1 Release 3.14.0 (?? ????????? 201?)
2 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~--
3 Release 3.14.0 is under development, not yet released.
5 3.14.0 is a feature release with many improvements and the usual collection of
8 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
9 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
10 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
11 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12. There is also preliminary
12 support for X86/macOS 10.13, AMD64/macOS 10.13.
14 * ==================== CORE CHANGES ===================
16 * The new option --keep-debuginfo=no|yes (default no) can be used to
17 keep symbols etc for unloaded code. This allows saved stack traces
18 (e.g. memory leaks) to include file/line info for code that has been
19 dlclose'd (or similar). See user manual for more information and
21 * Ability to specify suppressions based on source file name and line number.
23 * ================== PLATFORM CHANGES =================
25 * Preliminary support for macOS 10.13 has been added.
26 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
27 * mips: support for MIPS SIMD architecture (MSA) has been added.
28 * mips: support for MIPS N32 ABI support has been added.
30 * ==================== TOOL CHANGES ====================
34 - Addition of a flag --delta-stacktrace=no|yes [yes on linux amd64/x86],
35 specifying how full history stack traces are computed. Using yes can speed
36 up Helgrind by 25% when using --history-level=full.
38 * ==================== OTHER CHANGES ====================
40 * A new configure option --enable-lto=yes allows to build Valgrind
41 with link time optimisation. If the toolchain supports it,
42 this produces a smaller/faster Valgrind (up to 10%).
43 Note that if you are doing some Valgrind development, --enable-lto=yes
44 significantly slows down the build.
46 * ==================== FIXED BUGS ====================
48 The following bugs have been fixed or resolved. Note that "n-i-bz"
49 stands for "not in bugzilla" -- that is, a bug that was reported to us
50 but never got a bugzilla entry. We encourage you to file bugs in
51 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
52 than mailing the developers (or mailing lists) directly -- bugs that
53 are not entered into bugzilla tend to get forgotten about or ignored.
55 To see details of a given bug, visit
56 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
57 where XXXXXX is the bug number as listed below.
59 79362 Debug info is lost for .so files when they are dlclose'd
60 208052 strlcpy error when n = 0
61 255603 exp-sgcheck Assertion '!already_present' failed
62 338252 building valgrind with -flto (link time optimisation) fails
63 345763 MIPS N32 ABI support
64 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
65 372347 Replacement problem of the additional c++14/c++17 new and delete operators
66 376257 helgrind history full speed up using a cached stack
67 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
69 379748 Fix missing pselect syscall (OS X 10.11)
70 380397 s390x: __GI_strcspn() replacemenet needed
71 381162 possible array overrun in VEX register allocator
72 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
73 381289 epoll_pwait can have a NULL sigmask
74 381274 powerpc too chatty even with --sigill-diagnostics=no
75 381553 VEX register allocator v3
76 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
77 381769 Use ucontext_t instead of struct ucontext
78 381805 arm32 needs ld.so index hardwire for new glibc security fixes
79 382256 gz compiler flag test doesn't work for gold
80 382407 vg_perf needs "--terse" command line option
81 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/mscoree.c
82 382563 MIPS MSA ASE support
83 382998 xml-socket doesn't work
84 383275 massif valgrind: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
85 383723 Fix missing kevent_qos syscall (macOS 10.11)
86 384096 Mention AddrCheck at Memcheck's command line option --undef-value-errors=no
87 384230 vex x86->IR: unhandled instruction bytes: 0x67 0xE8 0xAB 0x68
88 384526 reduce number of spill instructions generated by VEX register allocator v3
89 384584 Callee saved registers listed first for AMD64, X86, and PPC architectures
90 384633 Add a simple progress-reporting facility
91 384987 VEX register allocator: allocate caller-save registers for short lived vregs
92 385182 PPC64 is missing support for the DSCR
93 385207 PPC64, generate_store_FPRF() generates too many Iops
94 385208 PPC64, xxperm instruction exhausts temporary memory
95 385210 PPC64, vpermr instruction could exhaust temporary memory
96 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
97 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp instructions
98 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
99 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
100 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
101 385939 Optionally exit on the first error
102 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
103 387410 MIPSr6 support
104 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
105 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
106 388862 Add replacements for wmemchr and wcsnlen on Linux
107 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
108 389065 valgrind meets gcc flag -Wlogical-op
109 390471 suppression by specification of source-file line number
110 390723 make xtree dump files world wide readable, similar to log files
111 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
112 392449 callgrind not clearing the number of calls properly
113 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
114 lxssp, stxssp and stxvl instructions.
115 393023 callgrind_control risks using the wrong vgdb
116 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
117 393099 posix_memalign() invalid write if alignment == 0
118 395709 PPC64 is missing support for the xvnegsp instruction
120 n-i-bz Fix missing workq_ops operations (macOS)
121 n-i-bz fix bug in strspn replacement
122 n-i-bz Add support for the Linux BLKFLSBUF ioctl
123 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
124 n-i-bz Fix possible stack trashing by semctl syscall wrapping
125 n-i-bz Add support for the Linux membarrier() system call
127 Release 3.13.0 (15 June 2017)
128 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
130 3.13.0 is a feature release with many improvements and the usual collection of
133 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
134 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
135 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
136 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
138 * ==================== CORE CHANGES ===================
140 * The translation cache size has been increased to keep up with the demands of
141 large applications. The maximum number of sectors has increased from 24 to
142 48. The default number of sectors has increased from 16 to 32 on all
143 targets except Android, where the increase is from 6 to 12.
