2 Release 3.15.0 (12 April 2019)
3 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
5 3.15.0 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 * The XTree Massif output format now makes use of the information obtained
17 when specifying --read-inline-info=yes.
19 * amd64 (x86_64): the RDRAND and F16C insn set extensions are now supported.
21 * ==================== TOOL CHANGES ====================
25 - DHAT been thoroughly overhauled, improved, and given a GUI. As a result,
26 it has been promoted from an experimental tool to a regular tool. Run it
27 with --tool=dhat instead of --tool=exp-dhat.
29 - DHAT now prints only minimal data when the program ends, instead writing
30 the bulk of the profiling data to a file. As a result, the --show-top-n
31 and --sort-by options have been removed.
33 - Profile results can be viewed with the new viewer, dh_view.html. When
34 a run ends, a short message is printed, explaining how to view the result.
36 - See the documentation for more details.
40 - cg_annotate has a new option, --show-percs, which prints percentages next
45 - callgrind_annotate has a new option, --show-percs, which prints percentages
46 next to all event counts.
48 - callgrind_annotate now inserts commas in call counts, and
49 sort the caller/callee lists in the call tree.
53 - The default value for --read-inline-info is now "yes" on
54 Linux/Android/Solaris. It is still "no" on other OS.
58 - The option --xtree-leak=yes (to output leak result in xtree format)
59 automatically activates the option --show-leak-kinds=all, as xtree
60 visualisation tools such as kcachegrind can in any case select what kind
63 - There has been further work to avoid false positives. In particular,
64 integer equality on partially defined inputs (C == and !=) is now handled
67 * ==================== OTHER CHANGES ====================
69 * The new option --show-error-list=no|yes displays, at the end of the run, the
70 list of detected errors and the used suppressions. Prior to this change,
71 showing this information could only be done by specifying "-v -v", which is
72 also produced a lot of other possibly-non-useful messages. The option -s is
73 equivalent to --show-error-list=yes.
75 * ==================== FIXED BUGS ====================
77 The following bugs have been fixed or resolved. Note that "n-i-bz"
78 stands for "not in bugzilla" -- that is, a bug that was reported to us
79 but never got a bugzilla entry. We encourage you to file bugs in
80 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
81 than mailing the developers (or mailing lists) directly -- bugs that
82 are not entered into bugzilla tend to get forgotten about or ignored.
84 To see details of a given bug, visit
85 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
86 where XXXXXX is the bug number as listed below.
88 385411 s390x: z13 vector floating-point instructions not implemented
89 397187 z13 vector register support for vgdb gdbserver
90 398183 Vex errors with _mm256_shuffle_epi8/vpshufb
91 398870 Please add support for instruction vcvtps2ph
92 399287 amd64 front end: Illegal Instruction vcmptrueps
93 399301 Use inlined frames in Massif XTree output.
94 399322 Improve callgrind_annotate output
95 399444 VEX/priv/guest_s390_toIR.c:17407: (style) Mismatching assignment [..]
96 400164 helgrind test encounters mips x-compiler warnings and assembler error
97 400490 s390x: VRs allocated as if separate from FPRs
98 400491 s390x: Operand of LOCH treated as unsigned integer
99 400975 Compile error: error: '-mips64r2' conflicts with the other architecture
100 options, which specify a mips64 processor
101 401112 LLVM 5.0 generates comparison against partially initialized data
102 401277 More bugs in z13 support
103 401454 Add a --show-percs option to cg_annotate and callgrind_annotate.
104 401578 drd: crashes sometimes on fork()
105 401627 memcheck errors with glibc avx2 optimized wcsncmp
106 401822 none/tests/ppc64/jm-vmx fails and produces assembler warnings
107 401827 none/tests/ppc64/test_isa_2_06_part3 failure on ppc64le (xvrsqrtesp)
108 401828 none/tests/ppc64/test_isa_2_06_part1 failure on ppc64le (fcfids and
110 402006 mark helper regs defined in final_tidyup before freeres_wrapper call
111 402048 WARNING: unhandled ppc64[be|le]-linux syscall: 26 (ptrace)
112 402123 invalid assembler opcodes for mips32r2
113 402134 assertion fail in mc_translate.c (noteTmpUsesIn) Iex_VECRET on arm64
114 402327 Warning: DWARF2 CFI reader: unhandled DW_OP_ opcode 0x13 (DW_OP_drop)
115 402341 drd/tests/tsan_thread_wrappers_pthread.h:369: suspicious code ?
116 402351 mips64 libvexmultiarch_test fails on s390x
118 402395 coregrind/vgdb-invoker-solaris.c: 2 * poor error checking
119 402480 Do not use %rsp in clobber list
120 402481 vbit-test fails on x86 for Iop_CmpEQ64 iselInt64Expr Sar64
121 402515 Implement new option --show-error-list=no|yes / -s
122 402519 POWER 3.0 addex instruction incorrectly implemented
123 402781 Redo the cache used to process indirect branch targets
124 403123 vex amd64->IR:0xF3 0x48 0xF 0xAE 0xD3 (wrfsbase)
125 403552 s390x: wrong facility bit checked for vector facility
126 404054 memcheck powerpc subfe x, x, x initializes x to 0 or -1 based on CA
127 404638 Add VG_(replaceIndexXA)
128 404843 s390x: backtrace sometimes ends prematurely
129 404888 autotools cleanup series
130 405079 unhandled ppc64le-linux syscall: 131 (quotactl)
131 405182 Valgrind fails to build with Clang
132 405356 PPC64, xvcvsxdsp, xvcvuxdsp are supposed to write the 32-bit result to
133 the upper and lower 32-bits of the 64-bit result
134 405362 PPC64, vmsummbm instruction doesn't handle overflow case correctly
135 405363 PPC64, xvcvdpsxws, xvcvdpuxws, do not handle NaN arguments correctly.
136 405365 PPC64, function _get_maxmin_fp_NaN() doesn't handle QNaN, SNaN case
138 405403 s390x disassembler cannot be used on x86
139 405430 Use gcc -Wimplicit-fallthrough=2 by default if available
140 405458 MIPS mkFormVEC arguments swapped?
141 405716 drd: Fix an integer overflow in the stack margin calculation
142 405722 Support arm64 core dump
143 405733 PPC64, xvcvdpsp should write 32-bit result to upper and lower 32-bits
144 of the 64-bit destination field.
145 405734 PPC64, vrlwnm, vrlwmi, vrldrm, vrldmi do not work properly when me < mb
146 405782 "VEX temporary storage exhausted" when attempting to debug slic3r-pe
147 406198 none/tests/ppc64/test_isa_3_0_other test sporadically including CA
149 406352 cachegrind/callgrind fails ann tests because of missing a.c
150 406354 dhat is broken on x86 (32bit)
151 406355 mcsignopass, mcsigpass, mcbreak fail due to difference in gdb output
152 406360 memcheck/tests/libstdc++.supp needs more supression variants
154 n-i-bz add syswrap for PTRACE_GET|SET_THREAD_AREA on amd64.
155 n-i-bz Fix callgrind_annotate non deterministic order for equal total
156 n-i-bz callgrind_annotate --threshold=100 does not print all functions.
157 n-i-bz callgrind_annotate Use of uninitialized value in numeric gt (>)
158 n-i-bz amd64 (x86_64): RDRAND and F16C insn set extensions are supported
160 (3.15.0.RC1: 8 April 2019, git ce94d674de5b99df173aad4c3ee48fc2a92e5d9c)
164 Release 3.14.0 (9 October 2018)
165 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
167 3.14.0 is a feature release with many improvements and the usual collection of
170 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
171 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
172 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
173 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
174 support for X86/macOS 10.13, AMD64/macOS 10.13.
176 * ==================== CORE CHANGES ===================
178 * The new option --keep-debuginfo=no|yes (default no) can be used to retain
179 debug info for unloaded code. This allows saved stack traces (e.g. for
180 memory leaks) to include file/line info for code that has been dlclose'd (or
181 similar). See the user manual for more information and known limitations.
183 * Ability to specify suppressions based on source file name and line number.
185 * Majorly overhauled register allocator. No end-user changes, but the JIT
186 generates code a bit more quickly now.
188 * ================== PLATFORM CHANGES =================
190 * Preliminary support for macOS 10.13 has been added.
192 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
194 * mips: support for MIPS SIMD architecture (MSA) has been added.
196 * mips: support for MIPS N32 ABI has been added.
198 * s390: partial support for vector instructions (integer and string) has been
201 * ==================== TOOL CHANGES ====================
203 * Helgrind: Addition of a flag
204 --delta-stacktrace=no|yes [yes on linux amd64/x86]
205 which specifies how full history stack traces should be computed.
206 Setting this to =yes can speed up Helgrind by 25% when using
207 --history-level=full.
209 * Memcheck: reduced false positive rate for optimised code created by Clang 6
210 / LLVM 6 on x86, amd64 and arm64. In particular, Memcheck analyses code
211 blocks more carefully to determine where it can avoid expensive definedness
212 checks without loss of precision. This is controlled by the flag
213 --expensive-definedness-checks=no|auto|yes [auto].
215 * ==================== OTHER CHANGES ====================
217 * Valgrind is now buildable with link-time optimisation (LTO). A new
218 configure option --enable-lto=yes allows building Valgrind with LTO. If the
219 toolchain supports it, this produces a smaller/faster Valgrind (up to 10%).
220 Note that if you are doing Valgrind development, --enable-lto=yes massively
221 slows down the build process.
223 * ==================== FIXED BUGS ====================
225 The following bugs have been fixed or resolved. Note that "n-i-bz"
226 stands for "not in bugzilla" -- that is, a bug that was reported to us
227 but never got a bugzilla entry. We encourage you to file bugs in
228 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
229 than mailing the developers (or mailing lists) directly -- bugs that
230 are not entered into bugzilla tend to get forgotten about or ignored.
232 To see details of a given bug, visit
233 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
234 where XXXXXX is the bug number as listed below.
236 79362 Debug info is lost for .so files when they are dlclose'd
237 208052 strlcpy error when n = 0
238 255603 exp-sgcheck Assertion '!already_present' failed
239 338252 building valgrind with -flto (link time optimisation) fails
240 345763 MIPS N32 ABI support
241 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
242 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
243 372347 Replacement problem of the additional c++14/c++17 new/delete operators
244 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
245 376257 helgrind history full speed up using a cached stack
246 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
248 379748 Fix missing pselect syscall (OS X 10.11)
249 379754 Fix missing syscall ulock_wait (OS X 10.12)
250 380397 s390x: __GI_strcspn() replacemenet needed
251 381162 possible array overrun in VEX register allocator
252 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
253 381274 powerpc too chatty even with --sigill-diagnostics=no
254 381289 epoll_pwait can have a NULL sigmask
255 381553 VEX register allocator v3
256 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
257 381769 Use ucontext_t instead of struct ucontext
258 381805 arm32 needs ld.so index hardwire for new glibc security fixes
259 382256 gz compiler flag test doesn't work for gold
260 382407 vg_perf needs "--terse" command line option
261 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
262 382563 MIPS MSA ASE support
263 382998 xml-socket doesn't work
264 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
265 383723 Fix missing kevent_qos syscall (macOS 10.11)
266 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
267 384096 Mention AddrCheck at Memcheck's command line option [..]
268 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
269 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
270 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
271 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
272 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
273 384337 performance improvements to VEX register allocator v2 and v3
274 384526 reduce number of spill insns generated by VEX register allocator v3
275 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
276 384631 Sanitise client args as printed with -v
277 384633 Add a simple progress-reporting facility
278 384987 VEX regalloc: allocate caller-save registers for short lived vregs
279 385055 PPC VEX temporary storage exhausted
280 385182 PPC64 is missing support for the DSCR
281 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
282 385207 PPC64, generate_store_FPRF() generates too many Iops
283 385208 PPC64, xxperm instruction exhausts temporary memory
284 385210 PPC64, vpermr instruction could exhaust temporary memory
285 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
286 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
287 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
288 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
289 385408 s390x: z13 vector "support" instructions not implemented
290 385409 s390x: z13 vector integer instructions not implemented
291 385410 s390x: z13 vector string instructions not implemented
292 385412 s390x: new non-vector z13 instructions not implemented
293 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
294 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
295 385939 Optionally exit on the first error
296 386318 valgrind.org/info/tools.html is missing SGCheck
297 386425 running valgrind + wine on armv7l gives illegal opcode
298 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
299 387410 MIPSr6 support
300 387664 Memcheck: make expensive-definedness-checks be the default
301 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
302 387766 asm shifts cause false positive "Conditional jump or move depends
303 on uninitialised value"
304 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
305 388174 valgrind with Wine quits with "Assertion 'cfsi_fits' failed"
306 388786 Support bpf syscall in amd64 Linux
307 388862 Add replacements for wmemchr and wcsnlen on Linux
308 389065 valgrind meets gcc flag -Wlogical-op
309 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
310 390471 suppression by specification of source-file line number
311 390723 make xtree dump files world wide readable, similar to log files
312 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
313 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
314 392118 unhandled amd64-linux syscall: 332 (statx)
315 392449 callgrind not clearing the number of calls properly
316 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
317 lxssp, stxssp and stxvl instructions.
318 393023 callgrind_control risks using the wrong vgdb
319 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
320 393099 posix_memalign() invalid write if alignment == 0
321 393146 failing assert "is_DebugInfo_active(di)"
322 395709 PPC64 is missing support for the xvnegsp instruction
323 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
325 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
326 395991 arm-linux: wine's unit tests enter a signal delivery loop [..]
327 396839 s390x: Trap instructions not implemented
328 396887 arch_prctl should return EINVAL on unknown option
329 == 397286 crash before launching binary (Unsupported arch_prctl option)
330 == 397393 valgrind: the 'impossible' happened: (Archlinux)
331 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
332 396906 compile tests failure on mips32-linux: broken inline asm in tests on
334 397012 glibc ld.so uses arch_prctl on i386
335 397089 amd64: Incorrect decoding of three-register vmovss/vmovsd opcode 11h
336 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
337 397424 glibc 2.27 and gdb_server tests
338 398028 Assertion `cfsi_fits` failing in simple C program
339 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
341 n-i-bz Fix missing workq_ops operations (macOS)
342 n-i-bz fix bug in strspn replacement
343 n-i-bz Add support for the Linux BLKFLSBUF ioctl
344 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
345 n-i-bz Fix possible stack trashing by semctl syscall wrapping
346 n-i-bz Add support for the Linux membarrier() system call
347 n-i-bz x86 front end: recognise and handle UD2 correctly
348 n-i-bz Signal delivery for x86-linux: ensure that the stack pointer is
349 correctly aligned before entering the handler.
351 (3.14.0.RC1: 30 September 2018, git c2aeea2d28acb0639bcc8cc1e4ab115067db1eae)
352 (3.14.0.RC2: 3 October 2018, git 3e214c4858a6fdd5697e767543a0c19e30505582)
353 (3.14.0: 9 October 2018, git 353a3587bb0e2757411f9138f5e936728ed6cc4f)
357 Release 3.13.0 (15 June 2017)
358 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
360 3.13.0 is a feature release with many improvements and the usual collection of
363 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
364 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
365 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
366 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
368 * ==================== CORE CHANGES ===================
370 * The translation cache size has been increased to keep up with the demands of
371 large applications. The maximum number of sectors has increased from 24 to
372 48. The default number of sectors has increased from 16 to 32 on all
373 targets except Android, where the increase is from 6 to 12.
375 * The amount of memory that Valgrind can use has been increased from 64GB to
376 128GB. In particular this means your application can allocate up to about
377 60GB when running on Memcheck.
379 * Valgrind's default load address has been changed from 0x3800'0000 to
380 0x5800'0000, so as to make it possible to load larger executables. This
381 should make it possible to load executables of size at least 1200MB.
383 * A massive spaceleak caused by reading compressed debuginfo files has been
384 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
387 * The C++ demangler has been updated.
389 * Support for demangling Rust symbols has been added.
391 * A new representation of stack traces, the "XTree", has been added. An XTree
392 is a tree of stacktraces with data associated with the stacktraces. This is
393 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
394 consumption of your program. Reporting is controlled by the new options
395 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
397 A report can also be produced on demand using the gdbserver monitor command
398 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
399 format' and 'massif format. The existing visualisers for these formats (e.g.
400 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
401 analyse these reports.
403 Memcheck can also produce XTree leak reports using the Callgrind file
404 format. For more details, see the user manual.
406 * ================== PLATFORM CHANGES =================
408 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
410 * amd64: fixes for JIT failure problems on long AVX2 code blocks
412 * amd64 and x86: support for CET prefixes has been added
414 * arm32: a few missing ARMv8 instructions have been implemented
416 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
417 Store-Conditional instructions has been added. This is to deal with
418 processor implementations that implement the LL/SC specifications strictly
419 and as a result cause Valgrind to hang in certain situations. The
420 alternative implementation is automatically enabled at startup, as required.
421 You can use the option --sim-hints=fallback-llsc to force-enable it if you
424 * Support for OSX 10.12 has been improved.
426 * On Linux, clone handling has been improved to honour CLONE_VFORK that
427 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
428 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
429 CLONE_VM exact semantics will (still) not work.
431 * The TileGX/Linux port has been removed because it appears to be both unused
434 * ==================== TOOL CHANGES ====================
438 - Memcheck should give fewer false positives when running optimised
439 Clang/LLVM generated code.
441 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
443 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
444 to produce the end of execution leak report in a xtree callgrind format
447 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
448 the leak report in an xtree file.
452 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
454 - For some workloads (typically, for big applications), Massif memory
455 consumption and CPU consumption has decreased significantly.
459 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
461 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
462 for Ada gnat compiled applications.
464 * ==================== OTHER CHANGES ====================
466 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
467 append the inner guest stacktrace to the inner host stacktrace. This helps
468 to investigate the errors reported by the outer, when they are caused by the
469 inner guest program (such as an inner regtest). See README_DEVELOPERS for
472 * To allow fast detection of callgrind files by desktop environments and file
473 managers, the format was extended to have an optional first line that
474 uniquely identifies the format ("# callgrind format"). Callgrind creates
475 this line now, as does the new xtree functionality.
