to store debug info (unwind
1610 info, line number information and symbol data) has been
1611 significantly reduced, even though Valgrind now reads more
1612 information in order to support unwinding of inlined function calls.
1614 * Dwarf3 handling with --read-var-info=yes has been improved:
1616 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
1618 - Code compiled with
1619 -ffunction-sections -fdata-sections -Wl,--gc-sections
1620 no longer causes assertion failures.
1622 * Improved checking for the --sim-hints= and --kernel-variant=
1623 options. Unknown strings are now detected and reported to the user
1626 * The semantics of stack start/end boundaries in the valgrind.h
1627 VALGRIND_STACK_REGISTER client request has been clarified and
1628 documented. The convention is that start and end are respectively
1629 the lowest and highest addressable bytes of the stack.
1631 * ==================== FIXED BUGS ====================
1633 The following bugs have been fixed or resolved. Note that "n-i-bz"
1634 stands for "not in bugzilla" -- that is, a bug that was reported to us
1635 but never got a bugzilla entry. We encourage you to file bugs in
1636 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1637 than mailing the developers (or mailing lists) directly -- bugs that
1638 are not entered into bugzilla tend to get forgotten about or ignored.
1640 To see details of a given bug, visit
1641 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1642 where XXXXXX is the bug number as listed below.
1644 175819 Support for ipv6 socket reporting with --track-fds
1645 232510 make distcheck fails
1646 249435 Analyzing wine programs with callgrind triggers a crash
1647 278972 support for inlined function calls in stacktraces and suppression
1649 291310 FXSAVE instruction marks memory as undefined on amd64
1650 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
1651 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
1652 315199 vgcore file for threaded app does not show which thread crashed
1653 315952 tun/tap ioctls are not supported
1654 323178 Unhandled instruction: PLDW register (ARM)
1655 323179 Unhandled instruction: PLDW immediate (ARM)
1656 324050 Helgrind: SEGV because of unaligned stack when using movdqa
1657 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
1658 325124 [MIPSEL] Compilation error
1659 325477 Phase 4 support for IBM Power ISA 2.07
1660 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
1661 325628 Phase 5 support for IBM Power ISA 2.07
1662 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
1663 325751 Missing the two privileged Power PC Transactional Memory Instructions
1664 325816 Phase 6 support for IBM Power ISA 2.07
1665 325856 Make SGCheck fail gracefully on unsupported platforms
1666 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
1667 326436 DRD: False positive in libstdc++ std::list::push_back
1668 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
1669 326462 Refactor vgdb to isolate invoker stuff into separate module
1670 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
1671 326623 DRD: false positive conflict report in a field assignment
1672 326724 Valgrind does not compile on OSX 1.9 Mavericks
1673 326816 Intercept for __strncpy_sse2_unaligned missing?
1674 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
1675 326983 Clear direction flag after tests on amd64.
1676 327212 Do not prepend the current directory to absolute path names.
1677 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
1678 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
1679 327284 s390x: Fix translation of the risbg instruction
1680 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
1681 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
1682 327916 DW_TAG_typedef may have no name
1683 327943 s390x: add a redirection for the 'index' function
1684 328100 XABORT not implemented
1685 328205 Implement additional Xen hypercalls
1686 328454 add support Backtraces with ARM unwind tables (EXIDX)
1687 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
1688 328711 valgrind.1 manpage "memcheck options" section is badly generated
1689 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
1690 329612 Incorrect handling of AT_BASE for image execution
1691 329694 clang warns about using uninitialized variable
1692 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
1693 330228 mmap must align to VKI_SHMLBA on mips32
1694 330257 LLVM does not support `-mno-dynamic-no-pic` option
1695 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
1696 330459 --track-fds=yes doesn't track eventfds
1697 330469 Add clock_adjtime syscall support
1698 330594 Missing sysalls on PowerPC / uClibc
1699 330622 Add test to regression suite for POWER instruction: dcbzl
1700 330939 Support for AMD's syscall instruction on x86
1702 330941 Typo in PRE(poll) syscall wrapper
1703 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
1704 331254 Fix expected output for memcheck/tests/dw4
1705 331255 Fix race condition in test none/tests/coolo_sigaction
1706 331257 Fix type of jump buffer in test none/tests/faultstatus
1707 331305 configure uses bash specific syntax
1708 331337 s390x WARNING: unhandled syscall: 326 (dup3)
1709 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
1710 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
1711 331829 Unexpected ioctl opcode sign extension
1712 331830 ppc64: WARNING: unhandled syscall: 96/97
1713 331839 drd/tests/sem_open specifies invalid semaphore name
1714 331847 outcome of drd/tests/thread_name is nondeterministic
1715 332037 Valgrind cannot handle Thumb "add pc, reg"
1716 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
1717 consistency checks enabled
1718 332263 intercepts for pthread_rwlock_timedrdlock and
1719 pthread_rwlock_timedwrlock are incorrect
1720 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
1722 332276 Implement additional Xen hypercalls
1723 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
1724 332765 Fix ms_print to create temporary files in a proper directory
1725 333072 drd: Add semaphore annotations
1726 333145 Tests for missaligned PC+#imm access for arm
1727 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
1728 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
1729 333248 WARNING: unhandled syscall: unix:443
1730 333428 ldr.w pc [rD, #imm] instruction leads to assertion
1731 333501 cachegrind: assertion: Cache set count is not a power of two.
1734 333666 Recognize MPX instructions and bnd prefix.
1735 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
1736 333817 Valgrind reports the memory areas written to by the SG_IO
1738 334049 lzcnt fails silently (x86_32)
1739 334384 Valgrind does not have support Little Endian support for
1741 334585 recvmmsg unhandled (+patch) (arm)
1742 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
1743 334727 Build fails with -Werror=format-security
1744 334788 clarify doc about --log-file initial program directory
1745 334834 PPC64 Little Endian support, patch 2
1746 334836 PPC64 Little Endian support, patch 3 testcase fixes
1747 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
1748 335034 Unhandled ioctl: HCIGETDEVLIST
1749 335155 vgdb, fix error print statement.
1750 335262 arm64: movi 8bit version is not supported
1751 335263 arm64: dmb instruction is not implemented
1752 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
1753 335496 arm64: sbc/abc instructions are not implemented
1754 335554 arm64: unhandled instruction: abs
1755 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
1756 335735 arm64: unhandled instruction: cnt
1757 335736 arm64: unhandled instruction: uaddlv
1758 335848 arm64: unhandled instruction: {s,u}cvtf
1759 335902 arm64: unhandled instruction: sli
1760 335903 arm64: unhandled instruction: umull (vector)
1761 336055 arm64: unhandled instruction: mov (element)
1762 336062 arm64: unhandled instruction: shrn{,2}
1763 336139 mip64: [...] valgrind hangs and spins on a single core [...]
1764 336189 arm64: unhandled Instruction: mvn
1765 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
1766 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
1767 336772 Make moans about unknown ioctls more informative
1768 336957 Add a section about the Solaris/illumos port on the webpage
1769 337094 ifunc wrapper is broken on ppc64
1770 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
1771 337528 leak check heuristic for block prefixed by length as 64bit number
1772 337740 Implement additional Xen hypercalls
1773 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
1774 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
1775 337871 deprecate --db-attach
1776 338023 Add support for all V4L2/media ioctls
1777 338024 inlined functions are not shown if DW_AT_ranges is used
1778 338106 Add support for 'kcmp' syscall
1779 338115 DRD: computed conflict set differs from actual after fork
1780 338160 implement display of thread local storage in gdbsrv
1781 338205 configure.ac and check for -Wno-tautological-compare
1782 338300 coredumps are missing one byte of every segment
1783 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
1784 338499 --sim-hints parsing broken due to wrong order in tokens
1785 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
1786 338681 Unable to unwind through clone thread created on i386-linux
1787 338698 race condition between gdbsrv and vgdb on startup
1788 338703 helgrind on arm-linux gets false positives in dynamic loader
1789 338791 alt dwz files can be relative of debug/main file
1790 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
1791 338932 build V-trunk with gcc-trunk
1792 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
1793 345079 Fix build problems in VEX/useful/test_main.c
1794 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
1795 n-i-bz s390x: Fix memory corruption for multithreaded applications
1796 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
1797 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
1798 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
1799 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
1800 n-i-bz Add command-line option --default-suppressions
1801 n-i-bz Add support for BLKDISCARDZEROES ioctl
1802 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
1803 n-i-bz Add support for sys_pivot_root and sys_unshare
1805 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
1806 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
1807 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
1811 Release 3.9.0 (31 October 2013)
1812 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1813 3.9.0 is a feature release with many improvements and the usual
1814 collection of bug fixes.
1816 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
1817 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
1818 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
1819 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
1821 * ================== PLATFORM CHANGES =================
1823 * Support for MIPS64 LE and BE running Linux. Valgrind has been
1824 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
1826 * Support for MIPS DSP ASE on MIPS32 platforms.
1828 * Support for s390x Decimal Floating Point instructions on hosts that
1829 have the DFP facility installed.
1831 * Support for POWER8 (Power ISA 2.07) instructions
1833 * Support for Intel AVX2 instructions. This is available only on 64
1836 * Initial support for Intel Transactional Synchronization Extensions,
1839 * Initial support for Hardware Transactional Memory on POWER.
1841 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
1842 run large GUI apps tolerably well.
1844 * ==================== TOOL CHANGES ====================
1848 - Improvements in handling of vectorised code, leading to
1849 significantly fewer false error reports. You need to use the flag
1850 --partial-loads-ok=yes to get the benefits of these changes.
1852 - Better control over the leak checker. It is now possible to
1853 specify which leak kinds (definite/indirect/possible/reachable)
1854 should be displayed, which should be regarded as errors, and which
1855 should be suppressed by a given leak suppression. This is done
1856 using the options --show-leak-kinds=kind1,kind2,..,
1857 --errors-for-leak-kinds=kind1,kind2,.. and an optional
1858 "match-leak-kinds:" line in suppression entries, respectively.
1860 Note that generated leak suppressions contain this new line and
1861 are therefore more specific than in previous releases. To get the
1862 same behaviour as previous releases, remove the "match-leak-kinds:"
1863 line from generated suppressions before using them.
1865 - Reduced "possible leak" reports from the leak checker by the use
1866 of better heuristics. The available heuristics provide detection
1867 of valid interior pointers to std::stdstring, to new[] allocated
1868 arrays with elements having destructors and to interior pointers
1869 pointing to an inner part of a C++ object using multiple
1870 inheritance. They can be selected individually using the
1871 option --leak-check-heuristics=heur1,heur2,...
1873 - Better control of stacktrace acquisition for heap-allocated
1874 blocks. Using the --keep-stacktraces option, it is possible to
1875 control independently whether a stack trace is acquired for each
1876 allocation and deallocation. This can be used to create better
1877 "use after free" errors or to decrease Valgrind's resource
1878 consumption by recording less information.
1880 - Better reporting of leak suppression usage. The list of used
1881 suppressions (shown when the -v option is given) now shows, for
1882 each leak suppressions, how many blocks and bytes it suppressed
1883 during the last leak search.
1887 - False errors resulting from the use of statically initialised
1888 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
1891 - False errors resulting from the use of pthread_cond_waits that
1892 timeout, have been removed.