145 * The amount of memory that Valgrind can use has been increased from 64GB to
146 128GB. In particular this means your application can allocate up to about
147 60GB when running on Memcheck.
149 * Valgrind's default load address has been changed from 0x3800'0000 to
150 0x5800'0000, so as to make it possible to load larger executables. This
151 should make it possible to load executables of size at least 1200MB.
153 * A massive spaceleak caused by reading compressed debuginfo files has been
154 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
157 * The C++ demangler has been updated.
159 * Support for demangling Rust symbols has been added.
161 * A new representation of stack traces, the "XTree", has been added. An XTree
162 is a tree of stacktraces with data associated with the stacktraces. This is
163 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
164 consumption of your program. Reporting is controlled by the new options
165 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
167 A report can also be produced on demand using the gdbserver monitor command
168 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
169 format' and 'massif format. The existing visualisers for these formats (e.g.
170 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
171 analyse these reports.
173 Memcheck can also produce XTree leak reports using the Callgrind file
174 format. For more details, see the user manual.
176 * ================== PLATFORM CHANGES =================
178 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
180 * amd64: fixes for JIT failure problems on long AVX2 code blocks
182 * amd64 and x86: support for CET prefixes has been added
184 * arm32: a few missing ARMv8 instructions have been implemented
186 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
187 Store-Conditional instructions has been added. This is to deal with
188 processor implementations that implement the LL/SC specifications strictly
189 and as a result cause Valgrind to hang in certain situations. The
190 alternative implementation is automatically enabled at startup, as required.
191 You can use the option --sim-hints=fallback-llsc to force-enable it if you
194 * Support for OSX 10.12 has been improved.
196 * On Linux, clone handling has been improved to honour CLONE_VFORK that
197 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
198 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
199 CLONE_VM exact semantics will (still) not work.
201 * The TileGX/Linux port has been removed because it appears to be both unused
204 * ==================== TOOL CHANGES ====================
208 - Memcheck should give fewer false positives when running optimised
209 Clang/LLVM generated code.
211 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
213 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
214 to produce the end of execution leak report in a xtree callgrind format
217 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
218 the leak report in an xtree file.
222 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
224 - For some workloads (typically, for big applications), Massif memory
225 consumption and CPU consumption has decreased significantly.
229 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
231 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
232 for Ada gnat compiled applications.
234 * ==================== OTHER CHANGES ====================
236 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
237 append the inner guest stacktrace to the inner host stacktrace. This helps
238 to investigate the errors reported by the outer, when they are caused by the
239 inner guest program (such as an inner regtest). See README_DEVELOPERS for
242 * To allow fast detection of callgrind files by desktop environments and file
243 managers, the format was extended to have an optional first line that
244 uniquely identifies the format ("# callgrind format"). Callgrind creates
245 this line now, as does the new xtree functionality.
247 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
248 have a new %n format letter that is replaced by a sequence number.
250 * "--version -v" now shows the SVN revision numbers from which Valgrind was
253 * ==================== FIXED BUGS ====================
255 The following bugs have been fixed or resolved. Note that "n-i-bz"
256 stands for "not in bugzilla" -- that is, a bug that was reported to us
257 but never got a bugzilla entry. We encourage you to file bugs in
258 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
259 than mailing the developers (or mailing lists) directly -- bugs that
260 are not entered into bugzilla tend to get forgotten about or ignored.
262 To see details of a given bug, visit
263 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
264 where XXXXXX is the bug number as listed below.
266 162848 --log-file output isn't split when a program forks
267 340777 Illegal instruction on mips (ar71xx)
268 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
269 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
270 to a different stack.
271 344139 x86 stack-seg overrides, needed by the Wine people
272 344524 store conditional of guest applications always fail - observed on
274 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
275 352395 Please provide SVN revision info in --version -v
276 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
277 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
278 INVALID_POSIX_THREADID' failed
279 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
280 barrier implementation
281 358697 valgrind.h: Some code remains even when defining NVALGRIND
282 359202 Add musl libc configure/compile
283 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
284 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
285 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
286 362223 assertion failed when .valgrindrc is a directory instead of a file
287 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
288 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
289 368507 can't malloc chunks larger than about 34GB
290 368529 Android arm target link error, missing atexit and pthread_atfork
291 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
292 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
293 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
294 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
295 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
296 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
297 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
298 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
299 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
300 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
301 370028 Reduce the number of compiler warnings on MIPS platforms
302 370635 arm64 missing syscall getcpu
303 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
304 371227 Clean AArch64 syscall table
305 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
306 371471 Valgrind complains about non legit memory leaks on placement new (C++)
307 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
308 371503 disInstr(arm64): unhandled instruction 0xF89F0000
309 371869 support '%' in symbol Z-encoding
310 371916 execution tree xtree concept
311 372120 c++ demangler demangles symbols which are not c++
312 372185 Support of valgrind on ARMv8 with 32 bit executable
313 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
314 372195 Power PC, xxsel instruction is not always recognized.