477 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
478 have a new %n format letter that is replaced by a sequence number.
480 * "--version -v" now shows the SVN revision numbers from which Valgrind was
483 * ==================== FIXED BUGS ====================
485 The following bugs have been fixed or resolved. Note that "n-i-bz"
486 stands for "not in bugzilla" -- that is, a bug that was reported to us
487 but never got a bugzilla entry. We encourage you to file bugs in
488 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
489 than mailing the developers (or mailing lists) directly -- bugs that
490 are not entered into bugzilla tend to get forgotten about or ignored.
492 To see details of a given bug, visit
493 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
494 where XXXXXX is the bug number as listed below.
496 162848 --log-file output isn't split when a program forks
497 340777 Illegal instruction on mips (ar71xx)
498 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
499 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
500 to a different stack.
501 344139 x86 stack-seg overrides, needed by the Wine people
502 344524 store conditional of guest applications always fail - observed on
504 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
505 352395 Please provide SVN revision info in --version -v
506 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
507 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
508 INVALID_POSIX_THREADID' failed
509 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
510 barrier implementation
511 358697 valgrind.h: Some code remains even when defining NVALGRIND
512 359202 Add musl libc configure/compile
513 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
514 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
515 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
516 362223 assertion failed when .valgrindrc is a directory instead of a file
517 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
518 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
519 368507 can't malloc chunks larger than about 34GB
520 368529 Android arm target link error, missing atexit and pthread_atfork
521 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
522 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
523 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
524 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
525 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
526 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
527 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
528 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
529 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
530 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
531 370028 Reduce the number of compiler warnings on MIPS platforms
532 370635 arm64 missing syscall getcpu
533 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
534 371227 Clean AArch64 syscall table
535 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
536 371471 Valgrind complains about non legit memory leaks on placement new (C++)
537 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
538 371503 disInstr(arm64): unhandled instruction 0xF89F0000
539 371869 support '%' in symbol Z-encoding
540 371916 execution tree xtree concept
541 372120 c++ demangler demangles symbols which are not c++
542 372185 Support of valgrind on ARMv8 with 32 bit executable
543 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
544 372195 Power PC, xxsel instruction is not always recognized.
545 372504 Hanging on exit_group
546 372600 process loops forever when fatal signals are arriving quickly
547 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
548 373046 Stacks registered by core are never deregistered
549 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
550 373086 Implement additional Xen hypercalls
551 373192 Calling posix_spawn in glibc 2.24 completely broken
552 373488 Support for fanotify API on ARM64 architecture
553 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
554 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
555 373938 const IRExpr arguments for matchIRExpr()
556 374719 some spelling fixes
557 374963 increase valgrind's load address to prevent mmap failure
558 375514 valgrind_get_tls_addr() does not work in case of static TLS
559 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
560 for ML_(find_rx_mapping)()
561 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
562 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
563 == 377159 "vex: the `impossible' happened" still present
564 == 375150 Assertion 'tres.status == VexTransOK' failed
565 == 378068 valgrind crashes on AVX2 function in FFmpeg
566 376142 Segfaults on MIPS Cavium Octeon boards
567 376279 disInstr(arm64): unhandled instruction 0xD50320FF
568 376455 Solaris: unhandled syscall lgrpsys(180)
569 376518 Solaris: unhandled fast trap getlgrp(6)
570 376611 ppc64 and arm64 don't know about prlimit64 syscall
571 376729 PPC64, remove R2 from the clobber list
573 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
574 to be wrongly marked as addressable
575 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
576 PIE enabled by default
577 377376 memcheck/tests/linux/getregset fails with glibc2.24
578 377427 PPC64, lxv instruction failing on odd destination register
579 377478 PPC64: ISA 3.0 setup fixes
580 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
581 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
582 and 2 args for FUTEX_TRYLOCK_PI
583 377717 Fix massive space leak when reading compressed debuginfo sections
584 377891 Update Xen 4.6 domctl wrappers
585 377930 fcntl syscall wrapper is missing flock structure check
586 378524 libvexmultiarch_test regression on s390x and ppc64
587 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
588 378673 Update libiberty demangler
589 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
590 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
591 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
592 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
593 (task_register_dyld_image_infos)
594 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
595 (task_register_dyld_shared_cache_image_info)
596 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
597 379473 MIPS: add support for rdhwr cycle counter register
598 379504 remove TileGX/Linux port
599 379525 Support more x86 nop opcodes
600 379838 disAMode(x86): not an addr!
601 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
602 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
603 379895 clock_gettime does not execute POST syscall wrapper
604 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
605 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
606 380200 xtree generated callgrind files refer to files without directory name
607 380202 Assertion failure for cache line size (cls == 64) on aarch64.
608 380397 s390x: __GI_strcspn() replacement needed
609 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
611 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
612 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
613 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
617 Release 3.12.0 (20 October 2016)
618 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
620 3.12.0 is a feature release with many improvements and the usual
621 collection of bug fixes.
623 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
624 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
625 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
626 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
627 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
628 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
630 * ================== PLATFORM CHANGES =================
632 * POWER: Support for ISA 3.0 has been added
634 * mips: support for O32 FPXX ABI has been added.
635 * mips: improved recognition of different processors
636 * mips: determination of page size now done at run time
638 * amd64: Partial support for AMD FMA4 instructions.
640 * arm, arm64: Support for v8 crypto and CRC instructions.
642 * Improvements and robustification of the Solaris port.
644 * Preliminary support for MacOS 10.12 (Sierra) has been added.
646 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
647 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
648 where possible. Valgrind's support for 32-bit x86 has stagnated in
649 recent years and has fallen far behind that for 64-bit x86
650 instructions. By contrast 64-bit x86 is well supported, up to and
653 * ==================== TOOL CHANGES ====================
657 - Added meta mempool support for describing a custom allocator which:
658 - Auto-frees all chunks assuming that destroying a pool destroys all
660 - Uses itself to allocate other memory blocks
662 - New flag --ignore-range-below-sp to ignore memory accesses below
663 the stack pointer, if you really have to. The related flag
664 --workaround-gcc296-bugs=yes is now deprecated. Use
665 --ignore-range-below-sp=1024-1 as a replacement.
669 - Improved thread startup time significantly on non-Linux platforms.
673 - Added collection of the metric "tot-blocks-allocd"
675 * ==================== OTHER CHANGES ====================
677 * Replacement/wrapping of malloc/new related functions is now done not just
678 for system libraries by default, but for any globally defined malloc/new
679 related function (both in shared libraries and statically linked alternative
680 malloc implementations). The dynamic (runtime) linker is excluded, though.
681 To only intercept malloc/new related functions in
682 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
683 "nouserintercepts" can be any non-existing library name).
684 This new functionality is not implemented for MacOS X.
686 * The maximum number of callers in a suppression entry is now equal to
687 the maximum size for --num-callers (500).
688 Note that --gen-suppressions=yes|all similarly generates suppressions
689 containing up to --num-callers frames.
691 * New and modified GDB server monitor features:
693 - Valgrind's gdbserver now accepts the command 'catch syscall'.
694 Note that you must have GDB >= 7.11 to use 'catch syscall' with
697 * New option --run-cxx-freeres=<yes|no> can be used to change whether
698 __gnu_cxx::__freeres() cleanup function is called or not. Default is
701 * Valgrind is able to read compressed debuginfo sections in two formats:
702 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
703 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
705 * Modest JIT-cost improvements: the cost of instrumenting code blocks
706 for the most common use case (x86_64-linux, Memcheck) has been
709 * Improved performance for programs that do a lot of discarding of
710 instruction address ranges of 8KB or less.
712 * The C++ symbol demangler has been updated.
714 * More robustness against invalid syscall parameters on Linux.
716 * ==================== FIXED BUGS ====================
718 The following bugs have been fixed or resolved. Note that "n-i-bz"
719 stands for "not in bugzilla" -- that is, a bug that was reported to us
720 but never got a bugzilla entry. We encourage you to file bugs in
721 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
722 than mailing the developers (or mailing lists) directly -- bugs that
723 are not entered into bugzilla tend to get forgotten about or ignored.
725 To see details of a given bug, visit
726 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
727 where XXXXXX is the bug number as listed below.
729 191069 Exiting due to signal not reported in XML output
730 199468 Suppressions: stack size limited to 25
731 while --num-callers allows more frames
732 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
733 278744 cvtps2pd with redundant RexW
734 303877 valgrind doesn't support compressed debuginfo sections.
735 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
736 348345 Assertion fails for negative lineno
737 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
738 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
739 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
740 351804 Crash on generating suppressions for "printf" call on OS X 10.10
741 352197 mips: mmap2() not wrapped correctly for page size > 4096
742 353083 arm64 doesn't implement various xattr system calls
743 353084 arm64 doesn't support sigpending system call
744 353137 www: update info for Supported Platforms
745 353138 www: update "The Valgrind Developers" page
746 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
749 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
750 353398 WARNING: unhandled amd64-solaris syscall: 207
751 353660 XML in auxwhat tag not escaping reserved symbols properly
752 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
753 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
754 353802 ELF debug info reader confused with multiple .rodata sections
755 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
756 353917 unhandled amd64-solaris syscall fchdir(120)
757 353920 unhandled amd64-solaris syscall: 170
758 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
759 354392 unhandled amd64-solaris syscall: 171
760 354797 Vbit test does not include Iops for Power 8 instruction support
761 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
765 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
766 355188 valgrind should intercept all malloc related global functions
767 355454 do not intercept malloc related symbols from the runtime linker
768 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
769 356044 Dwarf line info reader misinterprets is_stmt register
770 356112 mips: replace addi with addiu
771 356393 valgrind (vex) crashes because isZeroU happened
774 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
775 356678 arm64-linux: unhandled syscall 232 (mincore)
776 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
777 356823 Unsupported ARM instruction: stlex
778 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
779 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
780 357673 crash if I try to run valgrind with a binary link with libcurl
781 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
782 357871 pthread_spin_destroy not properly wrapped
783 357887 Calls to VG_(fclose) do not close the file descriptor
784 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
785 358030 support direct socket calls on x86 32bit (new in linux 4.3)
786 358478 drd/tests/std_thread.cpp doesn't build with GCC6
787 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
788 359181 Buffer Overflow during Demangling
789 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
790 359289 s390x: popcnt (B9E1) not implemented
791 359472 The Power PC vsubuqm instruction doesn't always give the correct result
792 359503 Add missing syscalls for aarch64 (arm64)
793 359645 "You need libc6-dbg" help message could be more helpful
794 359703 s390: wire up separate socketcalls system calls
795 359724 getsockname might crash - deref_UInt should call safe_to_deref
796 359733 amd64 implement ld.so strchr/index override like x86
797 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
798 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
800 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
801 359871 Incorrect mask handling in ppoll
802 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
803 360008 Contents of Power vr registers contents is not printed correctly when
804 the --vgdb-shadow-registers=yes option is used
805 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
807 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
808 360425 arm64 unsupported instruction ldpsw
810 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
811 360571 Error about the Android Runtime reading below the stack pointer on ARM
812 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
813 360749 kludge for multiple .rodata sections on Solaris no longer needed
814 360752 raise the number of reserved fds in m_main.c from 10 to 12
815 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
816 361226 s390x: risbgn (EC59) not implemented
817 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
818 361354 ppc64[le]: wire up separate socketcalls system calls
819 361615 Inconsistent termination for multithreaded process terminated by signal
820 361926 Unhandled Solaris syscall: sysfs(84)
821 362009 V dumps core on unimplemented functionality before threads are created
822 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
823 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
824 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
825 362953 Request for an update to the Valgrind Developers page
826 363680 add renameat2() support
827 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
828 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
829 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
830 364058 clarify in manual limitations of array overruns detections
831 364413 pselect sycallwrapper mishandles NULL sigmask
832 364728 Power PC, missing support for several HW registers in
833 get_otrack_shadow_offset_wrk()
834 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
835 365273 Invalid write to stack location reported after signal handler runs
836 365912 ppc64BE segfault during jm-insns test (RELRO)
837 366079 FPXX Support for MIPS32 Valgrind
838 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
839 366344 Multiple unhandled instruction for Aarch64
840 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
841 367995 Integration of memcheck with custom memory allocator
842 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
843 368412 False positive result for altivec capability check
844 368416 Add tc06_two_races_xml.exp output for ppc64
845 368419 Perf Events ioctls not implemented
846 368461 mmapunmap test fails on ppc64
847 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
848 369000 AMD64 fma4 instructions unsupported.
849 369169 ppc64 fails jm_int_isa_2_07 test
850 369175 jm_vec_isa_2_07 test crashes on ppc64
851 369209 valgrind loops and eats up all memory if cwd doesn't exist.
852 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
853 369359 msghdr_foreachfield can crash when handling bad iovec
854 369360 Bad sigprocmask old or new sets can crash valgrind
855 369361 vmsplice syscall wrapper crashes on bad iovec
856 369362 Bad sigaction arguments crash valgrind
857 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
858 369402 Bad set/get_thread_area pointer crashes valgrind
859 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
860 369446 valgrind crashes on unknown fcntl command
861 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
862 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
863 370265 ISA 3.0 HW cap stuff needs updating
864 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
865 372195 Power PC, xxsel instruction is not always recognized
867 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
868 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
869 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
870 n-i-bz Fix ppoll_alarm exclusion on OS X
871 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
872 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
873 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
874 n-i-bz amd64: memcheck false positive with shr %edx
875 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
876 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
877 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
878 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
880 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
881 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
882 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
886 Release 3.11.0 (22 September 2015)
887 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
889 3.11.0 is a feature release with many improvements and the usual
890 collection of bug fixes.
892 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
893 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
894 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
895 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
896 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
897 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
899 * ================== PLATFORM CHANGES =================
901 * Support for Solaris/x86 and Solaris/amd64 has been added.
903 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
905 * Preliminary support for the Tilera TileGX architecture has been added.
907 * s390x: It is now required for the host to have the "long displacement"
908 facility. The oldest supported machine model is z990.
910 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
911 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
913 * The JIT's register allocator is significantly faster, making the JIT
914 as a whole somewhat faster, so JIT-intensive activities, for example
915 program startup, are modestly faster, around 5%.
917 * There have been changes to the default settings of several command
918 line flags, as detailed below.
920 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
921 capable hosts, the simulated CPUID will now indicate AVX2 support.
923 * ==================== TOOL CHANGES ====================
927 - The default value for --leak-check-heuristics has been changed from
928 "none" to "all". This helps to reduce the number of possibly
929 lost blocks, in particular for C++ applications.
931 - The default value for --keep-stacktraces has been changed from
932 "malloc-then-free" to "malloc-and-free". This has a small cost in
933 memory (one word per malloc-ed block) but allows Memcheck to show the
934 3 stacktraces of a dangling reference: where the block was allocated,
935 where it was freed, and where it is acccessed after being freed.
937 - The default value for --partial-loads-ok has been changed from "no" to
938 "yes", so as to avoid false positive errors resulting from some kinds
941 - A new monitor command 'xb <addr> <len>' shows the validity bits of
942 <len> bytes at <addr>. The monitor command 'xb' is easier to use
943 than get_vbits when you need to associate byte data value with
944 their corresponding validity bits.
946 - The 'block_list' monitor command has been enhanced:
947 o it can print a range of loss records
948 o it now accepts an optional argument 'limited <max_blocks>'
949 to control the number of blocks printed.
950 o if a block has been found using a heuristic, then
951 'block_list' now shows the heuristic after the block size.
952 o the loss records/blocks to print can be limited to the blocks
953 found via specified heuristics.
955 - The C helper functions used to instrument loads on
956 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
957 replaced by handwritten assembly sequences. This gives speedups
958 in the region of 0% to 7% for those targets only.
960 - A new command line option, --expensive-definedness-checks=yes|no,
961 has been added. This is useful for avoiding occasional invalid
962 uninitialised-value errors in optimised code. Watch out for
963 runtime degradation, as this can be up to 25%. As always, though,
964 the slowdown is highly application specific. The default setting
969 - A new monitor command 'all_snapshots <filename>' dumps all
970 snapshots taken so far.
974 - Significant memory reduction and moderate speedups for
975 --history-level=full for applications accessing a lot of memory
976 with many different stacktraces.
978 - The default value for --conflict-cache-size=N has been doubled to
979 2000000. Users that were not using the default value should
980 preferably also double the value they give.
982 The default was changed due to the changes in the "full history"
983 implementation. Doubling the value gives on average a slightly more
984 complete history and uses similar memory (or significantly less memory
985 in the worst case) than the previous implementation.
987 - The Helgrind monitor command 'info locks' now accepts an optional
988 argument 'lock_addr', which shows information about the lock at the
991 - When using --history-level=full, the new Helgrind monitor command
992 'accesshistory <addr> [<len>]' will show the recorded accesses for
993 <len> (or 1) bytes at <addr>.
995 * ==================== OTHER CHANGES ====================
997 * The default value for the --smc-check option has been changed from
998 "stack" to "all-non-file" on targets that provide automatic D-I
999 cache coherence (x86, amd64 and s390x). The result is to provide,
1000 by default, transparent support for JIT generated and self-modifying
1001 code on all targets.
1003 * Mac OS X only: the default value for the --dsymutil option has been
1004 changed from "no" to "yes", since any serious usage on Mac OS X
1005 always required it to be "yes".
1007 * The command line options --db-attach and --db-command have been removed.
1008 They were deprecated in 3.10.0.
1010 * When a process dies due to a signal, Valgrind now shows the signal
1011 and the stacktrace at default verbosity (i.e. verbosity 1).
1013 * The address description logic used by Memcheck and Helgrind now
1014 describes addresses in anonymous segments, file mmap-ed segments,
1015 shared memory segments and the brk data segment.