1894 * ==================== OTHER CHANGES ====================
1896 * Some attempt to tune Valgrind's space requirements to the expected
1897 capabilities of the target:
1899 - The default size of the translation cache has been reduced from 8
1900 sectors to 6 on Android platforms, since each sector occupies
1901 about 40MB when using Memcheck.
1903 - The default size of the translation cache has been increased to 16
1904 sectors on all other platforms, reflecting the fact that large
1905 applications require instrumentation and storage of huge amounts
1906 of code. For similar reasons, the number of memory mapped
1907 segments that can be tracked has been increased by a factor of 6.
1909 - In all cases, the maximum number of sectors in the translation
1910 cache can be controlled by the new flag --num-transtab-sectors.
1912 * Changes in how debug info (line numbers, etc) is read:
1914 - Valgrind no longer temporarily mmaps the entire object to read
1915 from it. Instead, reading is done through a small fixed sized
1916 buffer. This avoids virtual memory usage spikes when Valgrind
1917 reads debuginfo from large shared objects.
1919 - A new experimental remote debug info server. Valgrind can read
1920 debug info from a different machine (typically, a build host)
1921 where debuginfo objects are stored. This can save a lot of time
1922 and hassle when running Valgrind on resource-constrained targets
1923 (phones, tablets) when the full debuginfo objects are stored
1924 somewhere else. This is enabled by the --debuginfo-server=
1927 - Consistency checking between main and debug objects can be
1928 disabled using the --allow-mismatched-debuginfo option.
1930 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
1931 scanning can recover stack traces in some cases when the normal
1932 unwind mechanisms fail. Stack scanning is best described as "a
1933 nasty, dangerous and misleading hack" and so is disabled by default.
1934 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
1937 * Detection and merging of recursive stack frame cycles. When your
1938 program has recursive algorithms, this limits the memory used by
1939 Valgrind for recorded stack traces and avoids recording
1940 uninteresting repeated calls. This is controlled by the command
1941 line option --merge-recursive-frame and by the monitor command
1942 "v.set merge-recursive-frames".
1944 * File name and line numbers for used suppressions. The list of used
1945 suppressions (shown when the -v option is given) now shows, for each
1946 used suppression, the file name and line number where the suppression
1949 * New and modified GDB server monitor features:
1951 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
1952 that can be used to execute gdbserver monitor commands from the
1955 - A new monitor command, "v.info open_fds", that gives the list of
1956 open file descriptors and additional details.
1958 - An optional message in the "v.info n_errs_found" monitor command,
1959 for example "v.info n_errs_found test 1234 finished", allowing a
1960 comment string to be added to the process output, perhaps for the
1961 purpose of separating errors of different tests or test phases.
1963 - A new monitor command "v.info execontext" that shows information
1964 about the stack traces recorded by Valgrind.
1966 - A new monitor command "v.do expensive_sanity_check_general" to run
1967 some internal consistency checks.
1969 * New flag --sigill-diagnostics to control whether a diagnostic
1970 message is printed when the JIT encounters an instruction it can't
1971 translate. The actual behavior -- delivery of SIGILL to the
1972 application -- is unchanged.
1974 * The maximum amount of memory that Valgrind can use on 64 bit targets
1975 has been increased from 32GB to 64GB. This should make it possible
1976 to run applications on Memcheck that natively require up to about 35GB.
1978 * ==================== FIXED BUGS ====================
1980 The following bugs have been fixed or resolved. Note that "n-i-bz"
1981 stands for "not in bugzilla" -- that is, a bug that was reported to us
1982 but never got a bugzilla entry. We encourage you to file bugs in
1983 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1984 than mailing the developers (or mailing lists) directly -- bugs that
1985 are not entered into bugzilla tend to get forgotten about or ignored.
1987 To see details of a given bug, visit
1988 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1989 where XXXXXX is the bug number as listed below.
1991 123837 system call: 4th argument is optional, depending on cmd
1992 135425 memcheck should tell you where Freed blocks were Mallocd
1993 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
1994 207815 Adds some of the drm ioctls to syswrap-linux.c
1995 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
1996 252955 Impossible to compile with ccache
1997 253519 Memcheck reports auxv pointer accesses as invalid reads.
1998 263034 Crash when loading some PPC64 binaries
1999 269599 Increase deepest backtrace
2000 274695 s390x: Support "compare to/from logical" instructions (z196)
2001 275800 s390x: Autodetect cache info (part 2)
2002 280271 Valgrind reports possible memory leaks on still-reachable std::string
2003 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
2004 289578 Backtraces with ARM unwind tables (stack scan flags)
2005 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
2006 304832 ppc32: build failure
2007 305431 Use find_buildid shdr fallback for separate .debug files
2008 305728 Add support for AVX2 instructions
2009 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
2010 306035 s390x: Fix IR generation for LAAG and friends
2011 306054 s390x: Condition code computation for convert-to-int/logical
2012 306098 s390x: alternate opcode form for convert to/from fixed
2013 306587 Fix cache line detection from auxiliary vector for PPC.
2014 306783 Mips unhandled syscall : 4025 / 4079 / 4182
2015 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
2016 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
2017 307101 sys_capget second argument can be NULL
2018 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
2019 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
2020 307113 s390x: DFP support
2021 307141 valgrind does't work in mips-linux system
2022 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
2023 307285 x86_amd64 feature test for avx in test suite is wrong
2024 307290 memcheck overlap testcase needs memcpy version filter
2025 307463 Please add "&limit=0" to the "all open bugs" link
2026 307465 --show-possibly-lost=no should reduce the error count / exit code
2027 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
2028 307729 pkgconfig support broken valgrind.pc
2029 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
2030 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
2031 308089 Unhandled syscall on ppc64: prctl
2032 308135 PPC32 MPC8xx has 16 bytes cache size
2033 308321 testsuite memcheck filter interferes with gdb_filter
2035 308341 vgdb should report process exit (or fatal signal)
2036 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
2037 308495 Remove build dependency on installed Xen headers
2038 308573 Internal error on 64-bit instruction executed in 32-bit mode
2040 308627 pmovmskb validity bit propagation is imprecise
2041 308644 vgdb command for having the info for the track-fds option
2042 308711 give more info about aspacemgr and arenas in out_of_memory
2043 308717 ARM: implement fixed-point VCVT.F64.[SU]32
2044 308718 ARM implement SMLALBB family of instructions
2045 308886 Missing support for PTRACE_SET/GETREGSET
2046 308930 syscall name_to_handle_at (303 on amd64) not handled
2047 309229 V-bit tester does not report number of tests generated
2048 309323 print unrecognized instuction on MIPS
2049 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
2050 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
2051 309430 Self hosting ppc64 encounters a vassert error on operand type
2052 309600 valgrind is a bit confused about 0-sized sections
2053 309823 Generate errors for still reachable blocks
2054 309921 PCMPISTRI validity bit propagation is imprecise
2055 309922 none/tests/ppc64/test_dfp5 sometimes fails
2056 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
2057 310424 --read-var-info does not properly describe static variables
2058 310792 search additional path for debug symbols
2059 310931 s390x: Message-security assist (MSA) instruction extension [..]
2060 311100 PPC DFP implementation of the integer operands is inconsistent [..]
2061 311318 ARM: "128-bit constant is not implemented" error message
2062 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
2063 311690 V crashes because it redirects branches inside of a redirected function
2064 311880 x86_64: make regtest hangs at shell_valid1
2065 311922 WARNING: unhandled syscall: 170
2067 312171 ppc: insn selection for DFP
2068 312571 Rounding mode call wrong for the DFP Iops [..]
2069 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
2070 312913 Dangling pointers error should also report the alloc stack trace
2071 312980 Building on Mountain Lion generates some compiler warnings
2072 313267 Adding MIPS64/Linux port to Valgrind
2075 313811 Buffer overflow in assert_fail
2076 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
2077 314269 ppc: dead code in insn selection
2078 314718 ARM: implement integer divide instruction (sdiv and udiv)
2079 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
2080 315441 sendmsg syscall should ignore unset msghdr msg_flags
2081 315534 msgrcv inside a thread causes valgrind to hang (block)
2082 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
2083 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
2084 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
2085 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
2086 316144 valgrind.1 manpage contains unknown ??? strings [..]
2087 316145 callgrind command line options in manpage reference (unknown) [..]
2088 316145 callgrind command line options in manpage reference [..]
2089 316181 drd: Fixed a 4x slowdown for certain applications
2090 316503 Valgrind does not support SSE4 "movntdqa" instruction
2091 316535 Use of |signed int| instead of |size_t| in valgrind messages
2092 316696 fluidanimate program of parsec 2.1 stuck
2093 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
2094 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
2095 317186 "Impossible happens" when occurs VCVT instruction on ARM
2096 317318 Support for Threading Building Blocks "scalable_malloc"
2097 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
2098 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
2099 317463 bmi testcase IR SANITY CHECK FAILURE
2100 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2101 318050 libmpiwrap fails to compile with out-of-source build
2102 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2103 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2104 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2105 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2106 318932 Add missing PPC64 and PPC32 system call support
2107 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2108 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2109 319494 VEX Makefile-gcc standalone build update after r2702
2110 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2111 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2112 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2113 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2114 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2115 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2116 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2117 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2118 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2119 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2120 320895 add fanotify support (patch included)
2121 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2122 321065 Valgrind updates for Xen 4.3
2123 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2124 321363 Unhandled instruction: SSAX (ARM + Thumb)
2125 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2126 321466 Unhandled instruction: SHASX (ARM + Thumb)
2127 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2128 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2129 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2130 321620 Unhandled instruction: UASX (ARM + Thumb)
2131 321621 Unhandled instruction: USAX (ARM + Thumb)
2132 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2133 321693 Unhandled instruction: LDRSBT (Thumb)
2134 321694 Unhandled instruction: UQASX (ARM + Thumb)
2135 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2136 321697 Unhandled instruction: UHASX (ARM + Thumb)
2137 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2138 321704 Unhandled instruction: REVSH (ARM + Thumb)
2139 321730 Add cg_diff and cg_merge man pages
2140 321738 Add vgdb and valgrind-listener man pages
2142 321891 Unhandled instruction: LDRHT (Thumb)
2143 321960 pthread_create() then alloca() causing invalid stack write errors
2144 321969 ppc32 and ppc64 don't support [lf]setxattr
2145 322254 Show threadname together with tid if set by application
2146 322294 Add initial support for IBM Power ISA 2.07
2147 322368 Assertion failure in wqthread_hijack under OS X 10.8
2148 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2149 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2150 322851 0bXXX binary literal syntax is not standard
2151 323035 Unhandled instruction: LDRSHT(Thumb)
2152 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2153 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2154 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2155 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2156 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2157 323437 Phase 2 support for IBM Power ISA 2.07
2158 323713 Support mmxext (integer sse) subset on i386 (athlon)
2159 323803 Transactional memory instructions are not supported for Power
2160 323893 SSE3 not available on amd cpus in valgrind
2161 323905 Probable false positive from Valgrind/drd on close()
2162 323912 valgrind.h header isn't compatible for mingw64
2163 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
2164 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
2165 324181 mmap does not handle MAP_32BIT
2166 324227 memcheck false positive leak when a thread calls exit+block [..]
2167 324421 Support for fanotify API on ARM architecture
2168 324514 gdbserver monitor cmd output behaviour consistency [..]