315 372504 Hanging on exit_group
316 372600 process loops forever when fatal signals are arriving quickly
317 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
318 373046 Stacks registered by core are never deregistered
319 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
320 373086 Implement additional Xen hypercalls
321 373192 Calling posix_spawn in glibc 2.24 completely broken
322 373488 Support for fanotify API on ARM64 architecture
323 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
324 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
325 373938 const IRExpr arguments for matchIRExpr()
326 374719 some spelling fixes
327 374963 increase valgrind's load address to prevent mmap failure
328 375514 valgrind_get_tls_addr() does not work in case of static TLS
329 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
330 for ML_(find_rx_mapping)()
331 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
332 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
333 == 377159 "vex: the `impossible' happened" still present
334 == 375150 Assertion 'tres.status == VexTransOK' failed
335 == 378068 valgrind crashes on AVX2 function in FFmpeg
336 376142 Segfaults on MIPS Cavium Octeon boards
337 376279 disInstr(arm64): unhandled instruction 0xD50320FF
338 376455 Solaris: unhandled syscall lgrpsys(180)
339 376518 Solaris: unhandled fast trap getlgrp(6)
340 376611 ppc64 and arm64 don't know about prlimit64 syscall
341 376729 PPC64, remove R2 from the clobber list
343 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
344 to be wrongly marked as addressable
345 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
346 PIE enabled by default
347 377376 memcheck/tests/linux/getregset fails with glibc2.24
348 377427 PPC64, lxv instruction failing on odd destination register
349 377478 PPC64: ISA 3.0 setup fixes
350 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
351 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
352 and 2 args for FUTEX_TRYLOCK_PI
353 377717 Fix massive space leak when reading compressed debuginfo sections
354 377891 Update Xen 4.6 domctl wrappers
355 377930 fcntl syscall wrapper is missing flock structure check
356 378524 libvexmultiarch_test regression on s390x and ppc64
357 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
358 378673 Update libiberty demangler
359 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
360 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
361 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
362 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
363 (task_register_dyld_image_infos)
364 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
365 (task_register_dyld_shared_cache_image_info)
366 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
367 379473 MIPS: add support for rdhwr cycle counter register
368 379504 remove TileGX/Linux port
369 379525 Support more x86 nop opcodes
370 379838 disAMode(x86): not an addr!
371 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
372 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
373 379895 clock_gettime does not execute POST syscall wrapper
374 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
375 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
376 380200 xtree generated callgrind files refer to files without directory name
377 380202 Assertion failure for cache line size (cls == 64) on aarch64.
378 380397 s390x: __GI_strcspn() replacement needed
379 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
381 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
382 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
383 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
387 Release 3.12.0 (20 October 2016)
388 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
390 3.12.0 is a feature release with many improvements and the usual
391 collection of bug fixes.
393 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
394 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
395 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
396 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
397 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
398 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
400 * ================== PLATFORM CHANGES =================
402 * POWER: Support for ISA 3.0 has been added
404 * mips: support for O32 FPXX ABI has been added.
405 * mips: improved recognition of different processors
406 * mips: determination of page size now done at run time
408 * amd64: Partial support for AMD FMA4 instructions.
410 * arm, arm64: Support for v8 crypto and CRC instructions.
412 * Improvements and robustification of the Solaris port.
414 * Preliminary support for MacOS 10.12 (Sierra) has been added.
416 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
417 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
418 where possible. Valgrind's support for 32-bit x86 has stagnated in
419 recent years and has fallen far behind that for 64-bit x86
420 instructions. By contrast 64-bit x86 is well supported, up to and
423 * ==================== TOOL CHANGES ====================
427 - Added meta mempool support for describing a custom allocator which:
428 - Auto-frees all chunks assuming that destroying a pool destroys all
430 - Uses itself to allocate other memory blocks
432 - New flag --ignore-range-below-sp to ignore memory accesses below
433 the stack pointer, if you really have to. The related flag
434 --workaround-gcc296-bugs=yes is now deprecated. Use
435 --ignore-range-below-sp=1024-1 as a replacement.
439 - Improved thread startup time significantly on non-Linux platforms.
443 - Added collection of the metric "tot-blocks-allocd"
445 * ==================== OTHER CHANGES ====================
447 * Replacement/wrapping of malloc/new related functions is now done not just
448 for system libraries by default, but for any globally defined malloc/new
449 related function (both in shared libraries and statically linked alternative
450 malloc implementations). The dynamic (runtime) linker is excluded, though.
451 To only intercept malloc/new related functions in
452 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
453 "nouserintercepts" can be any non-existing library name).
454 This new functionality is not implemented for MacOS X.
456 * The maximum number of callers in a suppression entry is now equal to
457 the maximum size for --num-callers (500).
458 Note that --gen-suppressions=yes|all similarly generates suppressions
459 containing up to --num-callers frames.
461 * New and modified GDB server monitor features:
463 - Valgrind's gdbserver now accepts the command 'catch syscall'.
464 Note that you must have GDB >= 7.11 to use 'catch syscall' with
467 * New option --run-cxx-freeres=<yes|no> can be used to change whether
468 __gnu_cxx::__freeres() cleanup function is called or not. Default is
471 * Valgrind is able to read compressed debuginfo sections in two formats:
472 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
473 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
475 * Modest JIT-cost improvements: the cost of instrumenting code blocks
476 for the most common use case (x86_64-linux, Memcheck) has been
479 * Improved performance for programs that do a lot of discarding of
480 instruction address ranges of 8KB or less.
482 * The C++ symbol demangler has been updated.
484 * More robustness against invalid syscall parameters on Linux.
486 * ==================== FIXED BUGS ====================
488 The following bugs have been fixed or resolved. Note that "n-i-bz"
489 stands for "not in bugzilla" -- that is, a bug that was reported to us
490 but never got a bugzilla entry. We encourage you to file bugs in
491 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
492 than mailing the developers (or mailing lists) directly -- bugs that
493 are not entered into bugzilla tend to get forgotten about or ignored.