1017 * The new option --error-markers=<begin>,<end> can be used to mark the
1018 begin/end of errors in textual output mode, to facilitate
1019 searching/extracting errors in output files that mix valgrind errors
1020 with program output.
1022 * The new option --max-threads=<number> can be used to change the number
1023 of threads valgrind can handle. The default is 500 threads which
1024 should be more than enough for most applications.
1026 * The new option --valgrind-stacksize=<number> can be used to change the
1027 size of the private thread stacks used by Valgrind. This is useful
1028 for reducing memory use or increasing the stack size if Valgrind
1029 segfaults due to stack overflow.
1031 * The new option --avg-transtab-entry-size=<number> can be used to specify
1032 the expected instrumented block size, either to reduce memory use or
1033 to avoid excessive retranslation.
1035 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
1037 * New and modified GDB server monitor features:
1039 - When a signal is reported in GDB, you can now use the GDB convenience
1040 variable $_siginfo to examine detailed signal information.
1042 - Valgrind's gdbserver now allows the user to change the signal
1043 to deliver to the process. So, use 'signal SIGNAL' to continue execution
1044 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
1045 continue without passing the signal to the process.
1047 - With GDB >= 7.10, the command 'target remote'
1048 will automatically load the executable file of the process running
1049 under Valgrind. This means you do not need to specify the executable
1050 file yourself, GDB will discover it itself. See GDB documentation about
1051 'qXfer:exec-file:read' packet for more info.
1053 * ==================== FIXED BUGS ====================
1055 The following bugs have been fixed or resolved. Note that "n-i-bz"
1056 stands for "not in bugzilla" -- that is, a bug that was reported to us
1057 but never got a bugzilla entry. We encourage you to file bugs in
1058 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1059 than mailing the developers (or mailing lists) directly -- bugs that
1060 are not entered into bugzilla tend to get forgotten about or ignored.
1062 To see details of a given bug, visit
1063 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1064 where XXXXXX is the bug number as listed below.
1066 116002 VG_(printf): Problems with justification of strings and integers
1067 155125 avoid cutting away file:lineno after long function name
1068 197259 Unsupported arch_prtctl PR_SET_GS option
1069 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
1070 201216 Fix Valgrind does not support pthread_sigmask() on OS X
1071 201435 Fix Darwin: -v does not show kernel version
1072 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
1073 211256 Fixed an outdated comment regarding the default platform.
1074 211529 Incomplete call stacks for code compiled by newer versions of MSVC
1075 211926 Avoid compilation warnings in valgrind.h with -pedantic
1076 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
1078 226609 Crediting upstream authors in man page
1079 231257 Valgrind omits path when executing script from shebang line
1080 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
1081 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
1082 269360 s390x: Fix addressing mode selection for compare-and-swap
1083 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
1085 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
1086 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
1087 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
1088 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
1089 330147 libmpiwrap PMPI_Get_count returns undefined value
1090 333051 mmap of huge pages fails due to incorrect alignment
1092 334802 valgrind does not always explain why a given option is bad
1093 335618 mov.w rN, pc/sp (ARM32)
1094 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
1099 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
1100 338602 AVX2 bit in CPUID missing
1101 338606 Strange message for scripts with invalid interpreter
1102 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
1103 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
1104 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
1106 339156 gdbsrv not called for fatal signal
1107 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
1108 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
1109 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
1110 339442 Fix testsuite build failure on OS X 10.9
1111 339542 Enable compilation with Intel's ICC compiler
1112 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
1113 339688 Mac-specific ASM does not support .version directive (cpuid,
1114 tronical and pushfpopf tests)
1115 339745 Valgrind crash when check Marmalade app (partial fix)
1116 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
1117 339778 Linux/TileGx platform support for Valgrind
1118 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
1119 339789 Fix none/tests/execve test on Mac OS X 10.9
1120 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
1121 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
1122 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
1123 340392 Allow user to select more accurate definedness checking in memcheck
1124 to avoid invalid complaints on optimised code
1125 340430 Fix some grammatical weirdness in the manual.
1126 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
1127 341419 Signal handler ucontext_t not filled out correctly on OS X
1128 341539 VG_(describe_addr) should not describe address as belonging to client
1129 segment if it is past the heap end
1130 341613 Enable building of manythreads and thread-exits tests on Mac OS X
1131 341615 Fix none/tests/darwin/access_extended test on Mac OS X
1132 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
1133 341789 aarch64: shmat fails with valgrind on ARMv8
1134 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
1135 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
1136 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
1137 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
1138 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
1139 342221 socket connect false positive uninit memory for unknown af family
1140 342353 Allow dumping full massif output while valgrind is still running
1141 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
1145 342603 Add I2C_SMBUS ioctl support
1146 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
1147 342683 Mark memory past the initial brk limit as unaddressable
1148 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
1149 342795 Internal glibc __GI_mempcpy call should be intercepted
1150 342841 s390x: Support instructions fiebr(a) and fidbr(a)
1151 343012 Unhandled syscall 319 (memfd_create)
1152 343069 Patch updating v4l2 API support
1153 343173 helgrind crash during stack unwind
1154 343219 fix GET_STARTREGS for arm
1155 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
1156 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
1157 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
1158 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
1159 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
1160 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
1161 343597 ppc64le: incorrect use of offseof macro
1162 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
1163 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
1164 343732 Unhandled syscall 144 (setgid) on aarch64
1165 343733 Unhandled syscall 187 (msgctl and related) on aarch64
1166 343802 s390x: False positive "conditional jump or move depends on [..]
1167 343902 --vgdb=yes doesn't break when --xml=yes is used
1168 343967 Don't warn about setuid/setgid/setcap executable for directories
1169 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
1170 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
1171 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
1172 344054 www - update info for Solaris/illumos
1173 344416 'make regtest' does not work cleanly on OS X
1174 344235 Remove duplicate include of pub_core_aspacemgr.h
1175 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
1176 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
1177 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
1178 344314 callgrind_annotate ... warnings about commands containing newlines
1179 344318 socketcall should wrap recvmmsg and sendmmsg
1180 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
1181 344416 Fix 'make regtest' does not work cleanly on OS X
1182 344499 Fix compilation for Linux kernel >= 4.0.0
1183 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
1184 unix:349 (__pthread_fchdir)
1185 344559 Garbage collection of unused segment names in address space manager
1186 344560 Fix stack traces missing penultimate frame on OS X
1187 344621 Fix memcheck/tests/err_disable4 test on OS X
1188 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
1189 344702 Fix missing libobjc suppressions on OS X 10.10
1191 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
1192 344939 Fix memcheck/tests/xml1 on OS X 10.10
1193 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
1194 345079 Fix build problems in VEX/useful/test_main.c
1195 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
1196 345177 arm64: prfm (reg) not implemented
1197 345215 Performance improvements for the register allocator
1198 345248 add support for Solaris OS in valgrind
1199 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
1200 345394 Fix memcheck/tests/strchr on OS X
1201 345637 Fix memcheck/tests/sendmsg on OS X
1202 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
1203 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
1204 345887 Fix an assertion in the address space manager
1205 345928 amd64: callstack only contains current function for small stacks
1206 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1207 345987 MIPS64: Implement cavium LHX instruction
1208 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1209 346185 Fix typo saving altivec register v24
1210 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1211 and LibVEX_GuestPPC64_get_CR()
1212 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1213 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1214 346307 fuse filesystem syscall deadlocks
1215 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1216 346411 MIPS: SysRes::_valEx handling is incorrect
1217 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1218 346474 PPC64 Power 8, spr TEXASRU register not supported
1219 346487 Compiler generates "note" about a future ABI change for PPC64
1220 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1221 and causing spurious "invalid read of size 8" warnings
1222 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1223 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1224 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1225 347322 Power PC regression test cleanup
1226 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1228 347389 unhandled syscall: 373 (Linux ARM syncfs)
1229 347686 Patch set to cleanup PPC64 regtests
1230 347978 Remove bash dependencies where not needed
1231 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1232 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1234 348102 Patch updating v4l2 API support
1235 348247 amd64 front end: jno jumps wrongly when overflow is not set
1236 348269 Improve mmap MAP_HUGETLB support.
1237 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1238 348345 Assertion fails for negative lineno
1239 348377 Unsupported ARM instruction: yield
1240 348565 Fix detection of command line option availability for clang
1241 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1242 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1243 348748 Fix redundant condition
1244 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1245 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1246 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1247 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1248 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1249 349626 Implemented additional Xen hypercalls
1250 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1251 349790 Clean up of the hardware capability checking utilities.
1252 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1253 349874 Fix typos in source code
1254 349879 memcheck: add handwritten assembly for helperc_LOADV*
1255 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1256 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1257 350202 Add limited param to 'monitor block_list'
1258 350290 s390x: Support instructions fixbr(a)
1259 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1260 350809 Fix none/tests/async-sigs for Solaris
1261 350811 Remove reference to --db-attach which has been removed.
1262 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1263 350854 hard-to-understand code in VG_(load_ELF)()
1264 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1265 351386 Solaris: Cannot run ld.so.1 under Valgrind
1266 351474 Fix VG_(iseqsigset) as obvious
1267 351531 Typo in /include/vki/vki-xen-physdev.h header guard
1268 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
1269 351858 ldsoexec support on Solaris
1270 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
1271 352130 helgrind reports false races for printfs using mempcpy on FILE* state
1272 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
1273 352320 arm64 crash on none/tests/nestedfs
1274 352765 Vbit test fails on Power 6
1275 352768 The mbar instruction is missing from the Power PC support
1276 352769 Power PC program priority register (PPR) is not supported
1277 n-i-bz Provide implementations of certain compiler builtins to support
1278 compilers that may not provide those
1279 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
1280 n-i-bz Fix compilation on distros with glibc < 2.5
1281 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
1282 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
1283 n-i-bz mremap did not work properly on shared memory
1284 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
1285 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
1287 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
1288 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
1289 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
1293 Release 3.10.1 (25 November 2014)
1294 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1295 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
1296 and backports fixes for all reported missing AArch64 ARMv8 instructions
1297 and syscalls from the trunk. If you package or deliver 3.10.0 for others
1298 to use, you might want to consider upgrading to 3.10.1 instead.
1300 The following bugs have been fixed or resolved. Note that "n-i-bz"
1301 stands for "not in bugzilla" -- that is, a bug that was reported to us
1302 but never got a bugzilla entry. We encourage you to file bugs in
1303 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1304 than mailing the developers (or mailing lists) directly -- bugs that
1305 are not entered into bugzilla tend to get forgotten about or ignored.
1307 To see details of a given bug, visit
1308 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1309 where XXXXXX is the bug number as listed below.
1311 335440 arm64: ld1 (single structure) is not implemented
1312 335713 arm64: unhanded instruction: prfm (immediate)
1313 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
1314 339182 ppc64: AvSplat ought to load destination vector register with [..]
1315 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
1316 339433 ppc64 lxvw4x instruction uses four 32-byte loads
1317 339645 Use correct tag names in sys_getdents/64 wrappers
1318 339706 Fix false positive for ioctl(TIOCSIG) on linux
1319 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
1320 339853 arm64 times syscall unknown
1321 339855 arm64 unhandled getsid/setsid syscalls
1322 339858 arm64 dmb sy not implemented
1323 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
1324 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
1325 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
1327 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
1328 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
1329 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
1330 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
1331 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
1332 340509 arm64: unhandled instruction fcvtas
1333 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
1334 340632 arm64: unhandled instruction fcvtas
1335 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
1336 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
1337 340788 warning: unhandled syscall: 318 (getrandom)
1338 340807 disInstr(arm): unhandled instruction: 0xEE989B20
1339 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
1340 340922 arm64: unhandled getgroups/setgroups syscalls
1341 350251 Fix typo in VEX utility program (test_main.c).
1342 350407 arm64: unhandled instruction ucvtf (vector, integer)
1343 350809 none/tests/async-sigs breaks when run under cron on Solaris
1344 350811 update README.solaris after r15445
1345 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
1346 350854 strange code in VG_(load_ELF)()
1347 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1348 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
1349 n-i-bz Add missing ]] to terminate CDATA.
1350 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
1351 n-i-bz Enable sys_fadvise64_64 on arm32.
1352 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
1353 n-i-bz Add test cases for all known arm64 load/store instructions.
1354 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
1355 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
1356 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
1357 n-i-bz arm64: Implement "fcvtpu w, s".
1358 n-i-bz arm64: implement ADDP and various others
1359 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
1360 n-i-bz arm64: enable FCVT{A,N}S X,S.
1362 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
1366 Release 3.10.0 (10 September 2014)
1367 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1369 3.10.0 is a feature release with many improvements and the usual
1370 collection of bug fixes.
1372 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1373 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1374 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
1375 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
1376 significantly improved relative to the 3.9.0 release.
1378 * ================== PLATFORM CHANGES =================
1380 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
1381 is mostly complete, and is usable, but some SIMD instructions are as
1384 * Support for little-endian variant of the 64-bit POWER architecture.
1386 * Support for Android on MIPS32.
1388 * Support for 64bit FPU on MIPS32 platforms.
1390 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
1392 * Configuration for and running on Android targets has changed.
1393 See README.android in the source tree for details.
1395 * ================== DEPRECATED FEATURES =================
1397 * --db-attach is now deprecated and will be removed in the next
1398 valgrind feature release. The built-in GDB server capabilities are
1399 superior and should be used instead. Learn more here:
1400 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
1402 * ==================== TOOL CHANGES ====================
1406 - Client code can now selectively disable and re-enable reporting of
1407 invalid address errors in specific ranges using the new client
1408 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
1409 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
1411 - Leak checker: there is a new leak check heuristic called
1412 "length64". This is used to detect interior pointers pointing 8
1413 bytes inside a block, on the assumption that the first 8 bytes
1414 holds the value "block size - 8". This is used by
1415 sqlite3MemMalloc, for example.
1417 - Checking of system call parameters: if a syscall parameter
1418 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
1419 several fields not initialised, an error is now reported for each
1420 field. Previously, an error was reported only for the first
1421 uninitialised field.
1423 - Mismatched alloc/free checking: a new flag
1424 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
1425 such checks if necessary.
1429 - Improvements to error messages:
1431 o Race condition error message involving heap allocated blocks also
1432 show the thread number that allocated the raced-on block.
1434 o All locks referenced by an error message are now announced.
1435 Previously, some error messages only showed the lock addresses.
1437 o The message indicating where a lock was first observed now also
1438 describes the address/location of the lock.
1440 - Helgrind now understands the Ada task termination rules and
1441 creates a happens-before relationship between a terminated task
1442 and its master. This avoids some false positives and avoids a big
1443 memory leak when a lot of Ada tasks are created and terminated.
1444 The interceptions are only activated with forthcoming releases of
1445 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
1447 - A new GDB server monitor command "info locks" giving the list of
1448 locks, their location, and their status.
1452 - callgrind_control now supports the --vgdb-prefix argument,
1453 which is needed if valgrind was started with this same argument.
1455 * ==================== OTHER CHANGES ====================
1457 * Unwinding through inlined function calls. Stack unwinding can now
1458 make use of Dwarf3 inlined-unwind information if it is available.
1459 The practical effect is that inlined calls become visible in stack
1460 traces. The suppression matching machinery has been adjusted
1461 accordingly. This is controlled by the new option
1462 --read-inline-info=yes|no. Currently this is enabled by default
1463 only on Linux and Android targets and only for the tools Memcheck,
1466 * Valgrind can now read EXIDX unwind information on 32-bit ARM
1467 targets. If an object contains both CFI and EXIDX unwind
1468 information, Valgrind will prefer the CFI over the EXIDX. This
1469 facilitates unwinding through system libraries on arm-android
1472 * Address description logic has been improved and is now common
1473 between Memcheck and Helgrind, resulting in better address
1474 descriptions for some kinds of error messages.
1476 * Error messages about dubious arguments (eg, to malloc or calloc) are
1477 output like other errors. This means that they can be suppressed
1478 and they have a stack trace.
1480 * The C++ demangler has been updated for better C++11 support.
1482 * New and modified GDB server monitor features:
1484 - Thread local variables/storage (__thread) can now be displayed.
1486 - The GDB server monitor command "v.info location <address>"
1487 displays information about an address. The information produced
1488 depends on the tool and on the options given to valgrind.
1489 Possibly, the following are described: global variables, local
1490 (stack) variables, allocated or freed blocks, ...
1492 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
1493 ask the GDB server to stop at the start of program execution, at
1494 the end of the program execution and on Valgrind internal errors.
1496 - A new monitor command "v.info stats" shows various Valgrind core
1497 and tool statistics.
1499 - A new monitor command "v.set hostvisibility" allows the GDB server
1500 to provide access to Valgrind internal host status/memory.
1502 * A new option "--aspace-minaddr=<address>" can in some situations
1503 allow the use of more memory by decreasing the address above which
1504 Valgrind maps memory. It can also be used to solve address
1505 conflicts with system libraries by increasing the default value.
1506 See user manual for details.
1508 * The amount of memory used by Valgrind to store debug info (unwind
1509 info, line number information and symbol data) has been
1510 significantly reduced, even though Valgrind now reads more
1511 information in order to support unwinding of inlined function calls.
1513 * Dwarf3 handling with --read-var-info=yes has been improved:
1515 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
1517 - Code compiled with
1518 -ffunction-sections -fdata-sections -Wl,--gc-sections
1519 no longer causes assertion failures.
1521 * Improved checking for the --sim-hints= and --kernel-variant=
1522 options. Unknown strings are now detected and reported to the user
1525 * The semantics of stack start/end boundaries in the valgrind.h
1526 VALGRIND_STACK_REGISTER client request has been clarified and
1527 documented. The convention is that start and end are respectively
1528 the lowest and highest addressable bytes of the stack.