2169 324518 ppc64: Emulation of dcbt instructions does not handle [..]
2170 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
2171 324582 When access is made to freed memory, report both allocation [..]
2172 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
2173 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
2174 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
2175 324834 Unhandled instructions in Microsoft C run-time for x86_64
2176 324894 Phase 3 support for IBM Power ISA 2.07
2177 326091 drd: Avoid false race reports from optimized strlen() impls
2178 326113 valgrind libvex hwcaps error on AMD64
2179 n-i-bz Some wrong command line options could be ignored
2180 n-i-bz patch to allow fair-sched on android
2181 n-i-bz report error for vgdb snapshot requested before execution
2182 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
2184 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
2188 Release 3.8.1 (19 September 2012)
2189 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2190 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
2191 that occur moderately frequently in real use cases, adds support for
2192 some missing instructions on ARM, and fixes a deadlock condition on
2193 MacOSX. If you package or deliver 3.8.0 for others to use, you might
2194 want to consider upgrading to 3.8.1 instead.
2196 The following bugs have been fixed or resolved. Note that "n-i-bz"
2197 stands for "not in bugzilla" -- that is, a bug that was reported to us
2198 but never got a bugzilla entry. We encourage you to file bugs in
2199 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2200 than mailing the developers (or mailing lists) directly -- bugs that
2201 are not entered into bugzilla tend to get forgotten about or ignored.
2203 To see details of a given bug, visit
2204 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2205 where XXXXXX is the bug number as listed below.
2208 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
2209 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
2210 298281 wcslen causes false(?) uninitialised value warnings
2211 301281 valgrind hangs on OS X when the process calls system()
2212 304035 disInstr(arm): unhandled instruction 0xE1023053
2213 304867 implement MOVBE instruction in x86 mode
2214 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
2215 305042 amd64: implement 0F 7F encoding of movq between two registers
2216 305199 ARM: implement QDADD and QDSUB
2217 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
2218 305513 killed by fatal signal: SIGSEGV
2219 305690 DRD reporting invalid semaphore when sem_trywait fails
2220 305926 Invalid alignment checks for some AVX instructions
2221 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
2222 306310 3.8.0 release tarball missing some files
2223 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
2224 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
2225 n-i-bz shmat of a segment > 4Gb does not work
2226 n-i-bz simulate_control_c script wrong USR1 signal number on mips
2227 n-i-bz vgdb ptrace calls wrong on mips [...]
2228 n-i-bz Fixes for more MPI false positives
2229 n-i-bz exp-sgcheck's memcpy causes programs to segfault
2230 n-i-bz OSX build w/ clang: asserts at startup
2231 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
2232 n-i-bz fix a couple of union tag-vs-field mixups
2233 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
2235 The following bugs were fixed in 3.8.0 but not listed in this NEWS
2238 254088 Valgrind should know about UD2 instruction
2241 304754 NEWS blows TeX's little mind
2243 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
2247 Release 3.8.0 (10 August 2012)
2248 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2249 3.8.0 is a feature release with many improvements and the usual
2250 collection of bug fixes.
2252 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2253 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
2254 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
2255 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
2256 There is initial support for MacOSX 10.8, but it is not usable for
2257 serious work at present.
2259 * ================== PLATFORM CHANGES =================
2261 * Support for MIPS32 platforms running Linux. Valgrind has been
2262 tested on MIPS32 and MIPS32r2 platforms running different Debian
2263 Squeeze and MeeGo distributions. Both little-endian and big-endian
2264 cores are supported. The tools Memcheck, Massif and Lackey have
2265 been tested and are known to work. See README.mips for more details.
2267 * Preliminary support for Android running on x86.
2269 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
2271 * Support for Intel AVX instructions and for AES instructions. This
2272 support is available only for 64 bit code.
2274 * Support for POWER Decimal Floating Point instructions.
2276 * ==================== TOOL CHANGES ====================
2278 * Non-libc malloc implementations are now supported. This is useful
2279 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
2280 Using the new option --soname-synonyms, such tools can be informed
2281 that the malloc implementation is either linked statically into the
2282 executable, or is present in some other shared library different
2283 from libc.so. This makes it possible to process statically linked
2284 programs, and programs using other malloc libraries, for example
2285 TCMalloc or JEMalloc.
2287 * For tools that provide their own replacement for malloc et al, the
2288 option --redzone-size=<number> allows users to specify the size of
2289 the padding blocks (redzones) added before and after each client
2290 allocated block. Smaller redzones decrease the memory needed by
2291 Valgrind. Bigger redzones increase the chance to detect blocks
2292 overrun or underrun. Prior to this change, the redzone size was
2293 hardwired to 16 bytes in Memcheck.
2297 - The leak_check GDB server monitor command now can
2298 control the maximum nr of loss records to output.
2300 - Reduction of memory use for applications allocating
2301 many blocks and/or having many partially defined bytes.
2303 - Addition of GDB server monitor command 'block_list' that lists
2304 the addresses/sizes of the blocks of a leak search loss record.
2306 - Addition of GDB server monitor command 'who_points_at' that lists
2307 the locations pointing at a block.
2309 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2310 detect an invalid access of these redzones, by marking them
2311 noaccess. Similarly, if a redzone size is given for a memory
2312 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2313 This still allows to find some bugs if the user has forgotten to
2314 mark the pool superblock noaccess.
2316 - Performance of memory leak check has been improved, especially in
2317 cases where there are many leaked blocks and/or many suppression
2318 rules used to suppress leak reports.
2320 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2321 more precise analysis, which is important for LLVM/Clang
2322 generated code. This is at the cost of somewhat reduced
2323 performance. Note there is no change to analysis precision or
2324 costs on Linux targets.
2328 - Added even more facilities that can help finding the cause of a data
2329 race, namely the command-line option --ptrace-addr and the macro
2330 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2332 - Fixed a subtle bug that could cause false positive data race reports.
2334 * ==================== OTHER CHANGES ====================
2336 * The C++ demangler has been updated so as to work well with C++
2337 compiled by up to at least g++ 4.6.
2339 * Tool developers can make replacement/wrapping more flexible thanks
2340 to the new option --soname-synonyms. This was reported above, but
2341 in fact is very general and applies to all function
2342 replacement/wrapping, not just to malloc-family functions.
2344 * Round-robin scheduling of threads can be selected, using the new
2345 option --fair-sched= yes. Prior to this change, the pipe-based
2346 thread serialisation mechanism (which is still the default) could
2347 give very unfair scheduling. --fair-sched=yes improves
2348 responsiveness of interactive multithreaded applications, and
2349 improves repeatability of results from the thread checkers Helgrind
2352 * For tool developers: support to run Valgrind on Valgrind has been
2353 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2355 * gdbserver now shows the float shadow registers as integer
2356 rather than float values, as the shadow values are mostly
2357 used as bit patterns.
2359 * Increased limit for the --num-callers command line flag to 500.
2361 * Performance improvements for error matching when there are many
2362 suppression records in use.
2364 * Improved support for DWARF4 debugging information (bug 284184).
2366 * Initial support for DWZ compressed Dwarf debug info.
2368 * Improved control over the IR optimiser's handling of the tradeoff
2369 between performance and precision of exceptions. Specifically,
2370 --vex-iropt-precise-memory-exns has been removed and replaced by
2371 --vex-iropt-register-updates, with extended functionality. This
2372 allows the Valgrind gdbserver to always show up to date register
2375 * Modest performance gains through the use of translation chaining for
2378 * ==================== FIXED BUGS ====================
2380 The following bugs have been fixed or resolved. Note that "n-i-bz"
2381 stands for "not in bugzilla" -- that is, a bug that was reported to us
2382 but never got a bugzilla entry. We encourage you to file bugs in
2383 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2384 than mailing the developers (or mailing lists) directly -- bugs that
2385 are not entered into bugzilla tend to get forgotten about or ignored.
2387 To see details of a given bug, visit
2388 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2389 where XXXXXX is the bug number as listed below.
2391 197914 Building valgrind from svn now requires automake-1.10
2392 203877 increase to 16Mb maximum allowed alignment for memalign et al
2393 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
2394 247386 make perf does not run all performance tests
2395 270006 Valgrind scheduler unfair
2396 270777 Adding MIPS/Linux port to Valgrind
2397 270796 s390x: Removed broken support for the TS insn
2398 271438 Fix configure for proper SSE4.2 detection
2399 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
2400 273475 Add support for AVX instructions
2401 274078 improved configure logic for mpicc
2402 276993 fix mremap 'no thrash checks'
2403 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
2404 281482 memcheck incorrect byte allocation count in realloc() for silly argument
2405 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
2406 283413 Fix wrong sanity check
2407 283671 Robustize alignment computation in LibVEX_Alloc
2408 283961 Adding support for some HCI IOCTLs
2409 284124 parse_type_DIE: confused by: DWARF 4
2410 284864 == 273475 (Add support for AVX instructions)
2411 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
2412 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
2413 285725 == 273475 (Add support for AVX instructions)
2414 286261 add wrapper for linux I2C_RDWR ioctl
2415 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
2416 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
2417 286384 configure fails "checking for a supported version of gcc"
2418 286497 == 273475 (Add support for AVX instructions)
2419 286596 == 273475 (Add support for AVX instructions)
2420 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
2421 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
2422 287260 Incorrect conditional jump or move depends on uninitialised value(s)
2423 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
2424 287307 == 273475 (Add support for AVX instructions)
2425 287858 VG_(strerror): unknown error
2426 288298 (MacOSX) unhandled syscall shm_unlink
2427 288995 == 273475 (Add support for AVX instructions)
2428 289470 Loading of large Mach-O thin binaries fails.
2429 289656 == 273475 (Add support for AVX instructions)
2430 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
2431 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
2432 289839 s390x: Provide support for unicode conversion instructions
2433 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
2434 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
2435 290655 Add support for AESKEYGENASSIST instruction
2436 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
2437 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
2438 291253 ES register not initialised in valgrind simulation
2439 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
2440 291865 s390x: Support the "Compare Double and Swap" family of instructions
2441 292300 == 273475 (Add support for AVX instructions)
2442 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
2443 292493 == 273475 (Add support for AVX instructions)
2444 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
2445 292627 Missing support for some SCSI ioctls
2446 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
2447 292841 == 273475 (Add support for AVX instructions)
2448 292993 implement the getcpu syscall on amd64-linux
2449 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
2450 293088 Add some VEX sanity checks for ppc64 unhandled instructions
2451 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
2452 293754 PCMPxSTRx not implemented for 16-bit characters
2453 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
2454 293808 CLFLUSH not supported by latest VEX for amd64
2455 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
2456 294048 MPSADBW instruction not implemented
2457 294055 regtest none/tests/shell fails when locale is not set to C
2458 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
2459 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
2460 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
2461 294260 disInstr_AMD64: disInstr miscalculated next %rip
2462 294523 --partial-loads-ok=yes causes false negatives
2463 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
2464 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
2465 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
2466 295089 can not annotate source for both helgrind and drd
2467 295221 POWER Processor decimal floating point instruction support missing
2468 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
2469 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
2470 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
2471 295617 ARM - Add some missing syscalls
2472 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
2473 296229 Linux user input device ioctls missing wrappers
2474 296318 ELF Debug info improvements (more than one rx/rw mapping)
2475 296422 Add translation chaining support
2476 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
2477 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
2478 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
2479 297078 gdbserver signal handling problems [..]