495 To see details of a given bug, visit
496 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
497 where XXXXXX is the bug number as listed below.
499 191069 Exiting due to signal not reported in XML output
500 199468 Suppressions: stack size limited to 25
501 while --num-callers allows more frames
502 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
503 278744 cvtps2pd with redundant RexW
504 303877 valgrind doesn't support compressed debuginfo sections.
505 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
506 348345 Assertion fails for negative lineno
507 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
508 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
509 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
510 351804 Crash on generating suppressions for "printf" call on OS X 10.10
511 352197 mips: mmap2() not wrapped correctly for page size > 4096
512 353083 arm64 doesn't implement various xattr system calls
513 353084 arm64 doesn't support sigpending system call
514 353137 www: update info for Supported Platforms
515 353138 www: update "The Valgrind Developers" page
516 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
519 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
520 353398 WARNING: unhandled amd64-solaris syscall: 207
521 353660 XML in auxwhat tag not escaping reserved symbols properly
522 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
523 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
524 353802 ELF debug info reader confused with multiple .rodata sections
525 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
526 353917 unhandled amd64-solaris syscall fchdir(120)
527 353920 unhandled amd64-solaris syscall: 170
528 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
529 354392 unhandled amd64-solaris syscall: 171
530 354797 Vbit test does not include Iops for Power 8 instruction support
531 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
535 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
536 355188 valgrind should intercept all malloc related global functions
537 355454 do not intercept malloc related symbols from the runtime linker
538 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
539 356044 Dwarf line info reader misinterprets is_stmt register
540 356112 mips: replace addi with addiu
541 356393 valgrind (vex) crashes because isZeroU happened
544 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
545 356678 arm64-linux: unhandled syscall 232 (mincore)
546 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
547 356823 Unsupported ARM instruction: stlex
548 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
549 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
550 357673 crash if I try to run valgrind with a binary link with libcurl
551 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
552 357871 pthread_spin_destroy not properly wrapped
553 357887 Calls to VG_(fclose) do not close the file descriptor
554 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
555 358030 support direct socket calls on x86 32bit (new in linux 4.3)
556 358478 drd/tests/std_thread.cpp doesn't build with GCC6
557 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
558 359181 Buffer Overflow during Demangling
559 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
560 359289 s390x: popcnt (B9E1) not implemented
561 359472 The Power PC vsubuqm instruction doesn't always give the correct result
562 359503 Add missing syscalls for aarch64 (arm64)
563 359645 "You need libc6-dbg" help message could be more helpful
564 359703 s390: wire up separate socketcalls system calls
565 359724 getsockname might crash - deref_UInt should call safe_to_deref
566 359733 amd64 implement ld.so strchr/index override like x86
567 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
568 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
570 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
571 359871 Incorrect mask handling in ppoll
572 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
573 360008 Contents of Power vr registers contents is not printed correctly when
574 the --vgdb-shadow-registers=yes option is used
575 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
577 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
578 360425 arm64 unsupported instruction ldpsw
580 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
581 360571 Error about the Android Runtime reading below the stack pointer on ARM
582 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
583 360749 kludge for multiple .rodata sections on Solaris no longer needed
584 360752 raise the number of reserved fds in m_main.c from 10 to 12
585 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
586 361226 s390x: risbgn (EC59) not implemented
587 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
588 361354 ppc64[le]: wire up separate socketcalls system calls
589 361615 Inconsistent termination for multithreaded process terminated by signal
590 361926 Unhandled Solaris syscall: sysfs(84)
591 362009 V dumps core on unimplemented functionality before threads are created
592 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
593 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
594 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
595 362953 Request for an update to the Valgrind Developers page
596 363680 add renameat2() support
597 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
598 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
599 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
600 364058 clarify in manual limitations of array overruns detections
601 364413 pselect sycallwrapper mishandles NULL sigmask
602 364728 Power PC, missing support for several HW registers in
603 get_otrack_shadow_offset_wrk()
604 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
605 365273 Invalid write to stack location reported after signal handler runs
606 365912 ppc64BE segfault during jm-insns test (RELRO)
607 366079 FPXX Support for MIPS32 Valgrind
608 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
609 366344 Multiple unhandled instruction for Aarch64
610 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
611 367995 Integration of memcheck with custom memory allocator
612 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
613 368412 False positive result for altivec capability check
614 368416 Add tc06_two_races_xml.exp output for ppc64
615 368419 Perf Events ioctls not implemented
616 368461 mmapunmap test fails on ppc64
617 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
618 369000 AMD64 fma4 instructions unsupported.
619 369169 ppc64 fails jm_int_isa_2_07 test
620 369175 jm_vec_isa_2_07 test crashes on ppc64
621 369209 valgrind loops and eats up all memory if cwd doesn't exist.
622 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
623 369359 msghdr_foreachfield can crash when handling bad iovec
624 369360 Bad sigprocmask old or new sets can crash valgrind
625 369361 vmsplice syscall wrapper crashes on bad iovec
626 369362 Bad sigaction arguments crash valgrind
627 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
628 369402 Bad set/get_thread_area pointer crashes valgrind
629 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
630 369446 valgrind crashes on unknown fcntl command
631 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
632 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
633 370265 ISA 3.0 HW cap stuff needs updating
634 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
635 372195 Power PC, xxsel instruction is not always recognized
637 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
638 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
639 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
640 n-i-bz Fix ppoll_alarm exclusion on OS X
641 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
642 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
643 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
644 n-i-bz amd64: memcheck false positive with shr %edx
645 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
646 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
647 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
648 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
650 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
651 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
652 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
656 Release 3.11.0 (22 September 2015)
657 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
659 3.11.0 is a feature release with many improvements and the usual
660 collection of bug fixes.