1530 * ==================== FIXED BUGS ====================
1532 The following bugs have been fixed or resolved. Note that "n-i-bz"
1533 stands for "not in bugzilla" -- that is, a bug that was reported to us
1534 but never got a bugzilla entry. We encourage you to file bugs in
1535 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1536 than mailing the developers (or mailing lists) directly -- bugs that
1537 are not entered into bugzilla tend to get forgotten about or ignored.
1539 To see details of a given bug, visit
1540 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1541 where XXXXXX is the bug number as listed below.
1543 175819 Support for ipv6 socket reporting with --track-fds
1544 232510 make distcheck fails
1545 249435 Analyzing wine programs with callgrind triggers a crash
1546 278972 support for inlined function calls in stacktraces and suppression
1548 291310 FXSAVE instruction marks memory as undefined on amd64
1549 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
1550 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
1551 315199 vgcore file for threaded app does not show which thread crashed
1552 315952 tun/tap ioctls are not supported
1553 323178 Unhandled instruction: PLDW register (ARM)
1554 323179 Unhandled instruction: PLDW immediate (ARM)
1555 324050 Helgrind: SEGV because of unaligned stack when using movdqa
1556 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
1557 325124 [MIPSEL] Compilation error
1558 325477 Phase 4 support for IBM Power ISA 2.07
1559 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
1560 325628 Phase 5 support for IBM Power ISA 2.07
1561 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
1562 325751 Missing the two privileged Power PC Transactional Memory Instructions
1563 325816 Phase 6 support for IBM Power ISA 2.07
1564 325856 Make SGCheck fail gracefully on unsupported platforms
1565 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
1566 326436 DRD: False positive in libstdc++ std::list::push_back
1567 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
1568 326462 Refactor vgdb to isolate invoker stuff into separate module
1569 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
1570 326623 DRD: false positive conflict report in a field assignment
1571 326724 Valgrind does not compile on OSX 1.9 Mavericks
1572 326816 Intercept for __strncpy_sse2_unaligned missing?
1573 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
1574 326983 Clear direction flag after tests on amd64.
1575 327212 Do not prepend the current directory to absolute path names.
1576 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
1577 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
1578 327284 s390x: Fix translation of the risbg instruction
1579 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
1580 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
1581 327916 DW_TAG_typedef may have no name
1582 327943 s390x: add a redirection for the 'index' function
1583 328100 XABORT not implemented
1584 328205 Implement additional Xen hypercalls
1585 328454 add support Backtraces with ARM unwind tables (EXIDX)
1586 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
1587 328711 valgrind.1 manpage "memcheck options" section is badly generated
1588 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
1589 329612 Incorrect handling of AT_BASE for image execution
1590 329694 clang warns about using uninitialized variable
1591 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
1592 330228 mmap must align to VKI_SHMLBA on mips32
1593 330257 LLVM does not support `-mno-dynamic-no-pic` option
1594 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
1595 330459 --track-fds=yes doesn't track eventfds
1596 330469 Add clock_adjtime syscall support
1597 330594 Missing sysalls on PowerPC / uClibc
1598 330622 Add test to regression suite for POWER instruction: dcbzl
1599 330939 Support for AMD's syscall instruction on x86
1601 330941 Typo in PRE(poll) syscall wrapper
1602 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
1603 331254 Fix expected output for memcheck/tests/dw4
1604 331255 Fix race condition in test none/tests/coolo_sigaction
1605 331257 Fix type of jump buffer in test none/tests/faultstatus
1606 331305 configure uses bash specific syntax
1607 331337 s390x WARNING: unhandled syscall: 326 (dup3)
1608 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
1609 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
1610 331829 Unexpected ioctl opcode sign extension
1611 331830 ppc64: WARNING: unhandled syscall: 96/97
1612 331839 drd/tests/sem_open specifies invalid semaphore name
1613 331847 outcome of drd/tests/thread_name is nondeterministic
1614 332037 Valgrind cannot handle Thumb "add pc, reg"
1615 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
1616 consistency checks enabled
1617 332263 intercepts for pthread_rwlock_timedrdlock and
1618 pthread_rwlock_timedwrlock are incorrect
1619 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
1621 332276 Implement additional Xen hypercalls
1622 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
1623 332765 Fix ms_print to create temporary files in a proper directory
1624 333072 drd: Add semaphore annotations
1625 333145 Tests for missaligned PC+#imm access for arm
1626 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
1627 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
1628 333248 WARNING: unhandled syscall: unix:443
1629 333428 ldr.w pc [rD, #imm] instruction leads to assertion
1630 333501 cachegrind: assertion: Cache set count is not a power of two.
1633 333666 Recognize MPX instructions and bnd prefix.
1634 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
1635 333817 Valgrind reports the memory areas written to by the SG_IO
1637 334049 lzcnt fails silently (x86_32)
1638 334384 Valgrind does not have support Little Endian support for
1640 334585 recvmmsg unhandled (+patch) (arm)
1641 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
1642 334727 Build fails with -Werror=format-security
1643 334788 clarify doc about --log-file initial program directory
1644 334834 PPC64 Little Endian support, patch 2
1645 334836 PPC64 Little Endian support, patch 3 testcase fixes
1646 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
1647 335034 Unhandled ioctl: HCIGETDEVLIST
1648 335155 vgdb, fix error print statement.
1649 335262 arm64: movi 8bit version is not supported
1650 335263 arm64: dmb instruction is not implemented
1651 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
1652 335496 arm64: sbc/abc instructions are not implemented
1653 335554 arm64: unhandled instruction: abs
1654 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
1655 335735 arm64: unhandled instruction: cnt
1656 335736 arm64: unhandled instruction: uaddlv
1657 335848 arm64: unhandled instruction: {s,u}cvtf
1658 335902 arm64: unhandled instruction: sli
1659 335903 arm64: unhandled instruction: umull (vector)
1660 336055 arm64: unhandled instruction: mov (element)
1661 336062 arm64: unhandled instruction: shrn{,2}
1662 336139 mip64: [...] valgrind hangs and spins on a single core [...]
1663 336189 arm64: unhandled Instruction: mvn
1664 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
1665 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
1666 336772 Make moans about unknown ioctls more informative
1667 336957 Add a section about the Solaris/illumos port on the webpage
1668 337094 ifunc wrapper is broken on ppc64
1669 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
1670 337528 leak check heuristic for block prefixed by length as 64bit number
1671 337740 Implement additional Xen hypercalls
1672 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
1673 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
1674 337871 deprecate --db-attach
1675 338023 Add support for all V4L2/media ioctls
1676 338024 inlined functions are not shown if DW_AT_ranges is used
1677 338106 Add support for 'kcmp' syscall
1678 338115 DRD: computed conflict set differs from actual after fork
1679 338160 implement display of thread local storage in gdbsrv
1680 338205 configure.ac and check for -Wno-tautological-compare
1681 338300 coredumps are missing one byte of every segment
1682 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
1683 338499 --sim-hints parsing broken due to wrong order in tokens
1684 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
1685 338681 Unable to unwind through clone thread created on i386-linux
1686 338698 race condition between gdbsrv and vgdb on startup
1687 338703 helgrind on arm-linux gets false positives in dynamic loader
1688 338791 alt dwz files can be relative of debug/main file
1689 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
1690 338932 build V-trunk with gcc-trunk
1691 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
1692 345079 Fix build problems in VEX/useful/test_main.c
1693 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
1694 n-i-bz s390x: Fix memory corruption for multithreaded applications
1695 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
1696 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
1697 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
1698 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
1699 n-i-bz Add command-line option --default-suppressions
1700 n-i-bz Add support for BLKDISCARDZEROES ioctl
1701 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
1702 n-i-bz Add support for sys_pivot_root and sys_unshare
1704 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
1705 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
1706 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
1710 Release 3.9.0 (31 October 2013)
1711 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1712 3.9.0 is a feature release with many improvements and the usual
1713 collection of bug fixes.
1715 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
1716 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
1717 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
1718 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
1720 * ================== PLATFORM CHANGES =================
1722 * Support for MIPS64 LE and BE running Linux. Valgrind has been
1723 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
1725 * Support for MIPS DSP ASE on MIPS32 platforms.
1727 * Support for s390x Decimal Floating Point instructions on hosts that
1728 have the DFP facility installed.
1730 * Support for POWER8 (Power ISA 2.07) instructions
1732 * Support for Intel AVX2 instructions. This is available only on 64
1735 * Initial support for Intel Transactional Synchronization Extensions,
1738 * Initial support for Hardware Transactional Memory on POWER.
1740 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
1741 run large GUI apps tolerably well.
1743 * ==================== TOOL CHANGES ====================
1747 - Improvements in handling of vectorised code, leading to
1748 significantly fewer false error reports. You need to use the flag
1749 --partial-loads-ok=yes to get the benefits of these changes.
1751 - Better control over the leak checker. It is now possible to
1752 specify which leak kinds (definite/indirect/possible/reachable)
1753 should be displayed, which should be regarded as errors, and which
1754 should be suppressed by a given leak suppression. This is done
1755 using the options --show-leak-kinds=kind1,kind2,..,
1756 --errors-for-leak-kinds=kind1,kind2,.. and an optional
1757 "match-leak-kinds:" line in suppression entries, respectively.
1759 Note that generated leak suppressions contain this new line and
1760 are therefore more specific than in previous releases. To get the
1761 same behaviour as previous releases, remove the "match-leak-kinds:"
1762 line from generated suppressions before using them.
1764 - Reduced "possible leak" reports from the leak checker by the use
1765 of better heuristics. The available heuristics provide detection
1766 of valid interior pointers to std::stdstring, to new[] allocated
1767 arrays with elements having destructors and to interior pointers
1768 pointing to an inner part of a C++ object using multiple
1769 inheritance. They can be selected individually using the
1770 option --leak-check-heuristics=heur1,heur2,...
1772 - Better control of stacktrace acquisition for heap-allocated
1773 blocks. Using the --keep-stacktraces option, it is possible to
1774 control independently whether a stack trace is acquired for each
1775 allocation and deallocation. This can be used to create better
1776 "use after free" errors or to decrease Valgrind's resource
1777 consumption by recording less information.
1779 - Better reporting of leak suppression usage. The list of used
1780 suppressions (shown when the -v option is given) now shows, for
1781 each leak suppressions, how many blocks and bytes it suppressed
1782 during the last leak search.
1786 - False errors resulting from the use of statically initialised
1787 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
1790 - False errors resulting from the use of pthread_cond_waits that
1791 timeout, have been removed.
1793 * ==================== OTHER CHANGES ====================
1795 * Some attempt to tune Valgrind's space requirements to the expected
1796 capabilities of the target:
1798 - The default size of the translation cache has been reduced from 8
1799 sectors to 6 on Android platforms, since each sector occupies
1800 about 40MB when using Memcheck.
1802 - The default size of the translation cache has been increased to 16
1803 sectors on all other platforms, reflecting the fact that large
1804 applications require instrumentation and storage of huge amounts
1805 of code. For similar reasons, the number of memory mapped
1806 segments that can be tracked has been increased by a factor of 6.
1808 - In all cases, the maximum number of sectors in the translation
1809 cache can be controlled by the new flag --num-transtab-sectors.
1811 * Changes in how debug info (line numbers, etc) is read:
1813 - Valgrind no longer temporarily mmaps the entire object to read
1814 from it. Instead, reading is done through a small fixed sized
1815 buffer. This avoids virtual memory usage spikes when Valgrind
1816 reads debuginfo from large shared objects.
1818 - A new experimental remote debug info server. Valgrind can read
1819 debug info from a different machine (typically, a build host)
1820 where debuginfo objects are stored. This can save a lot of time
1821 and hassle when running Valgrind on resource-constrained targets
1822 (phones, tablets) when the full debuginfo objects are stored
1823 somewhere else. This is enabled by the --debuginfo-server=
1826 - Consistency checking between main and debug objects can be
1827 disabled using the --allow-mismatched-debuginfo option.
1829 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
1830 scanning can recover stack traces in some cases when the normal
1831 unwind mechanisms fail. Stack scanning is best described as "a
1832 nasty, dangerous and misleading hack" and so is disabled by default.
1833 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
1836 * Detection and merging of recursive stack frame cycles. When your
1837 program has recursive algorithms, this limits the memory used by
1838 Valgrind for recorded stack traces and avoids recording
1839 uninteresting repeated calls. This is controlled by the command
1840 line option --merge-recursive-frame and by the monitor command
1841 "v.set merge-recursive-frames".
1843 * File name and line numbers for used suppressions. The list of used
1844 suppressions (shown when the -v option is given) now shows, for each
1845 used suppression, the file name and line number where the suppression
1848 * New and modified GDB server monitor features:
1850 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
1851 that can be used to execute gdbserver monitor commands from the
1854 - A new monitor command, "v.info open_fds", that gives the list of
1855 open file descriptors and additional details.
1857 - An optional message in the "v.info n_errs_found" monitor command,
1858 for example "v.info n_errs_found test 1234 finished", allowing a
1859 comment string to be added to the process output, perhaps for the
1860 purpose of separating errors of different tests or test phases.
1862 - A new monitor command "v.info execontext" that shows information
1863 about the stack traces recorded by Valgrind.
1865 - A new monitor command "v.do expensive_sanity_check_general" to run
1866 some internal consistency checks.
1868 * New flag --sigill-diagnostics to control whether a diagnostic
1869 message is printed when the JIT encounters an instruction it can't
1870 translate. The actual behavior -- delivery of SIGILL to the
1871 application -- is unchanged.
1873 * The maximum amount of memory that Valgrind can use on 64 bit targets
1874 has been increased from 32GB to 64GB. This should make it possible
1875 to run applications on Memcheck that natively require up to about 35GB.
1877 * ==================== FIXED BUGS ====================
1879 The following bugs have been fixed or resolved. Note that "n-i-bz"
1880 stands for "not in bugzilla" -- that is, a bug that was reported to us
1881 but never got a bugzilla entry. We encourage you to file bugs in
1882 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1883 than mailing the developers (or mailing lists) directly -- bugs that
1884 are not entered into bugzilla tend to get forgotten about or ignored.
1886 To see details of a given bug, visit
1887 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1888 where XXXXXX is the bug number as listed below.
1890 123837 system call: 4th argument is optional, depending on cmd
1891 135425 memcheck should tell you where Freed blocks were Mallocd
1892 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
1893 207815 Adds some of the drm ioctls to syswrap-linux.c
1894 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
1895 252955 Impossible to compile with ccache
1896 253519 Memcheck reports auxv pointer accesses as invalid reads.
1897 263034 Crash when loading some PPC64 binaries
1898 269599 Increase deepest backtrace
1899 274695 s390x: Support "compare to/from logical" instructions (z196)
1900 275800 s390x: Autodetect cache info (part 2)
1901 280271 Valgrind reports possible memory leaks on still-reachable std::string
1902 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
1903 289578 Backtraces with ARM unwind tables (stack scan flags)
1904 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
1905 304832 ppc32: build failure
1906 305431 Use find_buildid shdr fallback for separate .debug files
1907 305728 Add support for AVX2 instructions
1908 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
1909 306035 s390x: Fix IR generation for LAAG and friends
1910 306054 s390x: Condition code computation for convert-to-int/logical
1911 306098 s390x: alternate opcode form for convert to/from fixed
1912 306587 Fix cache line detection from auxiliary vector for PPC.
1913 306783 Mips unhandled syscall : 4025 / 4079 / 4182
1914 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
1915 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
1916 307101 sys_capget second argument can be NULL
1917 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
1918 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
1919 307113 s390x: DFP support
1920 307141 valgrind does't work in mips-linux system
1921 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
1922 307285 x86_amd64 feature test for avx in test suite is wrong
1923 307290 memcheck overlap testcase needs memcpy version filter
1924 307463 Please add "&limit=0" to the "all open bugs" link
1925 307465 --show-possibly-lost=no should reduce the error count / exit code
1926 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
1927 307729 pkgconfig support broken valgrind.pc
1928 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
1929 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
1930 308089 Unhandled syscall on ppc64: prctl
1931 308135 PPC32 MPC8xx has 16 bytes cache size
1932 308321 testsuite memcheck filter interferes with gdb_filter
1934 308341 vgdb should report process exit (or fatal signal)
1935 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
1936 308495 Remove build dependency on installed Xen headers
1937 308573 Internal error on 64-bit instruction executed in 32-bit mode
1939 308627 pmovmskb validity bit propagation is imprecise
1940 308644 vgdb command for having the info for the track-fds option
1941 308711 give more info about aspacemgr and arenas in out_of_memory
1942 308717 ARM: implement fixed-point VCVT.F64.[SU]32
1943 308718 ARM implement SMLALBB family of instructions
1944 308886 Missing support for PTRACE_SET/GETREGSET
1945 308930 syscall name_to_handle_at (303 on amd64) not handled
1946 309229 V-bit tester does not report number of tests generated
1947 309323 print unrecognized instuction on MIPS
1948 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
1949 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
1950 309430 Self hosting ppc64 encounters a vassert error on operand type
1951 309600 valgrind is a bit confused about 0-sized sections
1952 309823 Generate errors for still reachable blocks
1953 309921 PCMPISTRI validity bit propagation is imprecise
1954 309922 none/tests/ppc64/test_dfp5 sometimes fails
1955 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
1956 310424 --read-var-info does not properly describe static variables
1957 310792 search additional path for debug symbols
1958 310931 s390x: Message-security assist (MSA) instruction extension [..]
1959 311100 PPC DFP implementation of the integer operands is inconsistent [..]
1960 311318 ARM: "128-bit constant is not implemented" error message
1961 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
1962 311690 V crashes because it redirects branches inside of a redirected function
1963 311880 x86_64: make regtest hangs at shell_valid1
1964 311922 WARNING: unhandled syscall: 170
1966 312171 ppc: insn selection for DFP
1967 312571 Rounding mode call wrong for the DFP Iops [..]