2480 297147 drd false positives on newly allocated memory
2481 297329 disallow decoding of IBM Power DFP insns on some machines
2482 297497 POWER Processor decimal floating point instruction support missing
2483 297701 Another alias for strncasecmp_l in libc-2.13.so
2484 297911 'invalid write' not reported when using APIs for custom mem allocators.
2485 297976 s390x: revisit EX implementation
2486 297991 Valgrind interferes with mmap()+ftell()
2487 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
2488 297993 Fix compilation of valgrind with gcc -g3.
2489 298080 POWER Processor DFP support missing, part 3
2490 298227 == 273475 (Add support for AVX instructions)
2491 298335 == 273475 (Add support for AVX instructions)
2492 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
2493 298394 s390x: Don't bail out on an unknown machine model. [..]
2494 298421 accept4() syscall (366) support is missing for ARM
2495 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
2496 298732 valgrind installation problem in ubuntu with kernel version 3.x
2497 298862 POWER Processor DFP instruction support missing, part 4
2498 298864 DWARF reader mis-parses DW_FORM_ref_addr
2499 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
2500 299053 Support DWARF4 DW_AT_high_pc constant form
2501 299104 == 273475 (Add support for AVX instructions)
2502 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
2503 299629 dup3() syscall (358) support is missing for ARM
2504 299694 POWER Processor DFP instruction support missing, part 5
2505 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
2506 299803 == 273475 (Add support for AVX instructions)
2507 299804 == 273475 (Add support for AVX instructions)
2508 299805 == 273475 (Add support for AVX instructions)
2509 300140 ARM - Missing (T1) SMMUL
2510 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
2511 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
2512 300414 FCOM and FCOMP unimplemented for amd64 guest
2513 301204 infinite loop in canonicaliseSymtab with ifunc symbol
2514 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
2515 301265 add x86 support to Android build
2516 301984 configure script doesn't detect certain versions of clang
2517 302205 Fix compiler warnings for POWER VEX code and POWER test cases
2518 302287 Unhandled movbe instruction on Atom processors
2519 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
2520 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
2521 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
2522 302656 == 273475 (Add support for AVX instructions)
2523 302709 valgrind for ARM needs extra tls support for android emulator [..]
2524 302827 add wrapper for CDROM_GET_CAPABILITY
2525 302901 Valgrind crashes with dwz optimized debuginfo
2526 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
2527 303116 Add support for the POWER instruction popcntb
2528 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
2529 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
2530 303466 == 273475 (Add support for AVX instructions)
2531 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
2532 303963 strstr() function produces wrong results under valgrind callgrind
2533 304054 CALL_FN_xx macros need to enforce stack alignment
2534 304561 tee system call not supported
2535 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
2536 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
2537 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
2538 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
2539 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
2540 n-i-bz s390x: Shadow registers can now be examined using vgdb
2542 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
2543 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
2547 Release 3.7.0 (5 November 2011)
2548 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2549 3.7.0 is a feature release with many significant improvements and the
2550 usual collection of bug fixes.
2552 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2553 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
2554 Support for recent distros and toolchain components (glibc 2.14, gcc
2555 4.6, MacOSX 10.7) has been added.
2557 * ================== PLATFORM CHANGES =================
2559 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
2560 analyse 64-bit programs running on z/Architecture. Most user space
2561 instructions up to and including z10 are supported. Valgrind has
2562 been tested extensively on z9, z10, and z196 machines running SLES
2563 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
2564 known to work well. Callgrind, Helgrind, and DRD work reasonably
2565 well on z9 and later models. See README.s390 for more details.
2567 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
2568 64-bit processes are supported. Some complex threaded applications
2569 (Firefox) are observed to hang when run as 32 bit applications,
2570 whereas 64-bit versions run OK. The cause is unknown. Memcheck
2571 will likely report some false errors. In general, expect some rough
2572 spots. This release also supports MacOSX 10.6, but drops support
2575 * Preliminary support for Android (on ARM). Valgrind can now run
2576 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
2577 README.android for more details, plus instructions on how to get
2580 * Support for the IBM Power ISA 2.06 (Power7 instructions)
2582 * General correctness and performance improvements for ARM/Linux, and,
2583 by extension, ARM/Android.
2585 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
2586 instruction set support is under development but is not available in
2589 * Support for AIX5 has been removed.
2591 * ==================== TOOL CHANGES ====================
2593 * Memcheck: some incremental changes:
2595 - reduction of memory use in some circumstances
2597 - improved handling of freed memory, which in some circumstances
2598 can cause detection of use-after-free that would previously have
2601 - fix of a longstanding bug that could cause false negatives (missed
2602 errors) in programs doing vector saturated narrowing instructions.
2604 * Helgrind: performance improvements and major memory use reductions,
2605 particularly for large, long running applications which perform many
2606 synchronisation (lock, unlock, etc) events. Plus many smaller
2609 - display of locksets for both threads involved in a race
2611 - general improvements in formatting/clarity of error messages
2613 - addition of facilities and documentation regarding annotation
2614 of thread safe reference counted C++ classes
2616 - new flag --check-stack-refs=no|yes [yes], to disable race checking
2617 on thread stacks (a performance hack)
2619 - new flag --free-is-write=no|yes [no], to enable detection of races
2620 where one thread accesses heap memory but another one frees it,
2621 without any coordinating synchronisation event
2623 * DRD: enabled XML output; added support for delayed thread deletion
2624 in order to detect races that occur close to the end of a thread
2625 (--join-list-vol); fixed a memory leak triggered by repeated client
2626 memory allocatation and deallocation; improved Darwin support.
2628 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
2630 * exp-sgcheck: this tool has been reduced in scope so as to improve
2631 performance and remove checking that Memcheck does better.
2632 Specifically, the ability to check for overruns for stack and global
2633 arrays is unchanged, but the ability to check for overruns of heap
2634 blocks has been removed. The tool has accordingly been renamed to
2635 exp-sgcheck ("Stack and Global Array Checking").
2637 * ==================== OTHER CHANGES ====================
2639 * GDB server: Valgrind now has an embedded GDB server. That means it
2640 is possible to control a Valgrind run from GDB, doing all the usual
2641 things that GDB can do (single stepping, breakpoints, examining
2642 data, etc). Tool-specific functionality is also available. For
2643 example, it is possible to query the definedness state of variables
2644 or memory from within GDB when running Memcheck; arbitrarily large
2645 memory watchpoints are supported, etc. To use the GDB server, start
2646 Valgrind with the flag --vgdb-error=0 and follow the on-screen
2649 * Improved support for unfriendly self-modifying code: a new option
2650 --smc-check=all-non-file is available. This adds the relevant
2651 consistency checks only to code that originates in non-file-backed
2652 mappings. In effect this confines the consistency checking only to
2653 code that is or might be JIT generated, and avoids checks on code
2654 that must have been compiled ahead of time. This significantly
2655 improves performance on applications that generate code at run time.
2657 * It is now possible to build a working Valgrind using Clang-2.9 on
2660 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
2661 These enable and disable error reporting on a per-thread, and
2662 nestable, basis. This is useful for hiding errors in particularly
2663 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
2664 now uses this facility.
2666 * Added the --mod-funcname option to cg_diff.
2668 * ==================== FIXED BUGS ====================
2670 The following bugs have been fixed or resolved. Note that "n-i-bz"
2671 stands for "not in bugzilla" -- that is, a bug that was reported to us
2672 but never got a bugzilla entry. We encourage you to file bugs in
2673 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2674 mailing the developers (or mailing lists) directly -- bugs that are
2675 not entered into bugzilla tend to get forgotten about or ignored.
2677 To see details of a given bug, visit
2678 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2679 where XXXXXX is the bug number as listed below.
2681 79311 malloc silly arg warning does not give stack trace
2682 210935 port valgrind.h (not valgrind) to win32 to support client requests
2683 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
2684 243404 Port to zSeries
2685 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
2686 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
2687 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
2688 253206 Some fixes for the faultstatus testcase
2689 255223 capget testcase fails when running as root
2690 256703 xlc_dbl_u32.c testcase broken
2691 256726 Helgrind tests have broken inline asm
2692 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
2693 264800 testcase compile failure on zseries
2694 265762 make public VEX headers compilable by G++ 3.x
2695 265771 assertion in jumps.c (r11523) fails with glibc-2.3
2696 266753 configure script does not give the user the option to not use QtCore
2697 266931 gen_insn_test.pl is broken
2698 266961 ld-linux.so.2 i?86-linux strlen issues
2699 266990 setns instruction causes false positive
2700 267020 Make directory for temporary files configurable at run-time.
2701 267342 == 267997 (segmentation fault on Mac OS 10.6)
2702 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
2703 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
2704 267488 regtest: darwin support for 64-bit build
2705 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
2706 267630 Add support for IBM Power ISA 2.06 -- stage 1
2707 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
2708 267819 Add client request for informing the core about reallocation
2709 267925 laog data structure quadratic for a single sequence of lock
2710 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
2711 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
2712 268513 missed optimizations in fold_Expr
2713 268619 s390x: fpr - gpr transfer facility
2714 268620 s390x: reconsider "long displacement" requirement
2715 268621 s390x: improve IR generation for XC
2716 268715 s390x: FLOGR is not universally available
2717 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
2718 268930 s390x: MHY is not universally available
2719 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
2720 269079 Support ptrace system call on ARM
2721 269144 missing "Bad option" error message
2722 269209 conditional load and store facility (z196)
2723 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
2724 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
2725 269736 s390x: minor code generation tweaks
2726 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
2727 269863 s390x: remove unused function parameters
2728 269864 s390x: tweak s390_emit_load_cc
2729 269884 == 250101 (overhead for huge blocks exhausts space too soon)
2730 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
2731 270115 s390x: rewrite some testcases
2732 270309 == 267997 (valgrind crash on startup)
2733 270320 add support for Linux FIOQSIZE ioctl() call
2734 270326 segfault while trying to sanitize the environment passed to execle
2735 270794 IBM POWER7 support patch causes regression in none/tests
2736 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
2737 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
2738 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
2739 270959 s390x: invalid use of R0 as base register
2740 271042 VSX configure check fails when it should not
2741 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
2742 271259 s390x: fix code confusion
2743 271337 == 267997 (Valgrind segfaults on MacOS X)
2744 271385 s390x: Implement Ist_MBE
2745 271501 s390x: misc cleanups
2746 271504 s390x: promote likely and unlikely
2747 271579 ppc: using wrong enum type
2748 271615 unhandled instruction "popcnt" (arch=amd10h)
2749 271730 Fix bug when checking ioctls: duplicate check
2750 271776 s390x: provide STFLE instruction support
2751 271779 s390x: provide clock instructions like STCK
2752 271799 Darwin: ioctls without an arg report a memory error
2753 271820 arm: fix type confusion
2754 271917 pthread_cond_timedwait failure leads to not-locked false positive
2755 272067 s390x: fix DISP20 macro
2756 272615 A typo in debug output in mc_leakcheck.c
2757 272661 callgrind_annotate chokes when run from paths containing regex chars
2758 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
2759 272955 Unhandled syscall error for pwrite64 on ppc64 arch
2760 272967 make documentation build-system more robust
2761 272986 Fix gcc-4.6 warnings with valgrind.h
2762 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
2763 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
2764 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
2765 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
2766 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
2767 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
2768 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
2769 273778 exp-ptrcheck: unhandled sysno == 259
2770 274089 exp-ptrcheck: unhandled sysno == 208
2771 274378 s390x: Various dispatcher tweaks
2772 274447 WARNING: unhandled syscall: 340
2773 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
2774 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
2775 274926 valgrind does not build against linux-3
2776 275148 configure FAIL with glibc-2.14
2777 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
2778 275168 Make Valgrind work for MacOSX 10.7 Lion
2779 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
2780 275278 valgrind does not build on Linux kernel 3.0.* due to silly
2781 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
2782 275308 Fix implementation for ppc64 fres instruc
2783 275339 s390x: fix testcase compile warnings
2784 275517 s390x: Provide support for CKSM instruction
2785 275710 s390x: get rid of redundant address mode calculation
2786 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
2787 275852 == 250101 (valgrind uses all swap space and is killed)
2788 276784 Add support for IBM Power ISA 2.06 -- stage 3
2789 276987 gdbsrv: fix tests following recent commits
2790 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
2791 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
2792 277471 Unhandled syscall: 340
2793 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
2794 277653 ARM: support Thumb2 PLD instruction
2795 277663 ARM: NEON float VMUL by scalar incorrect
2796 277689 ARM: tests for VSTn with register post-index are broken
2797 277694 ARM: BLX LR instruction broken in ARM mode
2798 277780 ARM: VMOV.F32 (immediate) instruction is broken
2799 278057 fuse filesystem syscall deadlocks
2800 278078 Unimplemented syscall 280 on ppc32
2801 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
2802 278454 VALGRIND_STACK_DEREGISTER has wrong output type
2803 278502 == 275284 (Valgrind confuses memcpy() and memmove())
2804 278892 gdbsrv: factorize gdb version handling, fix doc and typos
2805 279027 Support for MVCL and CLCL instruction
2806 279027 s390x: Provide support for CLCL and MVCL instructions
2807 279062 Remove a redundant check in the insn selector for ppc.