662 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
663 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
664 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
665 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
666 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
667 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
669 * ================== PLATFORM CHANGES =================
671 * Support for Solaris/x86 and Solaris/amd64 has been added.
673 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
675 * Preliminary support for the Tilera TileGX architecture has been added.
677 * s390x: It is now required for the host to have the "long displacement"
678 facility. The oldest supported machine model is z990.
680 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
681 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
683 * The JIT's register allocator is significantly faster, making the JIT
684 as a whole somewhat faster, so JIT-intensive activities, for example
685 program startup, are modestly faster, around 5%.
687 * There have been changes to the default settings of several command
688 line flags, as detailed below.
690 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
691 capable hosts, the simulated CPUID will now indicate AVX2 support.
693 * ==================== TOOL CHANGES ====================
697 - The default value for --leak-check-heuristics has been changed from
698 "none" to "all". This helps to reduce the number of possibly
699 lost blocks, in particular for C++ applications.
701 - The default value for --keep-stacktraces has been changed from
702 "malloc-then-free" to "malloc-and-free". This has a small cost in
703 memory (one word per malloc-ed block) but allows Memcheck to show the
704 3 stacktraces of a dangling reference: where the block was allocated,
705 where it was freed, and where it is acccessed after being freed.
707 - The default value for --partial-loads-ok has been changed from "no" to
708 "yes", so as to avoid false positive errors resulting from some kinds
711 - A new monitor command 'xb <addr> <len>' shows the validity bits of
712 <len> bytes at <addr>. The monitor command 'xb' is easier to use
713 than get_vbits when you need to associate byte data value with
714 their corresponding validity bits.
716 - The 'block_list' monitor command has been enhanced:
717 o it can print a range of loss records
718 o it now accepts an optional argument 'limited <max_blocks>'
719 to control the number of blocks printed.
720 o if a block has been found using a heuristic, then
721 'block_list' now shows the heuristic after the block size.
722 o the loss records/blocks to print can be limited to the blocks
723 found via specified heuristics.
725 - The C helper functions used to instrument loads on
726 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
727 replaced by handwritten assembly sequences. This gives speedups
728 in the region of 0% to 7% for those targets only.
730 - A new command line option, --expensive-definedness-checks=yes|no,
731 has been added. This is useful for avoiding occasional invalid
732 uninitialised-value errors in optimised code. Watch out for
733 runtime degradation, as this can be up to 25%. As always, though,
734 the slowdown is highly application specific. The default setting
739 - A new monitor command 'all_snapshots <filename>' dumps all
740 snapshots taken so far.
744 - Significant memory reduction and moderate speedups for
745 --history-level=full for applications accessing a lot of memory
746 with many different stacktraces.
748 - The default value for --conflict-cache-size=N has been doubled to
749 2000000. Users that were not using the default value should
750 preferably also double the value they give.
752 The default was changed due to the changes in the "full history"
753 implementation. Doubling the value gives on average a slightly more
754 complete history and uses similar memory (or significantly less memory
755 in the worst case) than the previous implementation.
757 - The Helgrind monitor command 'info locks' now accepts an optional
758 argument 'lock_addr', which shows information about the lock at the
761 - When using --history-level=full, the new Helgrind monitor command
762 'accesshistory <addr> [<len>]' will show the recorded accesses for
763 <len> (or 1) bytes at <addr>.
765 * ==================== OTHER CHANGES ====================
767 * The default value for the --smc-check option has been changed from
768 "stack" to "all-non-file" on targets that provide automatic D-I
769 cache coherence (x86, amd64 and s390x). The result is to provide,
770 by default, transparent support for JIT generated and self-modifying
773 * Mac OS X only: the default value for the --dsymutil option has been
774 changed from "no" to "yes", since any serious usage on Mac OS X
775 always required it to be "yes".
777 * The command line options --db-attach and --db-command have been removed.
778 They were deprecated in 3.10.0.
780 * When a process dies due to a signal, Valgrind now shows the signal
781 and the stacktrace at default verbosity (i.e. verbosity 1).
783 * The address description logic used by Memcheck and Helgrind now
784 describes addresses in anonymous segments, file mmap-ed segments,
785 shared memory segments and the brk data segment.
787 * The new option --error-markers=<begin>,<end> can be used to mark the
788 begin/end of errors in textual output mode, to facilitate
789 searching/extracting errors in output files that mix valgrind errors
792 * The new option --max-threads=<number> can be used to change the number
793 of threads valgrind can handle. The default is 500 threads which
794 should be more than enough for most applications.
796 * The new option --valgrind-stacksize=<number> can be used to change the
797 size of the private thread stacks used by Valgrind. This is useful
798 for reducing memory use or increasing the stack size if Valgrind
799 segfaults due to stack overflow.
801 * The new option --avg-transtab-entry-size=<number> can be used to specify
802 the expected instrumented block size, either to reduce memory use or
803 to avoid excessive retranslation.
805 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
807 * New and modified GDB server monitor features:
809 - When a signal is reported in GDB, you can now use the GDB convenience
810 variable $_siginfo to examine detailed signal information.
812 - Valgrind's gdbserver now allows the user to change the signal
813 to deliver to the process. So, use 'signal SIGNAL' to continue execution
814 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
815 continue without passing the signal to the process.