1968 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
1969 312913 Dangling pointers error should also report the alloc stack trace
1970 312980 Building on Mountain Lion generates some compiler warnings
1971 313267 Adding MIPS64/Linux port to Valgrind
1974 313811 Buffer overflow in assert_fail
1975 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
1976 314269 ppc: dead code in insn selection
1977 314718 ARM: implement integer divide instruction (sdiv and udiv)
1978 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
1979 315441 sendmsg syscall should ignore unset msghdr msg_flags
1980 315534 msgrcv inside a thread causes valgrind to hang (block)
1981 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
1982 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
1983 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
1984 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
1985 316144 valgrind.1 manpage contains unknown ??? strings [..]
1986 316145 callgrind command line options in manpage reference (unknown) [..]
1987 316145 callgrind command line options in manpage reference [..]
1988 316181 drd: Fixed a 4x slowdown for certain applications
1989 316503 Valgrind does not support SSE4 "movntdqa" instruction
1990 316535 Use of |signed int| instead of |size_t| in valgrind messages
1991 316696 fluidanimate program of parsec 2.1 stuck
1992 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
1993 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
1994 317186 "Impossible happens" when occurs VCVT instruction on ARM
1995 317318 Support for Threading Building Blocks "scalable_malloc"
1996 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
1997 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
1998 317463 bmi testcase IR SANITY CHECK FAILURE
1999 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
2000 318050 libmpiwrap fails to compile with out-of-source build
2001 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
2002 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
2003 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
2004 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
2005 318932 Add missing PPC64 and PPC32 system call support
2006 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
2007 319395 Crash with unhandled instruction on STRT (Thumb) instructions
2008 319494 VEX Makefile-gcc standalone build update after r2702
2009 319505 [MIPSEL] Crash: unhandled UNRAY operator.
2010 319858 disInstr(thumb): unhandled instruction on instruction STRBT
2011 319932 disInstr(thumb): unhandled instruction on instruction STRHT
2012 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
2013 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
2014 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
2015 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
2016 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
2017 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
2018 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
2019 320895 add fanotify support (patch included)
2020 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
2021 321065 Valgrind updates for Xen 4.3
2022 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
2023 321363 Unhandled instruction: SSAX (ARM + Thumb)
2024 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
2025 321466 Unhandled instruction: SHASX (ARM + Thumb)
2026 321467 Unhandled instruction: SHSAX (ARM + Thumb)
2027 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
2028 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
2029 321620 Unhandled instruction: UASX (ARM + Thumb)
2030 321621 Unhandled instruction: USAX (ARM + Thumb)
2031 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
2032 321693 Unhandled instruction: LDRSBT (Thumb)
2033 321694 Unhandled instruction: UQASX (ARM + Thumb)
2034 321696 Unhandled instruction: UQSAX (Thumb + ARM)
2035 321697 Unhandled instruction: UHASX (ARM + Thumb)
2036 321703 Unhandled instruction: UHSAX (ARM + Thumb)
2037 321704 Unhandled instruction: REVSH (ARM + Thumb)
2038 321730 Add cg_diff and cg_merge man pages
2039 321738 Add vgdb and valgrind-listener man pages
2041 321891 Unhandled instruction: LDRHT (Thumb)
2042 321960 pthread_create() then alloca() causing invalid stack write errors
2043 321969 ppc32 and ppc64 don't support [lf]setxattr
2044 322254 Show threadname together with tid if set by application
2045 322294 Add initial support for IBM Power ISA 2.07
2046 322368 Assertion failure in wqthread_hijack under OS X 10.8
2047 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
2048 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
2049 322851 0bXXX binary literal syntax is not standard
2050 323035 Unhandled instruction: LDRSHT(Thumb)
2051 323036 Unhandled instruction: SMMLS (ARM and Thumb)
2052 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
2053 323175 Unhandled instruction: SMLALD (ARM + Thumb)
2054 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
2055 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
2056 323437 Phase 2 support for IBM Power ISA 2.07
2057 323713 Support mmxext (integer sse) subset on i386 (athlon)
2058 323803 Transactional memory instructions are not supported for Power
2059 323893 SSE3 not available on amd cpus in valgrind
2060 323905 Probable false positive from Valgrind/drd on close()
2061 323912 valgrind.h header isn't compatible for mingw64
2062 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
2063 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
2064 324181 mmap does not handle MAP_32BIT
2065 324227 memcheck false positive leak when a thread calls exit+block [..]
2066 324421 Support for fanotify API on ARM architecture
2067 324514 gdbserver monitor cmd output behaviour consistency [..]
2068 324518 ppc64: Emulation of dcbt instructions does not handle [..]
2069 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
2070 324582 When access is made to freed memory, report both allocation [..]
2071 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
2072 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
2073 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
2074 324834 Unhandled instructions in Microsoft C run-time for x86_64
2075 324894 Phase 3 support for IBM Power ISA 2.07
2076 326091 drd: Avoid false race reports from optimized strlen() impls
2077 326113 valgrind libvex hwcaps error on AMD64
2078 n-i-bz Some wrong command line options could be ignored
2079 n-i-bz patch to allow fair-sched on android
2080 n-i-bz report error for vgdb snapshot requested before execution
2081 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
2083 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
2087 Release 3.8.1 (19 September 2012)
2088 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2089 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
2090 that occur moderately frequently in real use cases, adds support for
2091 some missing instructions on ARM, and fixes a deadlock condition on
2092 MacOSX. If you package or deliver 3.8.0 for others to use, you might
2093 want to consider upgrading to 3.8.1 instead.
2095 The following bugs have been fixed or resolved. Note that "n-i-bz"
2096 stands for "not in bugzilla" -- that is, a bug that was reported to us
2097 but never got a bugzilla entry. We encourage you to file bugs in
2098 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2099 than mailing the developers (or mailing lists) directly -- bugs that
2100 are not entered into bugzilla tend to get forgotten about or ignored.
2102 To see details of a given bug, visit
2103 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2104 where XXXXXX is the bug number as listed below.
2107 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
2108 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
2109 298281 wcslen causes false(?) uninitialised value warnings
2110 301281 valgrind hangs on OS X when the process calls system()
2111 304035 disInstr(arm): unhandled instruction 0xE1023053
2112 304867 implement MOVBE instruction in x86 mode
2113 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
2114 305042 amd64: implement 0F 7F encoding of movq between two registers
2115 305199 ARM: implement QDADD and QDSUB
2116 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
2117 305513 killed by fatal signal: SIGSEGV
2118 305690 DRD reporting invalid semaphore when sem_trywait fails
2119 305926 Invalid alignment checks for some AVX instructions
2120 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
2121 306310 3.8.0 release tarball missing some files
2122 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
2123 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
2124 n-i-bz shmat of a segment > 4Gb does not work
2125 n-i-bz simulate_control_c script wrong USR1 signal number on mips
2126 n-i-bz vgdb ptrace calls wrong on mips [...]
2127 n-i-bz Fixes for more MPI false positives
2128 n-i-bz exp-sgcheck's memcpy causes programs to segfault
2129 n-i-bz OSX build w/ clang: asserts at startup
2130 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
2131 n-i-bz fix a couple of union tag-vs-field mixups
2132 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
2134 The following bugs were fixed in 3.8.0 but not listed in this NEWS
2137 254088 Valgrind should know about UD2 instruction
2140 304754 NEWS blows TeX's little mind
2142 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
2146 Release 3.8.0 (10 August 2012)
2147 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2148 3.8.0 is a feature release with many improvements and the usual
2149 collection of bug fixes.
2151 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2152 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
2153 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
2154 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
2155 There is initial support for MacOSX 10.8, but it is not usable for
2156 serious work at present.
2158 * ================== PLATFORM CHANGES =================
2160 * Support for MIPS32 platforms running Linux. Valgrind has been
2161 tested on MIPS32 and MIPS32r2 platforms running different Debian
2162 Squeeze and MeeGo distributions. Both little-endian and big-endian
2163 cores are supported. The tools Memcheck, Massif and Lackey have
2164 been tested and are known to work. See README.mips for more details.
2166 * Preliminary support for Android running on x86.
2168 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
2170 * Support for Intel AVX instructions and for AES instructions. This
2171 support is available only for 64 bit code.
2173 * Support for POWER Decimal Floating Point instructions.
2175 * ==================== TOOL CHANGES ====================
2177 * Non-libc malloc implementations are now supported. This is useful
2178 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
2179 Using the new option --soname-synonyms, such tools can be informed
2180 that the malloc implementation is either linked statically into the
2181 executable, or is present in some other shared library different
2182 from libc.so. This makes it possible to process statically linked
2183 programs, and programs using other malloc libraries, for example
2184 TCMalloc or JEMalloc.
2186 * For tools that provide their own replacement for malloc et al, the
2187 option --redzone-size=<number> allows users to specify the size of
2188 the padding blocks (redzones) added before and after each client
2189 allocated block. Smaller redzones decrease the memory needed by
2190 Valgrind. Bigger redzones increase the chance to detect blocks
2191 overrun or underrun. Prior to this change, the redzone size was
2192 hardwired to 16 bytes in Memcheck.
2196 - The leak_check GDB server monitor command now can
2197 control the maximum nr of loss records to output.
2199 - Reduction of memory use for applications allocating
2200 many blocks and/or having many partially defined bytes.
2202 - Addition of GDB server monitor command 'block_list' that lists
2203 the addresses/sizes of the blocks of a leak search loss record.
2205 - Addition of GDB server monitor command 'who_points_at' that lists
2206 the locations pointing at a block.
2208 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2209 detect an invalid access of these redzones, by marking them
2210 noaccess. Similarly, if a redzone size is given for a memory
2211 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2212 This still allows to find some bugs if the user has forgotten to
2213 mark the pool superblock noaccess.
2215 - Performance of memory leak check has been improved, especially in
2216 cases where there are many leaked blocks and/or many suppression
2217 rules used to suppress leak reports.
2219 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2220 more precise analysis, which is important for LLVM/Clang
2221 generated code. This is at the cost of somewhat reduced
2222 performance. Note there is no change to analysis precision or
2223 costs on Linux targets.
2227 - Added even more facilities that can help finding the cause of a data
2228 race, namely the command-line option --ptrace-addr and the macro
2229 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2231 - Fixed a subtle bug that could cause false positive data race reports.
2233 * ==================== OTHER CHANGES ====================
2235 * The C++ demangler has been updated so as to work well with C++
2236 compiled by up to at least g++ 4.6.
2238 * Tool developers can make replacement/wrapping more flexible thanks
2239 to the new option --soname-synonyms. This was reported above, but
2240 in fact is very general and applies to all function
2241 replacement/wrapping, not just to malloc-family functions.
2243 * Round-robin scheduling of threads can be selected, using the new
2244 option --fair-sched= yes. Prior to this change, the pipe-based
2245 thread serialisation mechanism (which is still the default) could
2246 give very unfair scheduling. --fair-sched=yes improves
2247 responsiveness of interactive multithreaded applications, and
2248 improves repeatability of results from the thread checkers Helgrind
2251 * For tool developers: support to run Valgrind on Valgrind has been
2252 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2254 * gdbserver now shows the float shadow registers as integer
2255 rather than float values, as the shadow values are mostly
2256 used as bit patterns.
2258 * Increased limit for the --num-callers command line flag to 500.
2260 * Performance improvements for error matching when there are many
2261 suppression records in use.
2263 * Improved support for DWARF4 debugging information (bug 284184).
2265 * Initial support for DWZ compressed Dwarf debug info.
2267 * Improved control over the IR optimiser's handling of the tradeoff
2268 between performance and precision of exceptions. Specifically,
2269 --vex-iropt-precise-memory-exns has been removed and replaced by
2270 --vex-iropt-register-updates, with extended functionality. This
2271 allows the Valgrind gdbserver to always show up to date register
2274 * Modest performance gains through the use of translation chaining for
2277 * ==================== FIXED BUGS ====================
2279 The following bugs have been fixed or resolved. Note that "n-i-bz"
2280 stands for "not in bugzilla" -- that is, a bug that was reported to us
2281 but never got a bugzilla entry. We encourage you to file bugs in
2282 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2283 than mailing the developers (or mailing lists) directly -- bugs that
2284 are not entered into bugzilla tend to get forgotten about or ignored.
2286 To see details of a given bug, visit
2287 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2288 where XXXXXX is the bug number as listed below.
2290 197914 Building valgrind from svn now requires automake-1.10
2291 203877 increase to 16Mb maximum allowed alignment for memalign et al
2292 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
2293 247386 make perf does not run all performance tests
2294 270006 Valgrind scheduler unfair
2295 270777 Adding MIPS/Linux port to Valgrind
2296 270796 s390x: Removed broken support for the TS insn
2297 271438 Fix configure for proper SSE4.2 detection
2298 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
2299 273475 Add support for AVX instructions
2300 274078 improved configure logic for mpicc
2301 276993 fix mremap 'no thrash checks'
2302 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
2303 281482 memcheck incorrect byte allocation count in realloc() for silly argument
2304 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
2305 283413 Fix wrong sanity check
2306 283671 Robustize alignment computation in LibVEX_Alloc
2307 283961 Adding support for some HCI IOCTLs
2308 284124 parse_type_DIE: confused by: DWARF 4
2309 284864 == 273475 (Add support for AVX instructions)
2310 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
2311 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
2312 285725 == 273475 (Add support for AVX instructions)
2313 286261 add wrapper for linux I2C_RDWR ioctl
2314 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
2315 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
2316 286384 configure fails "checking for a supported version of gcc"
2317 286497 == 273475 (Add support for AVX instructions)
2318 286596 == 273475 (Add support for AVX instructions)
2319 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
2320 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
2321 287260 Incorrect conditional jump or move depends on uninitialised value(s)
2322 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
2323 287307 == 273475 (Add support for AVX instructions)
2324 287858 VG_(strerror): unknown error
2325 288298 (MacOSX) unhandled syscall shm_unlink
2326 288995 == 273475 (Add support for AVX instructions)
2327 289470 Loading of large Mach-O thin binaries fails.
2328 289656 == 273475 (Add support for AVX instructions)
2329 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
2330 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
2331 289839 s390x: Provide support for unicode conversion instructions
2332 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
2333 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
2334 290655 Add support for AESKEYGENASSIST instruction
2335 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
2336 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
2337 291253 ES register not initialised in valgrind simulation
2338 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
2339 291865 s390x: Support the "Compare Double and Swap" family of instructions
2340 292300 == 273475 (Add support for AVX instructions)
2341 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
2342 292493 == 273475 (Add support for AVX instructions)
2343 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
2344 292627 Missing support for some SCSI ioctls
2345 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
2346 292841 == 273475 (Add support for AVX instructions)
2347 292993 implement the getcpu syscall on amd64-linux
2348 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
2349 293088 Add some VEX sanity checks for ppc64 unhandled instructions
2350 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
2351 293754 PCMPxSTRx not implemented for 16-bit characters
2352 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
2353 293808 CLFLUSH not supported by latest VEX for amd64
2354 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
2355 294048 MPSADBW instruction not implemented
2356 294055 regtest none/tests/shell fails when locale is not set to C
2357 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
2358 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
2359 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
2360 294260 disInstr_AMD64: disInstr miscalculated next %rip
2361 294523 --partial-loads-ok=yes causes false negatives
2362 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
2363 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
2364 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
2365 295089 can not annotate source for both helgrind and drd
2366 295221 POWER Processor decimal floating point instruction support missing
2367 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
2368 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
2369 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
2370 295617 ARM - Add some missing syscalls
2371 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
2372 296229 Linux user input device ioctls missing wrappers
2373 296318 ELF Debug info improvements (more than one rx/rw mapping)
2374 296422 Add translation chaining support
2375 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
2376 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
2377 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
2378 297078 gdbserver signal handling problems [..]
2379 297147 drd false positives on newly allocated memory
2380 297329 disallow decoding of IBM Power DFP insns on some machines
2381 297497 POWER Processor decimal floating point instruction support missing
2382 297701 Another alias for strncasecmp_l in libc-2.13.so
2383 297911 'invalid write' not reported when using APIs for custom mem allocators.
2384 297976 s390x: revisit EX implementation
2385 297991 Valgrind interferes with mmap()+ftell()
2386 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
2387 297993 Fix compilation of valgrind with gcc -g3.
2388 298080 POWER Processor DFP support missing, part 3
2389 298227 == 273475 (Add support for AVX instructions)
2390 298335 == 273475 (Add support for AVX instructions)
2391 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
2392 298394 s390x: Don't bail out on an unknown machine model. [..]
2393 298421 accept4() syscall (366) support is missing for ARM
2394 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
2395 298732 valgrind installation problem in ubuntu with kernel version 3.x
2396 298862 POWER Processor DFP instruction support missing, part 4
2397 298864 DWARF reader mis-parses DW_FORM_ref_addr
2398 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
2399 299053 Support DWARF4 DW_AT_high_pc constant form
2400 299104 == 273475 (Add support for AVX instructions)
2401 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
2402 299629 dup3() syscall (358) support is missing for ARM
2403 299694 POWER Processor DFP instruction support missing, part 5
2404 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
2405 299803 == 273475 (Add support for AVX instructions)
2406 299804 == 273475 (Add support for AVX instructions)
2407 299805 == 273475 (Add support for AVX instructions)
2408 300140 ARM - Missing (T1) SMMUL
2409 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
2410 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
2411 300414 FCOM and FCOMP unimplemented for amd64 guest
2412 301204 infinite loop in canonicaliseSymtab with ifunc symbol
2413 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
2414 301265 add x86 support to Android build
2415 301984 configure script doesn't detect certain versions of clang
2416 302205 Fix compiler warnings for POWER VEX code and POWER test cases
2417 302287 Unhandled movbe instruction on Atom processors
2418 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
2419 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
2420 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
2421 302656 == 273475 (Add support for AVX instructions)
2422 302709 valgrind for ARM needs extra tls support for android emulator [..]