2808 279071 JDK creates PTEST with redundant REX.W prefix
2809 279212 gdbsrv: add monitor cmd v.info scheduler.
2810 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
2811 279698 memcheck discards valid-bits for packuswb
2812 279795 memcheck reports uninitialised values for mincore on amd64
2813 279994 Add support for IBM Power ISA 2.06 -- stage 3
2814 280083 mempolicy syscall check errors
2815 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
2816 280710 s390x: config files for nightly builds
2817 280757 /tmp dir still used by valgrind even if TMPDIR is specified
2818 280965 Valgrind breaks fcntl locks when program does mmap
2819 281138 WARNING: unhandled syscall: 340
2820 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
2821 281304 == 275168 (Darwin: dyld "cannot load inserted library")
2822 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
2823 281468 s390x: handle do_clone and gcc clones in call traces
2824 281488 ARM: VFP register corruption
2825 281828 == 275284 (false memmove warning: "Source and destination overlap")
2826 281883 s390x: Fix system call wrapper for "clone".
2827 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
2828 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
2829 282238 SLES10: make check fails
2830 282979 strcasestr needs replacement with recent(>=2.12) glibc
2831 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
2832 283243 Regression in ppc64 memcheck tests
2833 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
2834 283427 re-connect epoll_pwait syscall on ARM linux
2835 283600 gdbsrv: android: port vgdb.c
2836 283709 none/tests/faultstatus needs to account for page size
2837 284305 filter_gdb needs enhancement to work on ppc64
2838 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
2839 284472 Thumb2 ROR.W encoding T2 not implemented
2840 284621 XML-escape process command line in XML output
2841 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
2842 that have non-power-of-2 sizes (also AMDs)
2843 n-i-bz don't be spooked by libraries mashed by elfhack
2844 n-i-bz don't be spooked by libxul.so linked with gold
2845 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
2847 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
2848 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
2849 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
2853 Release 3.6.1 (16 February 2011)
2854 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2855 3.6.1 is a bug fix release. It adds support for some SSE4
2856 instructions that were omitted in 3.6.0 due to lack of time. Initial
2857 support for glibc-2.13 has been added. A number of bugs causing
2858 crashing or assertion failures have been fixed.
2860 The following bugs have been fixed or resolved. Note that "n-i-bz"
2861 stands for "not in bugzilla" -- that is, a bug that was reported to us
2862 but never got a bugzilla entry. We encourage you to file bugs in
2863 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2864 mailing the developers (or mailing lists) directly -- bugs that are
2865 not entered into bugzilla tend to get forgotten about or ignored.
2867 To see details of a given bug, visit
2868 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2869 where XXXXXX is the bug number as listed below.
2871 188572 Valgrind on Mac should suppress setenv() mem leak
2872 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
2873 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
2874 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
2875 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
2876 254420 memory pool tracking broken
2877 254957 Test code failing to compile due to changes in memcheck.h
2878 255009 helgrind/drd: crash on chmod with invalid parameter
2879 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
2880 255355 helgrind/drd: crash on threaded programs doing fork
2882 255418 (SSE4.x) rint call compiled with ICC
2883 255822 --gen-suppressions can create invalid files: "too many callers [...]"
2884 255888 closing valgrindoutput tag outputted to log-stream on error
2885 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
2886 255966 Slowness when using mempool annotations
2887 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
2888 256600 super-optimized strcasecmp() false positive
2889 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
2890 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
2891 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
2892 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
2893 257276 Missing case in memcheck --track-origins=yes
2894 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
2895 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
2896 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
2897 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
2898 263099 callgrind_annotate counts Ir improperly [...]
2899 263877 undefined coprocessor instruction on ARMv7
2900 265964 configure FAIL with glibc-2.13
2901 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
2902 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
2903 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
2904 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
2905 n-i-bz DRD: disable --free-is-write due to implementation difficulties
2907 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
2911 Release 3.6.0 (21 October 2010)
2912 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2913 3.6.0 is a feature release with many significant improvements and the
2914 usual collection of bug fixes.
2916 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2917 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
2918 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
2920 -------------------------
2922 Here are some highlights. Details are shown further down:
2924 * Support for ARM/Linux.
2926 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
2928 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
2930 * Support for the SSE4.2 instruction set.
2932 * Enhancements to the Callgrind profiler, including the ability to
2933 handle CPUs with three levels of cache.
2935 * A new experimental heap profiler, DHAT.
2937 * A huge number of bug fixes and small enhancements.
2939 -------------------------
2941 Here are details of the above changes, together with descriptions of
2942 many other changes, and a list of fixed bugs.
2944 * ================== PLATFORM CHANGES =================
2946 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
2947 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
2948 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
2950 This requires a CPU capable of running the ARMv7-A instruction set
2951 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
2952 of the user space instruction set, including ARM and Thumb integer
2953 code, VFPv3, NEON and V6 media instructions. The Memcheck,
2954 Cachegrind and Massif tools work properly; other tools work to
2957 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
2958 with support for recent releases of the underlying toolchain
2959 components, notably gcc-4.5 and glibc-2.12.
2961 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
2962 support also works much better on OS X 10.5, and is as solid as
2965 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
2966 64-bit mode. In 32-bit mode, support is only available up to and
2967 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
2968 supported in 64-bit mode, and 32-bit mode does in fact support the
2969 bare minimum SSE4 instructions to needed to run programs on Mac OS X
2970 10.6 on 32-bit targets.
2972 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
2973 and including version 2.05 is supported.
2975 * ==================== TOOL CHANGES ====================
2977 * Cachegrind has a new processing script, cg_diff, which finds the
2978 difference between two profiles. It's very useful for evaluating
2979 the performance effects of a change in a program.
2981 Related to this change, the meaning of cg_annotate's (rarely-used)
2982 --threshold option has changed; this is unlikely to affect many
2983 people, if you do use it please see the user manual for details.
2985 * Callgrind now can do branch prediction simulation, similar to
2986 Cachegrind. In addition, it optionally can count the number of
2987 executed global bus events. Both can be used for a better
2988 approximation of a "Cycle Estimation" as derived event (you need to
2989 update the event formula in KCachegrind yourself).
2991 * Cachegrind and Callgrind now refer to the LL (last-level) cache
2992 rather than the L2 cache. This is to accommodate machines with
2993 three levels of caches -- if Cachegrind/Callgrind auto-detects the
2994 cache configuration of such a machine it will run the simulation as
2995 if the L2 cache isn't present. This means the results are less
2996 likely to match the true result for the machine, but
2997 Cachegrind/Callgrind's results are already only approximate, and
2998 should not be considered authoritative. The results are still
2999 useful for giving a general idea about a program's locality.
3001 * Massif has a new option, --pages-as-heap, which is disabled by
3002 default. When enabled, instead of tracking allocations at the level
3003 of heap blocks (as allocated with malloc/new/new[]), it instead
3004 tracks memory allocations at the level of memory pages (as mapped by
3005 mmap, brk, etc). Each mapped page is treated as its own block.
3006 Interpreting the page-level output is harder than the heap-level
3007 output, but this option is useful if you want to account for every
3008 byte of memory used by a program.
3010 * DRD has two new command-line options: --free-is-write and
3011 --trace-alloc. The former allows to detect reading from already freed
3012 memory, and the latter allows tracing of all memory allocations and
3015 * DRD has several new annotations. Custom barrier implementations can
3016 now be annotated, as well as benign races on static variables.
3018 * DRD's happens before / happens after annotations have been made more
3019 powerful, so that they can now also be used to annotate e.g. a smart
3020 pointer implementation.
3022 * Helgrind's annotation set has also been drastically improved, so as
3023 to provide to users a general set of annotations to describe locks,
3024 semaphores, barriers and condition variables. Annotations to
3025 describe thread-safe reference counted heap objects have also been
3028 * Memcheck has a new command-line option, --show-possibly-lost, which
3029 is enabled by default. When disabled, the leak detector will not
3030 show possibly-lost blocks.
3032 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
3033 has been added. DHAT keeps track of allocated heap blocks, and also
3034 inspects every memory reference to see which block (if any) is being
3035 accessed. This gives a lot of insight into block lifetimes,
3036 utilisation, turnover, liveness, and the location of hot and cold
3037 fields. You can use DHAT to do hot-field profiling.
3039 * ==================== OTHER CHANGES ====================
3041 * Improved support for unfriendly self-modifying code: the extra
3042 overhead incurred by --smc-check=all has been reduced by
3043 approximately a factor of 5 as compared with 3.5.0.
3045 * Ability to show directory names for source files in error messages.
3046 This is combined with a flexible mechanism for specifying which
3047 parts of the paths should be shown. This is enabled by the new flag
3050 * A new flag, --require-text-symbol, which will stop the run if a
3051 specified symbol is not found it a given shared object when it is
3052 loaded into the process. This makes advanced working with function
3053 intercepting and wrapping safer and more reliable.
3055 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
3056 and control of Valgrind is now available for the tools Memcheck and
3057 Helgrind. XML output from Valgrind is available for Memcheck,
3058 Helgrind and exp-Ptrcheck.
3060 * More reliable stack unwinding on amd64-linux, particularly in the
3061 presence of function wrappers, and with gcc-4.5 compiled code.