817 - With GDB >= 7.10, the command 'target remote'
818 will automatically load the executable file of the process running
819 under Valgrind. This means you do not need to specify the executable
820 file yourself, GDB will discover it itself. See GDB documentation about
821 'qXfer:exec-file:read' packet for more info.
823 * ==================== FIXED BUGS ====================
825 The following bugs have been fixed or resolved. Note that "n-i-bz"
826 stands for "not in bugzilla" -- that is, a bug that was reported to us
827 but never got a bugzilla entry. We encourage you to file bugs in
828 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
829 than mailing the developers (or mailing lists) directly -- bugs that
830 are not entered into bugzilla tend to get forgotten about or ignored.
832 To see details of a given bug, visit
833 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
834 where XXXXXX is the bug number as listed below.
836 116002 VG_(printf): Problems with justification of strings and integers
837 155125 avoid cutting away file:lineno after long function name
838 197259 Unsupported arch_prtctl PR_SET_GS option
839 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
840 201216 Fix Valgrind does not support pthread_sigmask() on OS X
841 201435 Fix Darwin: -v does not show kernel version
842 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
843 211256 Fixed an outdated comment regarding the default platform.
844 211529 Incomplete call stacks for code compiled by newer versions of MSVC
845 211926 Avoid compilation warnings in valgrind.h with -pedantic
846 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
848 226609 Crediting upstream authors in man page
849 231257 Valgrind omits path when executing script from shebang line
850 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
851 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
852 269360 s390x: Fix addressing mode selection for compare-and-swap
853 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
855 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
856 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
857 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
858 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
859 330147 libmpiwrap PMPI_Get_count returns undefined value
860 333051 mmap of huge pages fails due to incorrect alignment
862 334802 valgrind does not always explain why a given option is bad
863 335618 mov.w rN, pc/sp (ARM32)
864 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
869 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
870 338602 AVX2 bit in CPUID missing
871 338606 Strange message for scripts with invalid interpreter
872 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
873 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
874 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
876 339156 gdbsrv not called for fatal signal
877 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
878 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
879 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
880 339442 Fix testsuite build failure on OS X 10.9
881 339542 Enable compilation with Intel's ICC compiler
882 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
883 339688 Mac-specific ASM does not support .version directive (cpuid,
884 tronical and pushfpopf tests)
885 339745 Valgrind crash when check Marmalade app (partial fix)
886 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
887 339778 Linux/TileGx platform support for Valgrind
888 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
889 339789 Fix none/tests/execve test on Mac OS X 10.9
890 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
891 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
892 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
893 340392 Allow user to select more accurate definedness checking in memcheck
894 to avoid invalid complaints on optimised code
895 340430 Fix some grammatical weirdness in the manual.
896 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
897 341419 Signal handler ucontext_t not filled out correctly on OS X
898 341539 VG_(describe_addr) should not describe address as belonging to client
899 segment if it is past the heap end
900 341613 Enable building of manythreads and thread-exits tests on Mac OS X
901 341615 Fix none/tests/darwin/access_extended test on Mac OS X
902 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
903 341789 aarch64: shmat fails with valgrind on ARMv8
904 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
905 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
906 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
907 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
908 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
909 342221 socket connect false positive uninit memory for unknown af family
910 342353 Allow dumping full massif output while valgrind is still running
911 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
915 342603 Add I2C_SMBUS ioctl support
916 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
917 342683 Mark memory past the initial brk limit as unaddressable
918 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
919 342795 Internal glibc __GI_mempcpy call should be intercepted
920 342841 s390x: Support instructions fiebr(a) and fidbr(a)
921 343012 Unhandled syscall 319 (memfd_create)
922 343069 Patch updating v4l2 API support
923 343173 helgrind crash during stack unwind
924 343219 fix GET_STARTREGS for arm
925 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
926 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
927 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
928 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
929 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
930 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
931 343597 ppc64le: incorrect use of offseof macro
932 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
933 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
934 343732 Unhandled syscall 144 (setgid) on aarch64
935 343733 Unhandled syscall 187 (msgctl and related) on aarch64
936 343802 s390x: False positive "conditional jump or move depends on [..]
937 343902 --vgdb=yes doesn't break when --xml=yes is used
938 343967 Don't warn about setuid/setgid/setcap executable for directories
939 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
940 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
941 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
942 344054 www - update info for Solaris/illumos
943 344416 'make regtest' does not work cleanly on OS X
944 344235 Remove duplicate include of pub_core_aspacemgr.h
945 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
946 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
947 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
948 344314 callgrind_annotate ... warnings about commands containing newlines
949 344318 socketcall should wrap recvmmsg and sendmmsg
950 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
951 344416 Fix 'make regtest' does not work cleanly on OS X
952 344499 Fix compilation for Linux kernel >= 4.0.0
953 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
954 unix:349 (__pthread_fchdir)
955 344559 Garbage collection of unused segment names in address space manager
956 344560 Fix stack traces missing penultimate frame on OS X
957 344621 Fix memcheck/tests/err_disable4 test on OS X
958 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
959 344702 Fix missing libobjc suppressions on OS X 10.10
961 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
962 344939 Fix memcheck/tests/xml1 on OS X 10.10
963 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
964 345079 Fix build problems in VEX/useful/test_main.c
965 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
966 345177 arm64: prfm (reg) not implemented
967 345215 Performance improvements for the register allocator
968 345248 add support for Solaris OS in valgrind
969 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
970 345394 Fix memcheck/tests/strchr on OS X
971 345637 Fix memcheck/tests/sendmsg on OS X
972 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
973 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
974 345887 Fix an assertion in the address space manager
975 345928 amd64: callstack only contains current function for small stacks
976 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
977 345987 MIPS64: Implement cavium LHX instruction
978 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
979 346185 Fix typo saving altivec register v24
980 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
981 and LibVEX_GuestPPC64_get_CR()
982 346270 Regression tests none/tests/jm_vec/isa_2_07 and
983 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
984 346307 fuse filesystem syscall deadlocks
985 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
986 346411 MIPS: SysRes::_valEx handling is incorrect
987 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
988 346474 PPC64 Power 8, spr TEXASRU register not supported
989 346487 Compiler generates "note" about a future ABI change for PPC64
990 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
991 and causing spurious "invalid read of size 8" warnings
992 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
993 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
994 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
995 347322 Power PC regression test cleanup
996 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
998 347389 unhandled syscall: 373 (Linux ARM syncfs)
999 347686 Patch set to cleanup PPC64 regtests
1000 347978 Remove bash dependencies where not needed
1001 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1002 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1004 348102 Patch updating v4l2 API support
1005 348247 amd64 front end: jno jumps wrongly when overflow is not set
1006 348269 Improve mmap MAP_HUGETLB support.