2423 302827 add wrapper for CDROM_GET_CAPABILITY
2424 302901 Valgrind crashes with dwz optimized debuginfo
2425 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
2426 303116 Add support for the POWER instruction popcntb
2427 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
2428 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
2429 303466 == 273475 (Add support for AVX instructions)
2430 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
2431 303963 strstr() function produces wrong results under valgrind callgrind
2432 304054 CALL_FN_xx macros need to enforce stack alignment
2433 304561 tee system call not supported
2434 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
2435 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
2436 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
2437 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
2438 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
2439 n-i-bz s390x: Shadow registers can now be examined using vgdb
2441 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
2442 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
2446 Release 3.7.0 (5 November 2011)
2447 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2448 3.7.0 is a feature release with many significant improvements and the
2449 usual collection of bug fixes.
2451 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2452 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
2453 Support for recent distros and toolchain components (glibc 2.14, gcc
2454 4.6, MacOSX 10.7) has been added.
2456 * ================== PLATFORM CHANGES =================
2458 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
2459 analyse 64-bit programs running on z/Architecture. Most user space
2460 instructions up to and including z10 are supported. Valgrind has
2461 been tested extensively on z9, z10, and z196 machines running SLES
2462 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
2463 known to work well. Callgrind, Helgrind, and DRD work reasonably
2464 well on z9 and later models. See README.s390 for more details.
2466 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
2467 64-bit processes are supported. Some complex threaded applications
2468 (Firefox) are observed to hang when run as 32 bit applications,
2469 whereas 64-bit versions run OK. The cause is unknown. Memcheck
2470 will likely report some false errors. In general, expect some rough
2471 spots. This release also supports MacOSX 10.6, but drops support
2474 * Preliminary support for Android (on ARM). Valgrind can now run
2475 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
2476 README.android for more details, plus instructions on how to get
2479 * Support for the IBM Power ISA 2.06 (Power7 instructions)
2481 * General correctness and performance improvements for ARM/Linux, and,
2482 by extension, ARM/Android.
2484 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
2485 instruction set support is under development but is not available in
2488 * Support for AIX5 has been removed.
2490 * ==================== TOOL CHANGES ====================
2492 * Memcheck: some incremental changes:
2494 - reduction of memory use in some circumstances
2496 - improved handling of freed memory, which in some circumstances
2497 can cause detection of use-after-free that would previously have
2500 - fix of a longstanding bug that could cause false negatives (missed
2501 errors) in programs doing vector saturated narrowing instructions.
2503 * Helgrind: performance improvements and major memory use reductions,
2504 particularly for large, long running applications which perform many
2505 synchronisation (lock, unlock, etc) events. Plus many smaller
2508 - display of locksets for both threads involved in a race
2510 - general improvements in formatting/clarity of error messages
2512 - addition of facilities and documentation regarding annotation
2513 of thread safe reference counted C++ classes
2515 - new flag --check-stack-refs=no|yes [yes], to disable race checking
2516 on thread stacks (a performance hack)
2518 - new flag --free-is-write=no|yes [no], to enable detection of races
2519 where one thread accesses heap memory but another one frees it,
2520 without any coordinating synchronisation event
2522 * DRD: enabled XML output; added support for delayed thread deletion
2523 in order to detect races that occur close to the end of a thread
2524 (--join-list-vol); fixed a memory leak triggered by repeated client
2525 memory allocatation and deallocation; improved Darwin support.
2527 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
2529 * exp-sgcheck: this tool has been reduced in scope so as to improve
2530 performance and remove checking that Memcheck does better.
2531 Specifically, the ability to check for overruns for stack and global
2532 arrays is unchanged, but the ability to check for overruns of heap
2533 blocks has been removed. The tool has accordingly been renamed to
2534 exp-sgcheck ("Stack and Global Array Checking").
2536 * ==================== OTHER CHANGES ====================
2538 * GDB server: Valgrind now has an embedded GDB server. That means it
2539 is possible to control a Valgrind run from GDB, doing all the usual
2540 things that GDB can do (single stepping, breakpoints, examining
2541 data, etc). Tool-specific functionality is also available. For
2542 example, it is possible to query the definedness state of variables
2543 or memory from within GDB when running Memcheck; arbitrarily large
2544 memory watchpoints are supported, etc. To use the GDB server, start
2545 Valgrind with the flag --vgdb-error=0 and follow the on-screen
2548 * Improved support for unfriendly self-modifying code: a new option
2549 --smc-check=all-non-file is available. This adds the relevant
2550 consistency checks only to code that originates in non-file-backed
2551 mappings. In effect this confines the consistency checking only to
2552 code that is or might be JIT generated, and avoids checks on code
2553 that must have been compiled ahead of time. This significantly
2554 improves performance on applications that generate code at run time.
2556 * It is now possible to build a working Valgrind using Clang-2.9 on
2559 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
2560 These enable and disable error reporting on a per-thread, and
2561 nestable, basis. This is useful for hiding errors in particularly
2562 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
2563 now uses this facility.
2565 * Added the --mod-funcname option to cg_diff.
2567 * ==================== FIXED BUGS ====================
2569 The following bugs have been fixed or resolved. Note that "n-i-bz"
2570 stands for "not in bugzilla" -- that is, a bug that was reported to us
2571 but never got a bugzilla entry. We encourage you to file bugs in
2572 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2573 mailing the developers (or mailing lists) directly -- bugs that are
2574 not entered into bugzilla tend to get forgotten about or ignored.
2576 To see details of a given bug, visit
2577 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2578 where XXXXXX is the bug number as listed below.
2580 79311 malloc silly arg warning does not give stack trace
2581 210935 port valgrind.h (not valgrind) to win32 to support client requests
2582 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
2583 243404 Port to zSeries
2584 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
2585 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
2586 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
2587 253206 Some fixes for the faultstatus testcase
2588 255223 capget testcase fails when running as root
2589 256703 xlc_dbl_u32.c testcase broken
2590 256726 Helgrind tests have broken inline asm
2591 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
2592 264800 testcase compile failure on zseries
2593 265762 make public VEX headers compilable by G++ 3.x
2594 265771 assertion in jumps.c (r11523) fails with glibc-2.3
2595 266753 configure script does not give the user the option to not use QtCore
2596 266931 gen_insn_test.pl is broken
2597 266961 ld-linux.so.2 i?86-linux strlen issues
2598 266990 setns instruction causes false positive
2599 267020 Make directory for temporary files configurable at run-time.
2600 267342 == 267997 (segmentation fault on Mac OS 10.6)
2601 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
2602 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
2603 267488 regtest: darwin support for 64-bit build
2604 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
2605 267630 Add support for IBM Power ISA 2.06 -- stage 1
2606 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
2607 267819 Add client request for informing the core about reallocation
2608 267925 laog data structure quadratic for a single sequence of lock
2609 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
2610 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
2611 268513 missed optimizations in fold_Expr
2612 268619 s390x: fpr - gpr transfer facility
2613 268620 s390x: reconsider "long displacement" requirement
2614 268621 s390x: improve IR generation for XC
2615 268715 s390x: FLOGR is not universally available
2616 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
2617 268930 s390x: MHY is not universally available
2618 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
2619 269079 Support ptrace system call on ARM
2620 269144 missing "Bad option" error message
2621 269209 conditional load and store facility (z196)
2622 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
2623 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
2624 269736 s390x: minor code generation tweaks
2625 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
2626 269863 s390x: remove unused function parameters
2627 269864 s390x: tweak s390_emit_load_cc
2628 269884 == 250101 (overhead for huge blocks exhausts space too soon)
2629 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
2630 270115 s390x: rewrite some testcases
2631 270309 == 267997 (valgrind crash on startup)
2632 270320 add support for Linux FIOQSIZE ioctl() call
2633 270326 segfault while trying to sanitize the environment passed to execle
2634 270794 IBM POWER7 support patch causes regression in none/tests
2635 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
2636 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
2637 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
2638 270959 s390x: invalid use of R0 as base register
2639 271042 VSX configure check fails when it should not
2640 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
2641 271259 s390x: fix code confusion
2642 271337 == 267997 (Valgrind segfaults on MacOS X)
2643 271385 s390x: Implement Ist_MBE
2644 271501 s390x: misc cleanups
2645 271504 s390x: promote likely and unlikely
2646 271579 ppc: using wrong enum type
2647 271615 unhandled instruction "popcnt" (arch=amd10h)
2648 271730 Fix bug when checking ioctls: duplicate check
2649 271776 s390x: provide STFLE instruction support
2650 271779 s390x: provide clock instructions like STCK
2651 271799 Darwin: ioctls without an arg report a memory error
2652 271820 arm: fix type confusion
2653 271917 pthread_cond_timedwait failure leads to not-locked false positive
2654 272067 s390x: fix DISP20 macro
2655 272615 A typo in debug output in mc_leakcheck.c
2656 272661 callgrind_annotate chokes when run from paths containing regex chars
2657 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
2658 272955 Unhandled syscall error for pwrite64 on ppc64 arch
2659 272967 make documentation build-system more robust
2660 272986 Fix gcc-4.6 warnings with valgrind.h
2661 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
2662 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
2663 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
2664 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
2665 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
2666 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
2667 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
2668 273778 exp-ptrcheck: unhandled sysno == 259
2669 274089 exp-ptrcheck: unhandled sysno == 208
2670 274378 s390x: Various dispatcher tweaks
2671 274447 WARNING: unhandled syscall: 340
2672 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
2673 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
2674 274926 valgrind does not build against linux-3
2675 275148 configure FAIL with glibc-2.14
2676 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
2677 275168 Make Valgrind work for MacOSX 10.7 Lion
2678 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
2679 275278 valgrind does not build on Linux kernel 3.0.* due to silly
2680 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
2681 275308 Fix implementation for ppc64 fres instruc
2682 275339 s390x: fix testcase compile warnings
2683 275517 s390x: Provide support for CKSM instruction
2684 275710 s390x: get rid of redundant address mode calculation
2685 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
2686 275852 == 250101 (valgrind uses all swap space and is killed)
2687 276784 Add support for IBM Power ISA 2.06 -- stage 3
2688 276987 gdbsrv: fix tests following recent commits
2689 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
2690 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
2691 277471 Unhandled syscall: 340
2692 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
2693 277653 ARM: support Thumb2 PLD instruction
2694 277663 ARM: NEON float VMUL by scalar incorrect
2695 277689 ARM: tests for VSTn with register post-index are broken
2696 277694 ARM: BLX LR instruction broken in ARM mode
2697 277780 ARM: VMOV.F32 (immediate) instruction is broken
2698 278057 fuse filesystem syscall deadlocks
2699 278078 Unimplemented syscall 280 on ppc32
2700 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
2701 278454 VALGRIND_STACK_DEREGISTER has wrong output type
2702 278502 == 275284 (Valgrind confuses memcpy() and memmove())
2703 278892 gdbsrv: factorize gdb version handling, fix doc and typos
2704 279027 Support for MVCL and CLCL instruction
2705 279027 s390x: Provide support for CLCL and MVCL instructions
2706 279062 Remove a redundant check in the insn selector for ppc.
2707 279071 JDK creates PTEST with redundant REX.W prefix
2708 279212 gdbsrv: add monitor cmd v.info scheduler.
2709 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
2710 279698 memcheck discards valid-bits for packuswb
2711 279795 memcheck reports uninitialised values for mincore on amd64
2712 279994 Add support for IBM Power ISA 2.06 -- stage 3
2713 280083 mempolicy syscall check errors
2714 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
2715 280710 s390x: config files for nightly builds
2716 280757 /tmp dir still used by valgrind even if TMPDIR is specified
2717 280965 Valgrind breaks fcntl locks when program does mmap
2718 281138 WARNING: unhandled syscall: 340
2719 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
2720 281304 == 275168 (Darwin: dyld "cannot load inserted library")
2721 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
2722 281468 s390x: handle do_clone and gcc clones in call traces
2723 281488 ARM: VFP register corruption
2724 281828 == 275284 (false memmove warning: "Source and destination overlap")
2725 281883 s390x: Fix system call wrapper for "clone".
2726 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
2727 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
2728 282238 SLES10: make check fails
2729 282979 strcasestr needs replacement with recent(>=2.12) glibc
2730 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
2731 283243 Regression in ppc64 memcheck tests
2732 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
2733 283427 re-connect epoll_pwait syscall on ARM linux
2734 283600 gdbsrv: android: port vgdb.c
2735 283709 none/tests/faultstatus needs to account for page size
2736 284305 filter_gdb needs enhancement to work on ppc64
2737 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
2738 284472 Thumb2 ROR.W encoding T2 not implemented
2739 284621 XML-escape process command line in XML output
2740 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
2741 that have non-power-of-2 sizes (also AMDs)
2742 n-i-bz don't be spooked by libraries mashed by elfhack
2743 n-i-bz don't be spooked by libxul.so linked with gold
2744 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
2746 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
2747 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
2748 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
2752 Release 3.6.1 (16 February 2011)
2753 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2754 3.6.1 is a bug fix release. It adds support for some SSE4
2755 instructions that were omitted in 3.6.0 due to lack of time. Initial
2756 support for glibc-2.13 has been added. A number of bugs causing
2757 crashing or assertion failures have been fixed.
2759 The following bugs have been fixed or resolved. Note that "n-i-bz"
2760 stands for "not in bugzilla" -- that is, a bug that was reported to us
2761 but never got a bugzilla entry. We encourage you to file bugs in
2762 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2763 mailing the developers (or mailing lists) directly -- bugs that are
2764 not entered into bugzilla tend to get forgotten about or ignored.
2766 To see details of a given bug, visit
2767 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2768 where XXXXXX is the bug number as listed below.
2770 188572 Valgrind on Mac should suppress setenv() mem leak
2771 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
2772 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
2773 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
2774 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
2775 254420 memory pool tracking broken
2776 254957 Test code failing to compile due to changes in memcheck.h
2777 255009 helgrind/drd: crash on chmod with invalid parameter
2778 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
2779 255355 helgrind/drd: crash on threaded programs doing fork
2781 255418 (SSE4.x) rint call compiled with ICC
2782 255822 --gen-suppressions can create invalid files: "too many callers [...]"
2783 255888 closing valgrindoutput tag outputted to log-stream on error
2784 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
2785 255966 Slowness when using mempool annotations
2786 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
2787 256600 super-optimized strcasecmp() false positive
2788 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
2789 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
2790 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
2791 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
2792 257276 Missing case in memcheck --track-origins=yes
2793 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
2794 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
2795 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
2796 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
2797 263099 callgrind_annotate counts Ir improperly [...]
2798 263877 undefined coprocessor instruction on ARMv7
2799 265964 configure FAIL with glibc-2.13
2800 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
2801 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
2802 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
2803 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
2804 n-i-bz DRD: disable --free-is-write due to implementation difficulties
2806 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
2810 Release 3.6.0 (21 October 2010)
2811 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2812 3.6.0 is a feature release with many significant improvements and the
2813 usual collection of bug fixes.
2815 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2816 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
2817 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
2819 -------------------------
2821 Here are some highlights. Details are shown further down:
2823 * Support for ARM/Linux.
2825 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
2827 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
2829 * Support for the SSE4.2 instruction set.
2831 * Enhancements to the Callgrind profiler, including the ability to
2832 handle CPUs with three levels of cache.
2834 * A new experimental heap profiler, DHAT.
2836 * A huge number of bug fixes and small enhancements.
2838 -------------------------
2840 Here are details of the above changes, together with descriptions of
2841 many other changes, and a list of fixed bugs.
2843 * ================== PLATFORM CHANGES =================
2845 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
2846 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
2847 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
2849 This requires a CPU capable of running the ARMv7-A instruction set
2850 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
2851 of the user space instruction set, including ARM and Thumb integer
2852 code, VFPv3, NEON and V6 media instructions. The Memcheck,
2853 Cachegrind and Massif tools work properly; other tools work to
2856 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
2857 with support for recent releases of the underlying toolchain
2858 components, notably gcc-4.5 and glibc-2.12.
2860 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
2861 support also works much better on OS X 10.5, and is as solid as
2864 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
2865 64-bit mode. In 32-bit mode, support is only available up to and
2866 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
2867 supported in 64-bit mode, and 32-bit mode does in fact support the
2868 bare minimum SSE4 instructions to needed to run programs on Mac OS X
2869 10.6 on 32-bit targets.
2871 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
2872 and including version 2.05 is supported.
2874 * ==================== TOOL CHANGES ====================
2876 * Cachegrind has a new processing script, cg_diff, which finds the
2877 difference between two profiles. It's very useful for evaluating
2878 the performance effects of a change in a program.
2880 Related to this change, the meaning of cg_annotate's (rarely-used)
2881 --threshold option has changed; this is unlikely to affect many
2882 people, if you do use it please see the user manual for details.
2884 * Callgrind now can do branch prediction simulation, similar to
2885 Cachegrind. In addition, it optionally can count the number of
2886 executed global bus events. Both can be used for a better
2887 approximation of a "Cycle Estimation" as derived event (you need to
2888 update the event formula in KCachegrind yourself).
2890 * Cachegrind and Callgrind now refer to the LL (last-level) cache
2891 rather than the L2 cache. This is to accommodate machines with
2892 three levels of caches -- if Cachegrind/Callgrind auto-detects the
2893 cache configuration of such a machine it will run the simulation as
2894 if the L2 cache isn't present. This means the results are less
2895 likely to match the true result for the machine, but
2896 Cachegrind/Callgrind's results are already only approximate, and
2897 should not be considered authoritative. The results are still
2898 useful for giving a general idea about a program's locality.
2900 * Massif has a new option, --pages-as-heap, which is disabled by
2901 default. When enabled, instead of tracking allocations at the level
2902 of heap blocks (as allocated with malloc/new/new[]), it instead
2903 tracks memory allocations at the level of memory pages (as mapped by
2904 mmap, brk, etc). Each mapped page is treated as its own block.
2905 Interpreting the page-level output is harder than the heap-level
2906 output, but this option is useful if you want to account for every
2907 byte of memory used by a program.