3063 * Modest scalability (performance improvements) for massive
3064 long-running applications, particularly for those with huge amounts
3067 * Support for analyzing programs running under Wine with has been
3068 improved. The header files <valgrind/valgrind.h>,
3069 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
3070 Windows-programs compiled with MinGW or one of the Microsoft Visual
3073 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
3074 The 32-bit simulator was not affected. This did not occur often,
3075 but when it did would usually crash the program under test.
3078 * A large number of bugs were fixed. These are shown below.
3080 * A number of bugs were investigated, and were candidates for fixing,
3081 but are not fixed in 3.6.0, due to lack of developer time. They may
3082 get fixed in later releases. They are:
3084 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
3085 212419 false positive "lock order violated" (A+B vs A)
3086 213685 Undefined value propagates past dependency breaking instruction
3087 216837 Incorrect instrumentation of NSOperationQueue on Darwin
3088 237920 valgrind segfault on fork failure
3089 242137 support for code compiled by LLVM-2.8
3090 242423 Another unknown Intel cache config value
3091 243232 Inconsistent Lock Orderings report with trylock
3092 243483 ppc: callgrind triggers VEX assertion failure
3093 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
3094 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
3096 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
3097 249435 Analyzing wine programs with callgrind triggers a crash
3098 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
3099 250065 Handling large allocations
3100 250101 huge "free" memory usage due to m_mallocfree.c
3101 "superblocks fragmentation"
3102 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3103 252091 Callgrind on ARM does not detect function returns correctly
3104 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3105 254420 memory pool tracking broken
3106 n-i-bz support for adding symbols for JIT generated code
3109 The following bugs have been fixed or resolved. Note that "n-i-bz"
3110 stands for "not in bugzilla" -- that is, a bug that was reported to us
3111 but never got a bugzilla entry. We encourage you to file bugs in
3112 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3113 mailing the developers (or mailing lists) directly -- bugs that are
3114 not entered into bugzilla tend to get forgotten about or ignored.
3116 To see details of a given bug, visit
3117 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3118 where XXXXXX is the bug number as listed below.
3120 135264 dcbzl instruction missing
3122 153699 Valgrind should report unaligned reads with movdqa
3124 190429 Valgrind reports lost of errors in ld.so
3125 with x86_64 2.9.90 glibc
3126 197266 valgrind appears to choke on the xmms instruction
3128 197988 Crash when demangling very large symbol names
3129 202315 unhandled syscall: 332 (inotify_init1)
3130 203256 Add page-level profiling to Massif
3131 205093 dsymutil=yes needs quotes, locking (partial fix)
3132 205241 Snow Leopard 10.6 support (partial fix)
3133 206600 Leak checker fails to upgrade indirect blocks when their
3134 parent becomes reachable
3135 210935 port valgrind.h (not valgrind) to win32 so apps run under
3136 wine can make client requests
3137 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3138 within Linux ip-stack checksum functions
3139 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3141 213685 Undefined value propagates past dependency breaking instruction
3143 215914 Valgrind inserts bogus empty environment variable
3145 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3146 222545 shmat fails under valgind on some arm targets
3147 222560 ARM NEON support
3150 232509 Docs build fails with formatting inside <title></title> elements
3152 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3153 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3154 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3155 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3156 237485 symlink (syscall 57) is not supported on Mac OS
3157 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3159 238208 is_just_below_ESP doesn't take into account red-zone
3160 238345 valgrind passes wrong $0 when executing a shell script
3161 238679 mq_timedreceive syscall doesn't flag the reception buffer
3163 238696 fcntl command F_DUPFD_CLOEXEC not supported
3164 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3165 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3166 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
3168 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
3174 242606 unhandled syscall: setegid (in Ptrcheck)
3175 242814 Helgrind "Impossible has happened" during
3176 QApplication::initInstance();
3177 243064 Valgrind attempting to read debug information from iso
3178 243270 Make stack unwinding in Valgrind wrappers more reliable
3179 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
3180 sysno = 277 (mq_open)
3181 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
3182 244493 ARM VFP d16-d31 registers support
3183 244670 add support for audit_session_self syscall on Mac OS 10.6
3184 244921 The xml report of helgrind tool is not well format
3185 244923 In the xml report file, the <preamble> not escape the
3186 xml char, eg '<','&','>'
3187 245535 print full path names in plain text reports
3188 245925 x86-64 red zone handling problem
3189 246258 Valgrind not catching integer underruns + new [] s
3190 246311 reg/reg cmpxchg doesn't work on amd64
3191 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
3192 246888 Improve Makefile.vex.am
3193 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
3194 to [f]chmod_extended
3195 247526 IBM POWER6 (ISA 2.05) support is incomplete
3196 247561 Some leak testcases fails due to reachable addresses in
3198 247875 sizeofIRType to handle Ity_I128
3199 247894 [PATCH] unhandled syscall sys_readahead
3200 247980 Doesn't honor CFLAGS passed to configure
3201 248373 darwin10.supp is empty in the trunk
3202 248822 Linux FIBMAP ioctl has int parameter instead of long
3203 248893 [PATCH] make readdwarf.c big endianess safe to enable
3204 unwinding on big endian systems
3205 249224 Syscall 336 not supported (SYS_proc_info)
3207 249775 Incorrect scheme for detecting NEON capabilities of host CPU
3208 249943 jni JVM init fails when using valgrind
3209 249991 Valgrind incorrectly declares AESKEYGENASSIST support
3211 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
3212 250799 frexp$fenv_access_off function generates SIGILL
3213 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
3214 251251 support pclmulqdq insn
3215 251362 valgrind: ARM: attach to debugger either fails or provokes
3217 251674 Unhandled syscall 294
3220 254257 Add support for debugfiles found by build-id
3221 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
3222 254646 Wrapped functions cause stack misalignment on OS X
3223 (and possibly Linux)
3224 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
3226 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
3230 Release 3.5.0 (19 August 2009)
3231 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3232 3.5.0 is a feature release with many significant improvements and the
3233 usual collection of bug fixes. The main improvement is that Valgrind
3234 now works on Mac OS X.
3236 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
3237 and X86/Darwin. Support for recent distros and toolchain components
3238 (glibc 2.10, gcc 4.5) has been added.
3240 -------------------------
3242 Here is a short summary of the changes. Details are shown further
3245 * Support for Mac OS X (10.5.x).
3247 * Improvements and simplifications to Memcheck's leak checker.
3249 * Clarification and simplifications in various aspects of Valgrind's
3252 * XML output for Helgrind and Ptrcheck.
3254 * Performance and stability improvements for Helgrind and DRD.
3256 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
3258 * A new experimental tool, BBV, useful for computer architecture
3261 * Improved Wine support, including ability to read Windows PDB
3264 -------------------------
3266 Here are details of the above changes, followed by descriptions of
3267 many other minor changes, and a list of fixed bugs.
3270 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
3271 called "Darwin" because that is the name of the OS core, which is the
3272 level that Valgrind works at.)
3276 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
3277 because it would require work and 10.4 is only becoming less common.
3279 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
3280 fairly well. For 10.5.x, 32-bit programs are the default even on
3281 64-bit machines, so it handles most current programs.
3283 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
3284 officially supported, but simple programs at least will probably work.
3285 However, start-up is slow.
3287 - PowerPC machines are not supported.
3289 Things that don't work:
3291 - The Ptrcheck tool.
3293 - Objective-C garbage collection.
3297 - If you have Rogue Amoeba's "Instant Hijack" program installed,
3298 Valgrind will fail with a SIGTRAP at start-up. See
3299 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
3304 - You will likely find --dsymutil=yes a useful option, as error
3305 messages may be imprecise without it.
3307 - Mac OS X support is new and therefore will be less robust than the
3308 Linux support. Please report any bugs you find.
3310 - Threaded programs may run more slowly than on Linux.
3312 Many thanks to Greg Parker for developing this port over several years.
3315 * Memcheck's leak checker has been improved.
3317 - The results for --leak-check=summary now match the summary results
3318 for --leak-check=full. Previously they could differ because
3319 --leak-check=summary counted "indirectly lost" blocks and
3320 "suppressed" blocks as "definitely lost".
3322 - Blocks that are only reachable via at least one interior-pointer,
3323 but are directly pointed to by a start-pointer, were previously
3324 marked as "still reachable". They are now correctly marked as
3327 - The default value for the --leak-resolution option has been
3328 changed from "low" to "high". In general, this means that more
3329 leak reports will be produced, but each leak report will describe
3330 fewer leaked blocks.
3332 - With --leak-check=full, "definitely lost" and "possibly lost"
3333 leaks are now considered as proper errors, ie. they are counted
3334 for the "ERROR SUMMARY" and affect the behaviour of
3335 --error-exitcode. These leaks are not counted as errors if
3336 --leak-check=summary is specified, however.
3338 - Documentation for the leak checker has been improved.
3341 * Various aspects of Valgrind's text output have changed.
3343 - Valgrind's start-up message has changed. It is shorter but also
3344 includes the command being run, which makes it easier to use
3345 --trace-children=yes. An example:
3347 - Valgrind's shut-down messages have also changed. This is most
3348 noticeable with Memcheck, where the leak summary now occurs before
3349 the error summary. This change was necessary to allow leaks to be
3350 counted as proper errors (see the description of the leak checker
3351 changes above for more details). This was also necessary to fix a
3352 longstanding bug in which uses of suppressions against leaks were
3353 not "counted", leading to difficulties in maintaining suppression
3354 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3356 - Behavior of -v has changed. In previous versions, -v printed out
3357 a mixture of marginally-user-useful information, and tool/core
3358 statistics. The statistics printing has now been moved to its own
3359 flag, --stats=yes. This means -v is less verbose and more likely
3360 to convey useful end-user information.
3362 - The format of some (non-XML) stack trace entries has changed a
3363 little. Previously there were six possible forms:
3365 0x80483BF: really (a.c:20)
3366 0x80483BF: really (in /foo/a.out)
3368 0x80483BF: (within /foo/a.out)
3369 0x80483BF: ??? (a.c:20)
3372 The third and fourth of these forms have been made more consistent
3373 with the others. The six possible forms are now:
3375 0x80483BF: really (a.c:20)
3376 0x80483BF: really (in /foo/a.out)
3377 0x80483BF: really (in ???)
3378 0x80483BF: ??? (in /foo/a.out)
3379 0x80483BF: ??? (a.c:20)
3382 Stack traces produced when --xml=yes is specified are different
3386 * Helgrind and Ptrcheck now support XML output, so they can be used
3387 from GUI tools. Also, the XML output mechanism has been
3390 - The XML format has been overhauled and generalised, so it is more
3391 suitable for error reporting tools in general. The Memcheck
3392 specific aspects of it have been removed. The new format, which
3393 is an evolution of the old format, is described in
3394 docs/internals/xml-output-protocol4.txt.
3396 - Memcheck has been updated to use the new format.
3398 - Helgrind and Ptrcheck are now able to emit output in this format.
3400 - The XML output mechanism has been overhauled. XML is now output
3401 to its own file descriptor, which means that:
3403 * Valgrind can output text and XML independently.
3405 * The longstanding problem of XML output being corrupted by
3406 unexpected un-tagged text messages is solved.
3408 As before, the destination for text output is specified using
3409 --log-file=, --log-fd= or --log-socket=.