1007 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1008 348345 Assertion fails for negative lineno
1009 348377 Unsupported ARM instruction: yield
1010 348565 Fix detection of command line option availability for clang
1011 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1012 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1013 348748 Fix redundant condition
1014 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1015 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1016 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1017 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1018 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1019 349626 Implemented additional Xen hypercalls
1020 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1021 349790 Clean up of the hardware capability checking utilities.
1022 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1023 349874 Fix typos in source code
1024 349879 memcheck: add handwritten assembly for helperc_LOADV*
1025 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1026 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1027 350202 Add limited param to 'monitor block_list'
1028 350290 s390x: Support instructions fixbr(a)
1029 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1030 350809 Fix none/tests/async-sigs for Solaris
1031 350811 Remove reference to --db-attach which has been removed.
1032 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1033 350854 hard-to-understand code in VG_(load_ELF)()
1034 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1035 351386 Solaris: Cannot run ld.so.1 under Valgrind
1036 351474 Fix VG_(iseqsigset) as obvious
1037 351531 Typo in /include/vki/vki-xen-physdev.h header guard
1038 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
1039 351858 ldsoexec support on Solaris
1040 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
1041 352130 helgrind reports false races for printfs using mempcpy on FILE* state
1042 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
1043 352320 arm64 crash on none/tests/nestedfs
1044 352765 Vbit test fails on Power 6
1045 352768 The mbar instruction is missing from the Power PC support
1046 352769 Power PC program priority register (PPR) is not supported
1047 n-i-bz Provide implementations of certain compiler builtins to support
1048 compilers that may not provide those
1049 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
1050 n-i-bz Fix compilation on distros with glibc < 2.5
1051 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
1052 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
1053 n-i-bz mremap did not work properly on shared memory
1054 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
1055 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
1057 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
1058 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
1059 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
1063 Release 3.10.1 (25 November 2014)
1064 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1065 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
1066 and backports fixes for all reported missing AArch64 ARMv8 instructions
1067 and syscalls from the trunk. If you package or deliver 3.10.0 for others
1068 to use, you might want to consider upgrading to 3.10.1 instead.
1070 The following bugs have been fixed or resolved. Note that "n-i-bz"
1071 stands for "not in bugzilla" -- that is, a bug that was reported to us
1072 but never got a bugzilla entry. We encourage you to file bugs in
1073 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1074 than mailing the developers (or mailing lists) directly -- bugs that
1075 are not entered into bugzilla tend to get forgotten about or ignored.
1077 To see details of a given bug, visit
1078 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1079 where XXXXXX is the bug number as listed below.
1081 335440 arm64: ld1 (single structure) is not implemented
1082 335713 arm64: unhanded instruction: prfm (immediate)
1083 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
1084 339182 ppc64: AvSplat ought to load destination vector register with [..]
1085 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
1086 339433 ppc64 lxvw4x instruction uses four 32-byte loads
1087 339645 Use correct tag names in sys_getdents/64 wrappers
1088 339706 Fix false positive for ioctl(TIOCSIG) on linux
1089 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
1090 339853 arm64 times syscall unknown
1091 339855 arm64 unhandled getsid/setsid syscalls
1092 339858 arm64 dmb sy not implemented
1093 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
1094 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
1095 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
1097 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
1098 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
1099 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
1100 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
1101 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
1102 340509 arm64: unhandled instruction fcvtas
1103 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
1104 340632 arm64: unhandled instruction fcvtas
1105 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
1106 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
1107 340788 warning: unhandled syscall: 318 (getrandom)
1108 340807 disInstr(arm): unhandled instruction: 0xEE989B20
1109 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
1110 340922 arm64: unhandled getgroups/setgroups syscalls
1111 350251 Fix typo in VEX utility program (test_main.c).
1112 350407 arm64: unhandled instruction ucvtf (vector, integer)
1113 350809 none/tests/async-sigs breaks when run under cron on Solaris
1114 350811 update README.solaris after r15445
1115 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
1116 350854 strange code in VG_(load_ELF)()
1117 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1118 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
1119 n-i-bz Add missing ]] to terminate CDATA.
1120 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
1121 n-i-bz Enable sys_fadvise64_64 on arm32.
1122 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
1123 n-i-bz Add test cases for all known arm64 load/store instructions.