2909 * DRD has two new command-line options: --free-is-write and
2910 --trace-alloc. The former allows to detect reading from already freed
2911 memory, and the latter allows tracing of all memory allocations and
2914 * DRD has several new annotations. Custom barrier implementations can
2915 now be annotated, as well as benign races on static variables.
2917 * DRD's happens before / happens after annotations have been made more
2918 powerful, so that they can now also be used to annotate e.g. a smart
2919 pointer implementation.
2921 * Helgrind's annotation set has also been drastically improved, so as
2922 to provide to users a general set of annotations to describe locks,
2923 semaphores, barriers and condition variables. Annotations to
2924 describe thread-safe reference counted heap objects have also been
2927 * Memcheck has a new command-line option, --show-possibly-lost, which
2928 is enabled by default. When disabled, the leak detector will not
2929 show possibly-lost blocks.
2931 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
2932 has been added. DHAT keeps track of allocated heap blocks, and also
2933 inspects every memory reference to see which block (if any) is being
2934 accessed. This gives a lot of insight into block lifetimes,
2935 utilisation, turnover, liveness, and the location of hot and cold
2936 fields. You can use DHAT to do hot-field profiling.
2938 * ==================== OTHER CHANGES ====================
2940 * Improved support for unfriendly self-modifying code: the extra
2941 overhead incurred by --smc-check=all has been reduced by
2942 approximately a factor of 5 as compared with 3.5.0.
2944 * Ability to show directory names for source files in error messages.
2945 This is combined with a flexible mechanism for specifying which
2946 parts of the paths should be shown. This is enabled by the new flag
2949 * A new flag, --require-text-symbol, which will stop the run if a
2950 specified symbol is not found it a given shared object when it is
2951 loaded into the process. This makes advanced working with function
2952 intercepting and wrapping safer and more reliable.
2954 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
2955 and control of Valgrind is now available for the tools Memcheck and
2956 Helgrind. XML output from Valgrind is available for Memcheck,
2957 Helgrind and exp-Ptrcheck.
2959 * More reliable stack unwinding on amd64-linux, particularly in the
2960 presence of function wrappers, and with gcc-4.5 compiled code.
2962 * Modest scalability (performance improvements) for massive
2963 long-running applications, particularly for those with huge amounts
2966 * Support for analyzing programs running under Wine with has been
2967 improved. The header files <valgrind/valgrind.h>,
2968 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
2969 Windows-programs compiled with MinGW or one of the Microsoft Visual
2972 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
2973 The 32-bit simulator was not affected. This did not occur often,
2974 but when it did would usually crash the program under test.
2977 * A large number of bugs were fixed. These are shown below.
2979 * A number of bugs were investigated, and were candidates for fixing,
2980 but are not fixed in 3.6.0, due to lack of developer time. They may
2981 get fixed in later releases. They are:
2983 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
2984 212419 false positive "lock order violated" (A+B vs A)
2985 213685 Undefined value propagates past dependency breaking instruction
2986 216837 Incorrect instrumentation of NSOperationQueue on Darwin
2987 237920 valgrind segfault on fork failure
2988 242137 support for code compiled by LLVM-2.8
2989 242423 Another unknown Intel cache config value
2990 243232 Inconsistent Lock Orderings report with trylock
2991 243483 ppc: callgrind triggers VEX assertion failure
2992 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
2993 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
2995 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
2996 249435 Analyzing wine programs with callgrind triggers a crash
2997 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
2998 250065 Handling large allocations
2999 250101 huge "free" memory usage due to m_mallocfree.c
3000 "superblocks fragmentation"
3001 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
3002 252091 Callgrind on ARM does not detect function returns correctly
3003 252600 [PATCH] Allow lhs to be a pointer for shl/shr
3004 254420 memory pool tracking broken
3005 n-i-bz support for adding symbols for JIT generated code
3008 The following bugs have been fixed or resolved. Note that "n-i-bz"
3009 stands for "not in bugzilla" -- that is, a bug that was reported to us
3010 but never got a bugzilla entry. We encourage you to file bugs in
3011 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3012 mailing the developers (or mailing lists) directly -- bugs that are
3013 not entered into bugzilla tend to get forgotten about or ignored.
3015 To see details of a given bug, visit
3016 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3017 where XXXXXX is the bug number as listed below.
3019 135264 dcbzl instruction missing
3021 153699 Valgrind should report unaligned reads with movdqa
3023 190429 Valgrind reports lost of errors in ld.so
3024 with x86_64 2.9.90 glibc
3025 197266 valgrind appears to choke on the xmms instruction
3027 197988 Crash when demangling very large symbol names
3028 202315 unhandled syscall: 332 (inotify_init1)
3029 203256 Add page-level profiling to Massif
3030 205093 dsymutil=yes needs quotes, locking (partial fix)
3031 205241 Snow Leopard 10.6 support (partial fix)
3032 206600 Leak checker fails to upgrade indirect blocks when their
3033 parent becomes reachable
3034 210935 port valgrind.h (not valgrind) to win32 so apps run under
3035 wine can make client requests
3036 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
3037 within Linux ip-stack checksum functions
3038 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
3040 213685 Undefined value propagates past dependency breaking instruction
3042 215914 Valgrind inserts bogus empty environment variable
3044 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
3045 222545 shmat fails under valgind on some arm targets
3046 222560 ARM NEON support
3049 232509 Docs build fails with formatting inside <title></title> elements
3051 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
3052 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
3053 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
3054 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
3055 237485 symlink (syscall 57) is not supported on Mac OS
3056 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
3058 238208 is_just_below_ESP doesn't take into account red-zone
3059 238345 valgrind passes wrong $0 when executing a shell script
3060 238679 mq_timedreceive syscall doesn't flag the reception buffer
3062 238696 fcntl command F_DUPFD_CLOEXEC not supported
3063 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3064 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
3065 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
3067 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
3073 242606 unhandled syscall: setegid (in Ptrcheck)
3074 242814 Helgrind "Impossible has happened" during
3075 QApplication::initInstance();
3076 243064 Valgrind attempting to read debug information from iso
3077 243270 Make stack unwinding in Valgrind wrappers more reliable
3078 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
3079 sysno = 277 (mq_open)
3080 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
3081 244493 ARM VFP d16-d31 registers support
3082 244670 add support for audit_session_self syscall on Mac OS 10.6
3083 244921 The xml report of helgrind tool is not well format
3084 244923 In the xml report file, the <preamble> not escape the
3085 xml char, eg '<','&','>'
3086 245535 print full path names in plain text reports
3087 245925 x86-64 red zone handling problem
3088 246258 Valgrind not catching integer underruns + new [] s
3089 246311 reg/reg cmpxchg doesn't work on amd64
3090 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
3091 246888 Improve Makefile.vex.am
3092 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
3093 to [f]chmod_extended
3094 247526 IBM POWER6 (ISA 2.05) support is incomplete
3095 247561 Some leak testcases fails due to reachable addresses in
3097 247875 sizeofIRType to handle Ity_I128
3098 247894 [PATCH] unhandled syscall sys_readahead
3099 247980 Doesn't honor CFLAGS passed to configure
3100 248373 darwin10.supp is empty in the trunk
3101 248822 Linux FIBMAP ioctl has int parameter instead of long
3102 248893 [PATCH] make readdwarf.c big endianess safe to enable
3103 unwinding on big endian systems
3104 249224 Syscall 336 not supported (SYS_proc_info)
3106 249775 Incorrect scheme for detecting NEON capabilities of host CPU
3107 249943 jni JVM init fails when using valgrind
3108 249991 Valgrind incorrectly declares AESKEYGENASSIST support
3110 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
3111 250799 frexp$fenv_access_off function generates SIGILL
3112 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
3113 251251 support pclmulqdq insn
3114 251362 valgrind: ARM: attach to debugger either fails or provokes
3116 251674 Unhandled syscall 294
3119 254257 Add support for debugfiles found by build-id
3120 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
3121 254646 Wrapped functions cause stack misalignment on OS X
3122 (and possibly Linux)
3123 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
3125 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
3129 Release 3.5.0 (19 August 2009)
3130 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3131 3.5.0 is a feature release with many significant improvements and the
3132 usual collection of bug fixes. The main improvement is that Valgrind
3133 now works on Mac OS X.
3135 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
3136 and X86/Darwin. Support for recent distros and toolchain components
3137 (glibc 2.10, gcc 4.5) has been added.
3139 -------------------------
3141 Here is a short summary of the changes. Details are shown further
3144 * Support for Mac OS X (10.5.x).
3146 * Improvements and simplifications to Memcheck's leak checker.
3148 * Clarification and simplifications in various aspects of Valgrind's
3151 * XML output for Helgrind and Ptrcheck.
3153 * Performance and stability improvements for Helgrind and DRD.
3155 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
3157 * A new experimental tool, BBV, useful for computer architecture
3160 * Improved Wine support, including ability to read Windows PDB
3163 -------------------------
3165 Here are details of the above changes, followed by descriptions of
3166 many other minor changes, and a list of fixed bugs.
3169 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
3170 called "Darwin" because that is the name of the OS core, which is the
3171 level that Valgrind works at.)
3175 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
3176 because it would require work and 10.4 is only becoming less common.
3178 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
3179 fairly well. For 10.5.x, 32-bit programs are the default even on
3180 64-bit machines, so it handles most current programs.
3182 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
3183 officially supported, but simple programs at least will probably work.
3184 However, start-up is slow.
3186 - PowerPC machines are not supported.
3188 Things that don't work:
3190 - The Ptrcheck tool.
3192 - Objective-C garbage collection.
3196 - If you have Rogue Amoeba's "Instant Hijack" program installed,
3197 Valgrind will fail with a SIGTRAP at start-up. See
3198 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
3203 - You will likely find --dsymutil=yes a useful option, as error
3204 messages may be imprecise without it.
3206 - Mac OS X support is new and therefore will be less robust than the
3207 Linux support. Please report any bugs you find.
3209 - Threaded programs may run more slowly than on Linux.
3211 Many thanks to Greg Parker for developing this port over several years.
3214 * Memcheck's leak checker has been improved.
3216 - The results for --leak-check=summary now match the summary results
3217 for --leak-check=full. Previously they could differ because
3218 --leak-check=summary counted "indirectly lost" blocks and
3219 "suppressed" blocks as "definitely lost".
3221 - Blocks that are only reachable via at least one interior-pointer,
3222 but are directly pointed to by a start-pointer, were previously
3223 marked as "still reachable". They are now correctly marked as
3226 - The default value for the --leak-resolution option has been
3227 changed from "low" to "high". In general, this means that more
3228 leak reports will be produced, but each leak report will describe
3229 fewer leaked blocks.
3231 - With --leak-check=full, "definitely lost" and "possibly lost"
3232 leaks are now considered as proper errors, ie. they are counted
3233 for the "ERROR SUMMARY" and affect the behaviour of
3234 --error-exitcode. These leaks are not counted as errors if
3235 --leak-check=summary is specified, however.
3237 - Documentation for the leak checker has been improved.
3240 * Various aspects of Valgrind's text output have changed.
3242 - Valgrind's start-up message has changed. It is shorter but also
3243 includes the command being run, which makes it easier to use
3244 --trace-children=yes. An example:
3246 - Valgrind's shut-down messages have also changed. This is most
3247 noticeable with Memcheck, where the leak summary now occurs before
3248 the error summary. This change was necessary to allow leaks to be
3249 counted as proper errors (see the description of the leak checker
3250 changes above for more details). This was also necessary to fix a
3251 longstanding bug in which uses of suppressions against leaks were
3252 not "counted", leading to difficulties in maintaining suppression
3253 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3255 - Behavior of -v has changed. In previous versions, -v printed out
3256 a mixture of marginally-user-useful information, and tool/core
3257 statistics. The statistics printing has now been moved to its own
3258 flag, --stats=yes. This means -v is less verbose and more likely
3259 to convey useful end-user information.
3261 - The format of some (non-XML) stack trace entries has changed a
3262 little. Previously there were six possible forms:
3264 0x80483BF: really (a.c:20)
3265 0x80483BF: really (in /foo/a.out)
3267 0x80483BF: (within /foo/a.out)
3268 0x80483BF: ??? (a.c:20)
3271 The third and fourth of these forms have been made more consistent
3272 with the others. The six possible forms are now:
3274 0x80483BF: really (a.c:20)
3275 0x80483BF: really (in /foo/a.out)
3276 0x80483BF: really (in ???)
3277 0x80483BF: ??? (in /foo/a.out)
3278 0x80483BF: ??? (a.c:20)
3281 Stack traces produced when --xml=yes is specified are different
3285 * Helgrind and Ptrcheck now support XML output, so they can be used
3286 from GUI tools. Also, the XML output mechanism has been
3289 - The XML format has been overhauled and generalised, so it is more
3290 suitable for error reporting tools in general. The Memcheck
3291 specific aspects of it have been removed. The new format, which
3292 is an evolution of the old format, is described in
3293 docs/internals/xml-output-protocol4.txt.
3295 - Memcheck has been updated to use the new format.
3297 - Helgrind and Ptrcheck are now able to emit output in this format.
3299 - The XML output mechanism has been overhauled. XML is now output
3300 to its own file descriptor, which means that:
3302 * Valgrind can output text and XML independently.
3304 * The longstanding problem of XML output being corrupted by
3305 unexpected un-tagged text messages is solved.
3307 As before, the destination for text output is specified using
3308 --log-file=, --log-fd= or --log-socket=.
3310 As before, XML output for a tool is enabled using --xml=yes.
3312 Because there's a new XML output channel, the XML output
3313 destination is now specified by --xml-file=, --xml-fd= or
3316 Initial feedback has shown this causes some confusion. To
3317 clarify, the two envisaged usage scenarios are:
3319 (1) Normal text output. In this case, do not specify --xml=yes
3320 nor any of --xml-file=, --xml-fd= or --xml-socket=.
3322 (2) XML output. In this case, specify --xml=yes, and one of
3323 --xml-file=, --xml-fd= or --xml-socket= to select the XML
3324 destination, one of --log-file=, --log-fd= or --log-socket=
3325 to select the destination for any remaining text messages,
3326 and, importantly, -q.
3328 -q makes Valgrind completely silent on the text channel,
3329 except in the case of critical failures, such as Valgrind
3330 itself segfaulting, or failing to read debugging information.
3331 Hence, in this scenario, it suffices to check whether or not
3332 any output appeared on the text channel. If yes, then it is
3333 likely to be a critical error which should be brought to the
3334 attention of the user. If no (the text channel produced no
3335 output) then it can be assumed that the run was successful.
3337 This allows GUIs to make the critical distinction they need to
3338 make (did the run fail or not?) without having to search or
3339 filter the text output channel in any way.
3341 It is also recommended to use --child-silent-after-fork=yes in
3345 * Improvements and changes in Helgrind:
3347 - XML output, as described above
3349 - Checks for consistent association between pthread condition
3350 variables and their associated mutexes are now performed.
3352 - pthread_spinlock functions are supported.
3354 - Modest performance improvements.
3356 - Initial (skeletal) support for describing the behaviour of
3357 non-POSIX synchronisation objects through ThreadSanitizer
3358 compatible ANNOTATE_* macros.
3360 - More controllable tradeoffs between performance and the level of
3361 detail of "previous" accesses in a race. There are now three
3364 * --history-level=full. This is the default, and was also the
3365 default in 3.4.x. It shows both stacks involved in a race, but
3366 requires a lot of memory and can be very slow in programs that
3367 do many inter-thread synchronisation events.
3369 * --history-level=none. This only shows the later stack involved
3370 in a race. This can be much faster than --history-level=full,
3371 but makes it much more difficult to find the other access
3372 involved in the race.
3374 The new intermediate setting is
3376 * --history-level=approx
3378 For the earlier (other) access, two stacks are presented. The
3379 earlier access is guaranteed to be somewhere in between the two
3380 program points denoted by those stacks. This is not as useful
3381 as showing the exact stack for the previous access (as per
3382 --history-level=full), but it is better than nothing, and it's
3383 almost as fast as --history-level=none.
3386 * New features and improvements in DRD:
3388 - The error messages printed by DRD are now easier to interpret.
3389 Instead of using two different numbers to identify each thread
3390 (Valgrind thread ID and DRD thread ID), DRD does now identify
3391 threads via a single number (the DRD thread ID). Furthermore
3392 "first observed at" information is now printed for all error
3393 messages related to synchronization objects.
3395 - Added support for named semaphores (sem_open() and sem_close()).
3397 - Race conditions between pthread_barrier_wait() and
3398 pthread_barrier_destroy() calls are now reported.
3400 - Added support for custom allocators through the macros
3401 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
3402 in <valgrind/valgrind.h>). An alternative for these two macros is
3403 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
3406 - Added support for annotating non-POSIX synchronization objects
3407 through several new ANNOTATE_*() macros.
3409 - OpenMP: added support for the OpenMP runtime (libgomp) included
3410 with gcc versions 4.3.0 and 4.4.0.
3414 - Added two new command-line options (--first-race-only and
3415 --segment-merging-interval).
3418 * Genuinely atomic support for x86/amd64/ppc atomic instructions
3420 Valgrind will now preserve (memory-access) atomicity of LOCK-
3421 prefixed x86/amd64 instructions, and any others implying a global
3422 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
3424 This means that Valgrinded processes will "play nicely" in
3425 situations where communication with other processes, or the kernel,
3426 is done through shared memory and coordinated with such atomic
3427 instructions. Prior to this change, such arrangements usually
3428 resulted in hangs, races or other synchronisation failures, because
3429 Valgrind did not honour atomicity of such instructions.
3432 * A new experimental tool, BBV, has been added. BBV generates basic
3433 block vectors for use with the SimPoint analysis tool, which allows
3434 a program's overall behaviour to be approximated by running only a
3435 fraction of it. This is useful for computer architecture
3436 researchers. You can run BBV by specifying --tool=exp-bbv (the
3437 "exp-" prefix is short for "experimental"). BBV was written by
3441 * Modestly improved support for running Windows applications under
3442 Wine. In particular, initial support for reading Windows .PDB debug
3443 information has been added.