3411 As before, XML output for a tool is enabled using --xml=yes.
3413 Because there's a new XML output channel, the XML output
3414 destination is now specified by --xml-file=, --xml-fd= or
3417 Initial feedback has shown this causes some confusion. To
3418 clarify, the two envisaged usage scenarios are:
3420 (1) Normal text output. In this case, do not specify --xml=yes
3421 nor any of --xml-file=, --xml-fd= or --xml-socket=.
3423 (2) XML output. In this case, specify --xml=yes, and one of
3424 --xml-file=, --xml-fd= or --xml-socket= to select the XML
3425 destination, one of --log-file=, --log-fd= or --log-socket=
3426 to select the destination for any remaining text messages,
3427 and, importantly, -q.
3429 -q makes Valgrind completely silent on the text channel,
3430 except in the case of critical failures, such as Valgrind
3431 itself segfaulting, or failing to read debugging information.
3432 Hence, in this scenario, it suffices to check whether or not
3433 any output appeared on the text channel. If yes, then it is
3434 likely to be a critical error which should be brought to the
3435 attention of the user. If no (the text channel produced no
3436 output) then it can be assumed that the run was successful.
3438 This allows GUIs to make the critical distinction they need to
3439 make (did the run fail or not?) without having to search or
3440 filter the text output channel in any way.
3442 It is also recommended to use --child-silent-after-fork=yes in
3446 * Improvements and changes in Helgrind:
3448 - XML output, as described above
3450 - Checks for consistent association between pthread condition
3451 variables and their associated mutexes are now performed.
3453 - pthread_spinlock functions are supported.
3455 - Modest performance improvements.
3457 - Initial (skeletal) support for describing the behaviour of
3458 non-POSIX synchronisation objects through ThreadSanitizer
3459 compatible ANNOTATE_* macros.
3461 - More controllable tradeoffs between performance and the level of
3462 detail of "previous" accesses in a race. There are now three
3465 * --history-level=full. This is the default, and was also the
3466 default in 3.4.x. It shows both stacks involved in a race, but
3467 requires a lot of memory and can be very slow in programs that
3468 do many inter-thread synchronisation events.
3470 * --history-level=none. This only shows the later stack involved
3471 in a race. This can be much faster than --history-level=full,
3472 but makes it much more difficult to find the other access
3473 involved in the race.
3475 The new intermediate setting is
3477 * --history-level=approx
3479 For the earlier (other) access, two stacks are presented. The
3480 earlier access is guaranteed to be somewhere in between the two
3481 program points denoted by those stacks. This is not as useful
3482 as showing the exact stack for the previous access (as per
3483 --history-level=full), but it is better than nothing, and it's
3484 almost as fast as --history-level=none.
3487 * New features and improvements in DRD:
3489 - The error messages printed by DRD are now easier to interpret.
3490 Instead of using two different numbers to identify each thread
3491 (Valgrind thread ID and DRD thread ID), DRD does now identify
3492 threads via a single number (the DRD thread ID). Furthermore
3493 "first observed at" information is now printed for all error
3494 messages related to synchronization objects.
3496 - Added support for named semaphores (sem_open() and sem_close()).
3498 - Race conditions between pthread_barrier_wait() and
3499 pthread_barrier_destroy() calls are now reported.
3501 - Added support for custom allocators through the macros
3502 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
3503 in <valgrind/valgrind.h>). An alternative for these two macros is
3504 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
3507 - Added support for annotating non-POSIX synchronization objects
3508 through several new ANNOTATE_*() macros.
3510 - OpenMP: added support for the OpenMP runtime (libgomp) included
3511 with gcc versions 4.3.0 and 4.4.0.
3515 - Added two new command-line options (--first-race-only and
3516 --segment-merging-interval).
3519 * Genuinely atomic support for x86/amd64/ppc atomic instructions
3521 Valgrind will now preserve (memory-access) atomicity of LOCK-
3522 prefixed x86/amd64 instructions, and any others implying a global
3523 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
3525 This means that Valgrinded processes will "play nicely" in
3526 situations where communication with other processes, or the kernel,
3527 is done through shared memory and coordinated with such atomic
3528 instructions. Prior to this change, such arrangements usually
3529 resulted in hangs, races or other synchronisation failures, because
3530 Valgrind did not honour atomicity of such instructions.
3533 * A new experimental tool, BBV, has been added. BBV generates basic
3534 block vectors for use with the SimPoint analysis tool, which allows
3535 a program's overall behaviour to be approximated by running only a
3536 fraction of it. This is useful for computer architecture
3537 researchers. You can run BBV by specifying --tool=exp-bbv (the
3538 "exp-" prefix is short for "experimental"). BBV was written by
3542 * Modestly improved support for running Windows applications under
3543 Wine. In particular, initial support for reading Windows .PDB debug
3544 information has been added.
3547 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
3548 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
3552 * The Valgrind client requests VALGRIND_PRINTF and
3553 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
3554 the string was always printed immediately on its own line. Now, the
3555 string will be added to a buffer but not printed until a newline is
3556 encountered, or other Valgrind output is printed (note that for
3557 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
3558 "other Valgrind output"). This allows you to use multiple
3559 VALGRIND_PRINTF calls to build up a single output line, and also to
3560 print multiple output lines with a single request (by embedding
3561 multiple newlines in the string).
3564 * The graphs drawn by Massif's ms_print program have changed slightly:
3566 - The half-height chars '.' and ',' are no longer drawn, because
3567 they are confusing. The --y option can be used if the default
3568 y-resolution is not high enough.
3570 - Horizontal lines are now drawn after the top of a snapshot if
3571 there is a gap until the next snapshot. This makes it clear that
3572 the memory usage has not dropped to zero between snapshots.
3575 * Something that happened in 3.4.0, but wasn't clearly announced: the
3576 option --read-var-info=yes can be used by some tools (Memcheck,
3577 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
3578 variable type and location information. This makes those tools
3579 start up more slowly and increases memory consumption, but
3580 descriptions of data addresses in error messages become more
3584 * exp-Omega, an experimental instantaneous leak-detecting tool, was
3585 disabled in 3.4.0 due to a lack of interest and maintenance,
3586 although the source code was still in the distribution. The source
3587 code has now been removed from the distribution. For anyone
3588 interested, the removal occurred in SVN revision r10247.
3591 * Some changes have been made to the build system.
3593 - VEX/ is now integrated properly into the build system. This means
3594 that dependency tracking within VEX/ now works properly, "make
3595 install" will work without requiring "make" before it, and
3596 parallel builds (ie. 'make -j') now work (previously a
3597 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
3598 was effectively ignored).
3600 - The --with-vex configure option has been removed. It was of
3601 little use and removing it simplified the build system.
3603 - The location of some install files has changed. This should not
3604 affect most users. Those who might be affected:
3606 * For people who use Valgrind with MPI programs, the installed
3607 libmpiwrap.so library has moved from
3608 $(INSTALL)/<platform>/libmpiwrap.so to
3609 $(INSTALL)/libmpiwrap-<platform>.so.
3611 * For people who distribute standalone Valgrind tools, the
3612 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
3613 have moved to $(INSTALL)/libcoregrind-<platform>.a.
3615 These changes simplify the build system.
3617 - Previously, all the distributed suppression (*.supp) files were
3618 installed. Now, only default.supp is installed. This should not
3619 affect users as the other installed suppression files were not
3620 read; the fact that they were installed was a mistake.
3623 * KNOWN LIMITATIONS:
3625 - Memcheck is unusable with the Intel compiler suite version 11.1,
3626 when it generates code for SSE2-and-above capable targets. This
3627 is because of icc's use of highly optimised inlined strlen
3628 implementations. It causes Memcheck to report huge numbers of
3629 false errors even in simple programs. Helgrind and DRD may also
3632 Versions 11.0 and earlier may be OK, but this has not been
3636 The following bugs have been fixed or resolved. Note that "n-i-bz"
3637 stands for "not in bugzilla" -- that is, a bug that was reported to us
3638 but never got a bugzilla entry. We encourage you to file bugs in
3639 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3640 mailing the developers (or mailing lists) directly -- bugs that are
3641 not entered into bugzilla tend to get forgotten about or ignored.
3643 To see details of a given bug, visit
3644 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3645 where XXXXXX is the bug number as listed below.
3647 84303 How about a LockCheck tool?
3648 91633 dereference of null ptr in vgPlain_st_basetype
3649 97452 Valgrind doesn't report any pthreads problems
3650 100628 leak-check gets assertion failure when using
3651 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
3652 108528 NPTL pthread cleanup handlers not called
3653 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
3654 110128 mallinfo is not implemented...
3655 110770 VEX: Generated files not always updated when making valgrind
3656 111102 Memcheck: problems with large (memory footprint) applications
3657 115673 Vex's decoder should never assert
3658 117564 False positive: Syscall param clone(child_tidptr) contains
3659 uninitialised byte(s)
3660 119404 executing ssh from inside valgrind fails
3661 133679 Callgrind does not write path names to sources with dwarf debug
3663 135847 configure.in problem with non gnu compilers (and possible fix)
3664 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
3665 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
3666 136230 memcheck reports "possibly lost", should be "still reachable"
3667 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
3668 137904 Valgrind reports a memory leak when using POSIX threads,
3670 139076 valgrind VT_GETSTATE error
3671 142228 complaint of elf_dynamic_do_rela in trivial usage
3672 145347 spurious warning with USBDEVFS_REAPURB
3673 148441 (wine) can't find memory leak in Wine, win32 binary
3675 148742 Leak-check fails assert on exit
3676 149878 add (proper) check for calloc integer overflow
3677 150606 Call graph is broken when using callgrind control
3678 152393 leak errors produce an exit code of 0. I need some way to
3679 cause leak errors to result in a nonzero exit code.
3680 157154 documentation (leak-resolution doc speaks about num-callers
3681 def=4) + what is a loss record
3682 159501 incorrect handling of ALSA ioctls
3683 162020 Valgrinding an empty/zero-byte file crashes valgrind
3684 162482 ppc: Valgrind crashes while reading stabs information
3685 162718 x86: avoid segment selector 0 in sys_set_thread_area()
3686 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
3687 163560 VEX/test_main.c is missing from valgrind-3.3.1
3688 164353 malloc_usable_size() doesn't return a usable size
3689 165468 Inconsistent formatting in memcheck manual -- please fix
3690 169505 main.c:286 (endOfInstr):
3691 Assertion 'ii->cost_offset == *cost_offset' failed
3692 177206 Generate default.supp during compile instead of configure
3693 177209 Configure valt_load_address based on arch+os
3694 177305 eventfd / syscall 323 patch lost
3695 179731 Tests fail to build because of inlining of non-local asm labels
3696 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
3697 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
3698 181594 Bogus warning for empty text segment
3699 181707 dwarf doesn't require enumerations to have name
3700 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3701 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3702 Assertion '!already_present' failed.
3703 185359 exp-ptrcheck: unhandled syscall getresuid()
3704 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
3705 185816 Valgrind is unable to handle debug info for files with split
3706 debug info that are prelinked afterwards
3707 185980 [darwin] unhandled syscall: sem_open
3708 186238 bbToIR_AMD64: disInstr miscalculated next %rip
3709 186507 exp-ptrcheck unhandled syscalls prctl, etc.
3710 186790 Suppression pattern used for leaks are not reported
3711 186796 Symbols with length>200 in suppression files are ignored
3712 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
3713 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
3714 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
3715 188046 bashisms in the configure script
3716 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
3717 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
3718 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
3719 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
3720 assertion fail "!lock->heldBy"
3721 188427 Add support for epoll_create1 (with patch)
3722 188530 Support for SIOCGSTAMPNS
3723 188560 Include valgrind.spec in the tarball
3724 188572 Valgrind on Mac should suppress setenv() mem leak
3725 189054 Valgrind fails to build because of duplicate non-local asm labels
3726 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
3727 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
3728 189763 drd assertion failure: s_threadinfo[tid].is_recording
3729 190219 unhandled syscall: 328 (x86-linux)
3730 190391 dup of 181394; see above
3731 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
3732 190820 No debug information on powerpc-linux
3733 191095 PATCH: Improve usbdevfs ioctl handling
3734 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
3736 191189 --xml=yes should obey --gen-suppressions=all
3737 191192 syslog() needs a suppression on macosx
3738 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
3739 191761 getrlimit on MacOSX
3740 191992 multiple --fn-skip only works sometimes; dependent on order
3741 192634 V. reports "aspacem sync_check_mapping_callback:
3742 segment mismatch" on Darwin
3743 192954 __extension__ missing on 2 client requests
3744 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
3745 194474 "INSTALL" file has different build instructions than "README"
3746 194671 Unhandled syscall (sem_wait?) from mac valgrind
3747 195069 memcheck: reports leak (memory still reachable) for
3749 195169 drd: (vgDrd_barrier_post_wait):
3750 Assertion 'r->sg[p->post_iteration]' failed.
3751 195268 valgrind --log-file doesn't accept ~/...
3752 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
3753 195860 WARNING: unhandled syscall: unix:223
3754 196528 need a error suppression for pthread_rwlock_init under os x?
3755 197227 Support aio_* syscalls on Darwin
3756 197456 valgrind should reject --suppressions=(directory)
3757 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
3758 197591 unhandled syscall 27 (mincore)
3759 197793 Merge DCAS branch to the trunk == 85756, 142103
3760 197794 Avoid duplicate filenames in Vex
3761 197898 make check fails on current SVN
3762 197901 make check fails also under exp-ptrcheck in current SVN
3763 197929 Make --leak-resolution=high the default
3764 197930 Reduce spacing between leak reports
3765 197933 Print command line of client at start-up, and shorten preamble
3766 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
3767 198395 add BBV to the distribution as an experimental tool
3768 198624 Missing syscalls on Darwin: 82, 167, 281, 347
3769 198649 callgrind_annotate doesn't cumulate counters
3770 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
3771 199977 Valgrind complains about an unrecognized instruction in the
3772 atomic_incs test program
3773 200029 valgrind isn't able to read Fedora 12 debuginfo
3774 200760 darwin unhandled syscall: unix:284
3775 200827 DRD doesn't work on Mac OS X
3776 200990 VG_(read_millisecond_timer)() does not work correctly
3777 201016 Valgrind does not support pthread_kill() on Mac OS
3778 201169 Document --read-var-info
3779 201323 Pre-3.5.0 performance sanity checking
3780 201384 Review user manual for the 3.5.0 release
3781 201585 mfpvr not implemented on ppc
3782 201708 tests failing because x86 direction flag is left set
3783 201757 Valgrind doesn't handle any recent sys_futex additions
3784 204377 64-bit valgrind can not start a shell script
3785 (with #!/path/to/shell) if the shell is a 32-bit executable
3786 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
3787 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
3788 about memory allocation events with memory access tracing enabled
3789 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
3791 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
3795 Release 3.4.1 (28 February 2009)
3796 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3797 3.4.1 is a bug-fix release that fixes some regressions and assertion
3798 failures in debug info reading in 3.4.0, most notably incorrect stack
3799 traces on amd64-linux on older (glibc-2.3 based) systems. Various
3800 other debug info problems are also fixed. A number of bugs in the
3801 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
3803 In view of the fact that 3.4.0 contains user-visible regressions
3804 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
3805 encouraged to ship 3.4.1 in preference to 3.4.0.
3807 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
3808 bugzilla" -- that is, a bug that was reported to us but never got a
3809 bugzilla entry. We encourage you to file bugs in bugzilla
3810 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
3811 developers (or mailing lists) directly -- bugs that are not entered
3812 into bugzilla tend to get forgotten about or ignored.
3814 n-i-bz Fix various bugs reading icc-11 generated debug info
3815 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
3816 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
3817 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
3818 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
3819 179618 exp-ptrcheck crashed / exit prematurely
3820 179624 helgrind: false positive races with pthread_create and
3821 recv/open/close/read
3822 134207 pkg-config output contains @VG_PLATFORM@
3823 176926 floating point exception at valgrind startup with PPC 440EPX
3824 181594 Bogus warning for empty text segment
3825 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
3826 181707 Dwarf3 doesn't require enumerations to have name
3827 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3828 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3829 Assertion '!already_present' failed.
3830 185359 exp-ptrcheck unhandled syscall getresuid()
3832 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
3833 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
3837 Release 3.4.0 (2 January 2009)
3838 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3839 3.4.0 is a feature release with many significant improvements and the
3840 usual collection of bug fixes. This release supports X86/Linux,
3841 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
3842 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
3844 3.4.0 brings some significant tool improvements. Memcheck can now
3845 report the origin of uninitialised values, the thread checkers
3846 Helgrind and DRD are much improved, and we have a new experimental
3847 tool, exp-Ptrcheck, which is able to detect overruns of stack and
3848 global arrays. In detail:
3850 * Memcheck is now able to track the origin of uninitialised values.
3851 When it reports an uninitialised value error, it will try to show
3852 the origin of the value, as either a heap or stack allocation.
3853 Origin tracking is expensive and so is not enabled by default. To
3854 use it, specify --track-origins=yes. Memcheck's speed will be
3855 essentially halved, and memory usage will be significantly
3856 increased. Nevertheless it can drastically reduce the effort
3857 required to identify the root cause of uninitialised value errors,
3858 and so is often a programmer productivity win, despite running more
3861 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
3862 3.4.0, will be released shortly.
3864 * Helgrind's race detection algorithm has been completely redesigned
3865 and reimplemented, to address usability and scalability concerns:
3867 - The new algorithm has a lower false-error rate: it is much less
3868 likely to report races that do not really exist.
3870 - Helgrind will display full call stacks for both accesses involved
3871 in a race. This makes it easier to identify the root causes of
3874 - Limitations on the size of program that can run have been removed.
3876 - Performance has been modestly improved, although that is very
3879 - Direct support for Qt4 threading has been added.
3881 - pthread_barriers are now directly supported.
3883 - Helgrind works well on all supported Linux targets.
3885 * The DRD thread debugging tool has seen major improvements:
3887 - Greatly improved performance and significantly reduced memory
3890 - Support for several major threading libraries (Boost.Thread, Qt4,
3891 glib, OpenMP) has been added.
3893 - Support for atomic instructions, POSIX semaphores, barriers and
3894 reader-writer locks has been added.
3896 - Works now on PowerPC CPUs too.
3898 - Added support for printing thread stack usage at thread exit time.
3900 - Added support for debugging lock contention.
3902 - Added a manual for Drd.
3904 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
3905 checks for misuses of pointers. In that sense it is a bit like
3906 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
3907 detect overruns of stack and global arrays, it can detect
3908 arbitrarily far out-of-bounds accesses to heap blocks, and it can
3909 detect accesses heap blocks that have been freed a very long time
3910 ago (millions of blocks in the past).
3912 Ptrcheck currently works only on x86-linux and amd64-linux. To use
3913 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
3914 of the main Valgrind documentation. As this is an experimental
3915 tool, we would be particularly interested in hearing about your
3916 experiences with it.
3918 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
3919 longer built by default, although the code remains in the repository
3920 and the tarball. This is due to three factors: a perceived lack of
3921 users, a lack of maintenance, and concerns that it may not be
3922 possible to achieve reliable operation using the existing design.
3924 * As usual, support for the latest Linux distros and toolchain
3925 components has been added. It should work well on Fedora Core 10,
3926 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
3927 state) is supported, as is glibc-2.9. The C++ demangler has been
3928 updated so as to work well with C++ compiled by even the most recent
3931 * You can now use frame-level wildcards in suppressions. This was a
3932 frequently-requested enhancement. A line "..." in a suppression now
3933 matches zero or more frames. This makes it easier to write
3934 suppressions which are precise yet insensitive to changes in
3937 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
3939 * Very basic support for IBM Power6 has been added (64-bit processes only).
3941 * Valgrind is now cross-compilable. For example, it is possible to
3942 cross compile Valgrind on an x86/amd64-linux host, so that it runs
3943 on a ppc32/64-linux target.
3945 * You can set the main thread's stack size at startup using the
3946 new --main-stacksize= flag (subject of course to ulimit settings).
3947 This is useful for running apps that need a lot of stack space.
3949 * The limitation that you can't use --trace-children=yes together
3950 with --db-attach=yes has been removed.
3952 * The following bugs have been fixed. Note that "n-i-bz" stands for
3953 "not in bugzilla" -- that is, a bug that was reported to us but
3954 never got a bugzilla entry. We encourage you to file bugs in
3955 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3956 mailing the developers (or mailing lists) directly.
3958 n-i-bz Make return types for some client requests 64-bit clean
3959 n-i-bz glibc 2.9 support
3960 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
3961 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
3962 n-i-bz Building in an env without gdb gives bogus gdb attach
3963 92456 Tracing the origin of uninitialised memory
3964 106497 Valgrind does not demangle some C++ template symbols
3966 151612 Suppression with "..." (frame-level wildcards in .supp files)
3967 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
3968 159285 unhandled syscall:25 (stime, on x86-linux)
3969 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
3970 160954 ppc build of valgrind crashes with illegal instruction (isel)
3971 160956 mallinfo implementation, w/ patch
3972 162092 Valgrind fails to start gnome-system-monitor
3973 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
3974 163794 assertion failure with "--track-origins=yes"
3975 163933 sigcontext.err and .trapno must be set together
3976 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
3977 164476 Missing kernel module loading system calls
3978 164669 SVN regression: mmap() drops posix file locks
3979 166581 Callgrind output corruption when program forks
3980 167288 Patch file for missing system calls on Cell BE
3981 168943 unsupported scas instruction pentium
3982 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
3983 172417 x86->IR: 0x82 ...
3984 172563 amd64->IR: 0xD9 0xF5 - fprem1
3985 173099 .lds linker script generation error
3986 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
3987 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
3989 174908 --log-file value not expanded correctly for core file
3990 175044 Add lookup_dcookie for amd64
3991 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
3993 Developer-visible changes:
3995 * Valgrind's debug-info reading machinery has been majorly overhauled.
3996 It can now correctly establish the addresses for ELF data symbols,
3997 which is something that has never worked properly before now.
3999 Also, Valgrind can now read DWARF3 type and location information for
4000 stack and global variables. This makes it possible to use the
4001 framework to build tools that rely on knowing the type and locations
4002 of stack and global variables, for example exp-Ptrcheck.
4004 Reading of such information is disabled by default, because most
4005 tools don't need it, and because it is expensive in space and time.
4006 However, you can force Valgrind to read it, using the
4007 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
4008 make use of such information, if present, to provide source-level
4009 descriptions of data addresses in the error messages they create.
4011 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
4012 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).