3446 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
3447 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
3451 * The Valgrind client requests VALGRIND_PRINTF and
3452 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
3453 the string was always printed immediately on its own line. Now, the
3454 string will be added to a buffer but not printed until a newline is
3455 encountered, or other Valgrind output is printed (note that for
3456 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
3457 "other Valgrind output"). This allows you to use multiple
3458 VALGRIND_PRINTF calls to build up a single output line, and also to
3459 print multiple output lines with a single request (by embedding
3460 multiple newlines in the string).
3463 * The graphs drawn by Massif's ms_print program have changed slightly:
3465 - The half-height chars '.' and ',' are no longer drawn, because
3466 they are confusing. The --y option can be used if the default
3467 y-resolution is not high enough.
3469 - Horizontal lines are now drawn after the top of a snapshot if
3470 there is a gap until the next snapshot. This makes it clear that
3471 the memory usage has not dropped to zero between snapshots.
3474 * Something that happened in 3.4.0, but wasn't clearly announced: the
3475 option --read-var-info=yes can be used by some tools (Memcheck,
3476 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
3477 variable type and location information. This makes those tools
3478 start up more slowly and increases memory consumption, but
3479 descriptions of data addresses in error messages become more
3483 * exp-Omega, an experimental instantaneous leak-detecting tool, was
3484 disabled in 3.4.0 due to a lack of interest and maintenance,
3485 although the source code was still in the distribution. The source
3486 code has now been removed from the distribution. For anyone
3487 interested, the removal occurred in SVN revision r10247.
3490 * Some changes have been made to the build system.
3492 - VEX/ is now integrated properly into the build system. This means
3493 that dependency tracking within VEX/ now works properly, "make
3494 install" will work without requiring "make" before it, and
3495 parallel builds (ie. 'make -j') now work (previously a
3496 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
3497 was effectively ignored).
3499 - The --with-vex configure option has been removed. It was of
3500 little use and removing it simplified the build system.
3502 - The location of some install files has changed. This should not
3503 affect most users. Those who might be affected:
3505 * For people who use Valgrind with MPI programs, the installed
3506 libmpiwrap.so library has moved from
3507 $(INSTALL)/<platform>/libmpiwrap.so to
3508 $(INSTALL)/libmpiwrap-<platform>.so.
3510 * For people who distribute standalone Valgrind tools, the
3511 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
3512 have moved to $(INSTALL)/libcoregrind-<platform>.a.
3514 These changes simplify the build system.
3516 - Previously, all the distributed suppression (*.supp) files were
3517 installed. Now, only default.supp is installed. This should not
3518 affect users as the other installed suppression files were not
3519 read; the fact that they were installed was a mistake.
3522 * KNOWN LIMITATIONS:
3524 - Memcheck is unusable with the Intel compiler suite version 11.1,
3525 when it generates code for SSE2-and-above capable targets. This
3526 is because of icc's use of highly optimised inlined strlen
3527 implementations. It causes Memcheck to report huge numbers of
3528 false errors even in simple programs. Helgrind and DRD may also
3531 Versions 11.0 and earlier may be OK, but this has not been
3535 The following bugs have been fixed or resolved. Note that "n-i-bz"
3536 stands for "not in bugzilla" -- that is, a bug that was reported to us
3537 but never got a bugzilla entry. We encourage you to file bugs in
3538 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3539 mailing the developers (or mailing lists) directly -- bugs that are
3540 not entered into bugzilla tend to get forgotten about or ignored.
3542 To see details of a given bug, visit
3543 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3544 where XXXXXX is the bug number as listed below.
3546 84303 How about a LockCheck tool?
3547 91633 dereference of null ptr in vgPlain_st_basetype
3548 97452 Valgrind doesn't report any pthreads problems
3549 100628 leak-check gets assertion failure when using
3550 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
3551 108528 NPTL pthread cleanup handlers not called
3552 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
3553 110128 mallinfo is not implemented...
3554 110770 VEX: Generated files not always updated when making valgrind
3555 111102 Memcheck: problems with large (memory footprint) applications
3556 115673 Vex's decoder should never assert
3557 117564 False positive: Syscall param clone(child_tidptr) contains
3558 uninitialised byte(s)
3559 119404 executing ssh from inside valgrind fails
3560 133679 Callgrind does not write path names to sources with dwarf debug
3562 135847 configure.in problem with non gnu compilers (and possible fix)
3563 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
3564 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
3565 136230 memcheck reports "possibly lost", should be "still reachable"
3566 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
3567 137904 Valgrind reports a memory leak when using POSIX threads,
3569 139076 valgrind VT_GETSTATE error
3570 142228 complaint of elf_dynamic_do_rela in trivial usage
3571 145347 spurious warning with USBDEVFS_REAPURB
3572 148441 (wine) can't find memory leak in Wine, win32 binary
3574 148742 Leak-check fails assert on exit
3575 149878 add (proper) check for calloc integer overflow
3576 150606 Call graph is broken when using callgrind control
3577 152393 leak errors produce an exit code of 0. I need some way to
3578 cause leak errors to result in a nonzero exit code.
3579 157154 documentation (leak-resolution doc speaks about num-callers
3580 def=4) + what is a loss record
3581 159501 incorrect handling of ALSA ioctls
3582 162020 Valgrinding an empty/zero-byte file crashes valgrind
3583 162482 ppc: Valgrind crashes while reading stabs information
3584 162718 x86: avoid segment selector 0 in sys_set_thread_area()
3585 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
3586 163560 VEX/test_main.c is missing from valgrind-3.3.1
3587 164353 malloc_usable_size() doesn't return a usable size
3588 165468 Inconsistent formatting in memcheck manual -- please fix
3589 169505 main.c:286 (endOfInstr):
3590 Assertion 'ii->cost_offset == *cost_offset' failed
3591 177206 Generate default.supp during compile instead of configure
3592 177209 Configure valt_load_address based on arch+os
3593 177305 eventfd / syscall 323 patch lost
3594 179731 Tests fail to build because of inlining of non-local asm labels
3595 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
3596 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
3597 181594 Bogus warning for empty text segment
3598 181707 dwarf doesn't require enumerations to have name
3599 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3600 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3601 Assertion '!already_present' failed.
3602 185359 exp-ptrcheck: unhandled syscall getresuid()
3603 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
3604 185816 Valgrind is unable to handle debug info for files with split
3605 debug info that are prelinked afterwards
3606 185980 [darwin] unhandled syscall: sem_open
3607 186238 bbToIR_AMD64: disInstr miscalculated next %rip
3608 186507 exp-ptrcheck unhandled syscalls prctl, etc.
3609 186790 Suppression pattern used for leaks are not reported
3610 186796 Symbols with length>200 in suppression files are ignored
3611 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
3612 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
3613 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
3614 188046 bashisms in the configure script
3615 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
3616 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
3617 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
3618 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
3619 assertion fail "!lock->heldBy"
3620 188427 Add support for epoll_create1 (with patch)
3621 188530 Support for SIOCGSTAMPNS
3622 188560 Include valgrind.spec in the tarball
3623 188572 Valgrind on Mac should suppress setenv() mem leak
3624 189054 Valgrind fails to build because of duplicate non-local asm labels
3625 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
3626 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
3627 189763 drd assertion failure: s_threadinfo[tid].is_recording
3628 190219 unhandled syscall: 328 (x86-linux)
3629 190391 dup of 181394; see above
3630 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
3631 190820 No debug information on powerpc-linux
3632 191095 PATCH: Improve usbdevfs ioctl handling
3633 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
3635 191189 --xml=yes should obey --gen-suppressions=all
3636 191192 syslog() needs a suppression on macosx
3637 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
3638 191761 getrlimit on MacOSX
3639 191992 multiple --fn-skip only works sometimes; dependent on order
3640 192634 V. reports "aspacem sync_check_mapping_callback:
3641 segment mismatch" on Darwin
3642 192954 __extension__ missing on 2 client requests
3643 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
3644 194474 "INSTALL" file has different build instructions than "README"
3645 194671 Unhandled syscall (sem_wait?) from mac valgrind
3646 195069 memcheck: reports leak (memory still reachable) for
3648 195169 drd: (vgDrd_barrier_post_wait):
3649 Assertion 'r->sg[p->post_iteration]' failed.
3650 195268 valgrind --log-file doesn't accept ~/...
3651 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
3652 195860 WARNING: unhandled syscall: unix:223
3653 196528 need a error suppression for pthread_rwlock_init under os x?
3654 197227 Support aio_* syscalls on Darwin
3655 197456 valgrind should reject --suppressions=(directory)
3656 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
3657 197591 unhandled syscall 27 (mincore)
3658 197793 Merge DCAS branch to the trunk == 85756, 142103
3659 197794 Avoid duplicate filenames in Vex
3660 197898 make check fails on current SVN
3661 197901 make check fails also under exp-ptrcheck in current SVN
3662 197929 Make --leak-resolution=high the default
3663 197930 Reduce spacing between leak reports
3664 197933 Print command line of client at start-up, and shorten preamble
3665 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
3666 198395 add BBV to the distribution as an experimental tool
3667 198624 Missing syscalls on Darwin: 82, 167, 281, 347
3668 198649 callgrind_annotate doesn't cumulate counters
3669 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
3670 199977 Valgrind complains about an unrecognized instruction in the
3671 atomic_incs test program
3672 200029 valgrind isn't able to read Fedora 12 debuginfo
3673 200760 darwin unhandled syscall: unix:284
3674 200827 DRD doesn't work on Mac OS X
3675 200990 VG_(read_millisecond_timer)() does not work correctly
3676 201016 Valgrind does not support pthread_kill() on Mac OS
3677 201169 Document --read-var-info
3678 201323 Pre-3.5.0 performance sanity checking
3679 201384 Review user manual for the 3.5.0 release
3680 201585 mfpvr not implemented on ppc
3681 201708 tests failing because x86 direction flag is left set
3682 201757 Valgrind doesn't handle any recent sys_futex additions
3683 204377 64-bit valgrind can not start a shell script
3684 (with #!/path/to/shell) if the shell is a 32-bit executable
3685 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
3686 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
3687 about memory allocation events with memory access tracing enabled
3688 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
3690 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
3694 Release 3.4.1 (28 February 2009)
3695 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3696 3.4.1 is a bug-fix release that fixes some regressions and assertion
3697 failures in debug info reading in 3.4.0, most notably incorrect stack
3698 traces on amd64-linux on older (glibc-2.3 based) systems. Various
3699 other debug info problems are also fixed. A number of bugs in the
3700 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
3702 In view of the fact that 3.4.0 contains user-visible regressions
3703 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
3704 encouraged to ship 3.4.1 in preference to 3.4.0.
3706 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
3707 bugzilla" -- that is, a bug that was reported to us but never got a
3708 bugzilla entry. We encourage you to file bugs in bugzilla
3709 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
3710 developers (or mailing lists) directly -- bugs that are not entered
3711 into bugzilla tend to get forgotten about or ignored.
3713 n-i-bz Fix various bugs reading icc-11 generated debug info
3714 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
3715 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
3716 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
3717 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
3718 179618 exp-ptrcheck crashed / exit prematurely
3719 179624 helgrind: false positive races with pthread_create and
3720 recv/open/close/read
3721 134207 pkg-config output contains @VG_PLATFORM@
3722 176926 floating point exception at valgrind startup with PPC 440EPX
3723 181594 Bogus warning for empty text segment
3724 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
3725 181707 Dwarf3 doesn't require enumerations to have name
3726 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3727 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3728 Assertion '!already_present' failed.
3729 185359 exp-ptrcheck unhandled syscall getresuid()
3731 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
3732 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
3736 Release 3.4.0 (2 January 2009)
3737 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3738 3.4.0 is a feature release with many significant improvements and the
3739 usual collection of bug fixes. This release supports X86/Linux,
3740 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
3741 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
3743 3.4.0 brings some significant tool improvements. Memcheck can now
3744 report the origin of uninitialised values, the thread checkers
3745 Helgrind and DRD are much improved, and we have a new experimental
3746 tool, exp-Ptrcheck, which is able to detect overruns of stack and
3747 global arrays. In detail:
3749 * Memcheck is now able to track the origin of uninitialised values.
3750 When it reports an uninitialised value error, it will try to show
3751 the origin of the value, as either a heap or stack allocation.
3752 Origin tracking is expensive and so is not enabled by default. To
3753 use it, specify --track-origins=yes. Memcheck's speed will be
3754 essentially halved, and memory usage will be significantly
3755 increased. Nevertheless it can drastically reduce the effort
3756 required to identify the root cause of uninitialised value errors,
3757 and so is often a programmer productivity win, despite running more
3760 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
3761 3.4.0, will be released shortly.
3763 * Helgrind's race detection algorithm has been completely redesigned
3764 and reimplemented, to address usability and scalability concerns:
3766 - The new algorithm has a lower false-error rate: it is much less
3767 likely to report races that do not really exist.
3769 - Helgrind will display full call stacks for both accesses involved
3770 in a race. This makes it easier to identify the root causes of
3773 - Limitations on the size of program that can run have been removed.
3775 - Performance has been modestly improved, although that is very
3778 - Direct support for Qt4 threading has been added.
3780 - pthread_barriers are now directly supported.
3782 - Helgrind works well on all supported Linux targets.
3784 * The DRD thread debugging tool has seen major improvements:
3786 - Greatly improved performance and significantly reduced memory
3789 - Support for several major threading libraries (Boost.Thread, Qt4,
3790 glib, OpenMP) has been added.
3792 - Support for atomic instructions, POSIX semaphores, barriers and
3793 reader-writer locks has been added.
3795 - Works now on PowerPC CPUs too.
3797 - Added support for printing thread stack usage at thread exit time.
3799 - Added support for debugging lock contention.
3801 - Added a manual for Drd.
3803 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
3804 checks for misuses of pointers. In that sense it is a bit like
3805 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
3806 detect overruns of stack and global arrays, it can detect
3807 arbitrarily far out-of-bounds accesses to heap blocks, and it can
3808 detect accesses heap blocks that have been freed a very long time
3809 ago (millions of blocks in the past).
3811 Ptrcheck currently works only on x86-linux and amd64-linux. To use
3812 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
3813 of the main Valgrind documentation. As this is an experimental
3814 tool, we would be particularly interested in hearing about your
3815 experiences with it.
3817 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
3818 longer built by default, although the code remains in the repository
3819 and the tarball. This is due to three factors: a perceived lack of
3820 users, a lack of maintenance, and concerns that it may not be
3821 possible to achieve reliable operation using the existing design.
3823 * As usual, support for the latest Linux distros and toolchain
3824 components has been added. It should work well on Fedora Core 10,
3825 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
3826 state) is supported, as is glibc-2.9. The C++ demangler has been
3827 updated so as to work well with C++ compiled by even the most recent
3830 * You can now use frame-level wildcards in suppressions. This was a
3831 frequently-requested enhancement. A line "..." in a suppression now
3832 matches zero or more frames. This makes it easier to write
3833 suppressions which are precise yet insensitive to changes in
3836 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
3838 * Very basic support for IBM Power6 has been added (64-bit processes only).
3840 * Valgrind is now cross-compilable. For example, it is possible to
3841 cross compile Valgrind on an x86/amd64-linux host, so that it runs
3842 on a ppc32/64-linux target.
3844 * You can set the main thread's stack size at startup using the
3845 new --main-stacksize= flag (subject of course to ulimit settings).
3846 This is useful for running apps that need a lot of stack space.
3848 * The limitation that you can't use --trace-children=yes together
3849 with --db-attach=yes has been removed.
3851 * The following bugs have been fixed. Note that "n-i-bz" stands for
3852 "not in bugzilla" -- that is, a bug that was reported to us but
3853 never got a bugzilla entry. We encourage you to file bugs in
3854 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3855 mailing the developers (or mailing lists) directly.
3857 n-i-bz Make return types for some client requests 64-bit clean
3858 n-i-bz glibc 2.9 support
3859 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
3860 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
3861 n-i-bz Building in an env without gdb gives bogus gdb attach
3862 92456 Tracing the origin of uninitialised memory
3863 106497 Valgrind does not demangle some C++ template symbols
3865 151612 Suppression with "..." (frame-level wildcards in .supp files)
3866 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
3867 159285 unhandled syscall:25 (stime, on x86-linux)
3868 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
3869 160954 ppc build of valgrind crashes with illegal instruction (isel)
3870 160956 mallinfo implementation, w/ patch
3871 162092 Valgrind fails to start gnome-system-monitor
3872 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
3873 163794 assertion failure with "--track-origins=yes"
3874 163933 sigcontext.err and .trapno must be set together
3875 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
3876 164476 Missing kernel module loading system calls
3877 164669 SVN regression: mmap() drops posix file locks
3878 166581 Callgrind output corruption when program forks
3879 167288 Patch file for missing system calls on Cell BE
3880 168943 unsupported scas instruction pentium
3881 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
3882 172417 x86->IR: 0x82 ...
3883 172563 amd64->IR: 0xD9 0xF5 - fprem1
3884 173099 .lds linker script generation error
3885 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
3886 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
3888 174908 --log-file value not expanded correctly for core file
3889 175044 Add lookup_dcookie for amd64
3890 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
3892 Developer-visible changes:
3894 * Valgrind's debug-info reading machinery has been majorly overhauled.
3895 It can now correctly establish the addresses for ELF data symbols,
3896 which is something that has never worked properly before now.
3898 Also, Valgrind can now read DWARF3 type and location information for
3899 stack and global variables. This makes it possible to use the
3900 framework to build tools that rely on knowing the type and locations
3901 of stack and global variables, for example exp-Ptrcheck.
3903 Reading of such information is disabled by default, because most
3904 tools don't need it, and because it is expensive in space and time.
3905 However, you can force Valgrind to read it, using the
3906 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
3907 make use of such information, if present, to provide source-level
3908 descriptions of data addresses in the error messages they create.
3910 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
3911 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).