1124 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
1125 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
1126 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
1127 n-i-bz arm64: Implement "fcvtpu w, s".
1128 n-i-bz arm64: implement ADDP and various others
1129 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
1130 n-i-bz arm64: enable FCVT{A,N}S X,S.
1132 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
1136 Release 3.10.0 (10 September 2014)
1137 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1139 3.10.0 is a feature release with many improvements and the usual
1140 collection of bug fixes.
1142 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1143 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1144 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
1145 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
1146 significantly improved relative to the 3.9.0 release.
1148 * ================== PLATFORM CHANGES =================
1150 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
1151 is mostly complete, and is usable, but some SIMD instructions are as
1154 * Support for little-endian variant of the 64-bit POWER architecture.
1156 * Support for Android on MIPS32.
1158 * Support for 64bit FPU on MIPS32 platforms.
1160 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
1162 * Configuration for and running on Android targets has changed.
1163 See README.android in the source tree for details.
1165 * ================== DEPRECATED FEATURES =================
1167 * --db-attach is now deprecated and will be removed in the next
1168 valgrind feature release. The built-in GDB server capabilities are
1169 superior and should be used instead. Learn more here:
1170 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
1172 * ==================== TOOL CHANGES ====================
1176 - Client code can now selectively disable and re-enable reporting of
1177 invalid address errors in specific ranges using the new client
1178 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
1179 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
1181 - Leak checker: there is a new leak check heuristic called
1182 "length64". This is used to detect interior pointers pointing 8
1183 bytes inside a block, on the assumption that the first 8 bytes
1184 holds the value "block size - 8". This is used by
1185 sqlite3MemMalloc, for example.
1187 - Checking of system call parameters: if a syscall parameter
1188 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
1189 several fields not initialised, an error is now reported for each
1190 field. Previously, an error was reported only for the first
1191 uninitialised field.
1193 - Mismatched alloc/free checking: a new flag
1194 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
1195 such checks if necessary.
1199 - Improvements to error messages:
1201 o Race condition error message involving heap allocated blocks also
1202 show the thread number that allocated the raced-on block.
1204 o All locks referenced by an error message are now announced.
1205 Previously, some error messages only showed the lock addresses.
1207 o The message indicating where a lock was first observed now also
1208 describes the address/location of the lock.
1210 - Helgrind now understands the Ada task termination rules and
1211 creates a happens-before relationship between a terminated task
1212 and its master. This avoids some false positives and avoids a big
1213 memory leak when a lot of Ada tasks are created and terminated.
1214 The interceptions are only activated with forthcoming releases of
1215 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
1217 - A new GDB server monitor command "info locks" giving the list of
1218 locks, their location, and their status.
1222 - callgrind_control now supports the --vgdb-prefix argument,
1223 which is needed if valgrind was started with this same argument.
1225 * ==================== OTHER CHANGES ====================
1227 * Unwinding through inlined function calls. Stack unwinding can now
1228 make use of Dwarf3 inlined-unwind information if it is available.
1229 The practical effect is that inlined calls become visible in stack
1230 traces. The suppression matching machinery has been adjusted
1231 accordingly. This is controlled by the new option
1232 --read-inline-info=yes|no. Currently this is enabled by default
1233 only on Linux and Android targets and only for the tools Memcheck,
1236 * Valgrind can now read EXIDX unwind information on 32-bit ARM
1237 targets. If an object contains both CFI and EXIDX unwind
1238 information, Valgrind will prefer the CFI over the EXIDX. This
1239 facilitates unwinding through system libraries on arm-android
1242 * Address description logic has been improved and is now common
1243 between Memcheck and Helgrind, resulting in better address
1244 descriptions for some kinds of error messages.
1246 * Error messages about dubious arguments (eg, to malloc or calloc) are
1247 output like other errors. This means that they can be suppressed
1248 and they have a stack trace.
1250 * The C++ demangler has been updated for better C++11 support.
1252 * New and modified GDB server monitor features:
1254 - Thread local variables/storage (__thread) can now be displayed.
1256 - The GDB server monitor command "v.info location <address>"
1257 displays information about an address. The information produced
1258 depends on the tool and on the options given to valgrind.
1259 Possibly, the following are described: global variables, local
1260 (stack) variables, allocated or freed blocks, ...
1262 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
1263 ask the GDB server to stop at the start of program execution, at
1264 the end of the program execution and on Valgrind internal errors.
1266 - A new monitor command "v.info stats" shows various Valgrind core
1267 and tool statistics.
1269 - A new monitor command "v.set hostvisibility" allows the GDB server
1270 to provide access to Valgrind internal host status/memory.
1272 * A new option "--aspace-minaddr=<address>" can in some situations
1273 allow the use of more memory by decreasing the address above which
1274 Valgrind maps memory. It can also be used to solve address
1275 conflicts with system libraries by increasing the default value.
1276 See user manual for details.
1278 * The amount of memory used by Valgrind to store debug info (unwind
1279 info, line number information and symbol data) has been
1280 significantly reduced, even though Valgrind now reads more
1281 information in order to support unwinding of inlined function calls.
1283 * Dwarf3 handling with --read-var-info=yes has been improved:
1285 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
1287 - Code compiled with
1288 -ffunction-sections -fdata-sections -Wl,--gc-sections
1289 no longer causes assertion failures.
1291 * Improved checking for the --sim-hints= and --kernel-variant=
1292 options. Unknown strings are now detected and reported to the user
1295 * The semantics of stack start/end boundaries in the valgrind.h
1296 VALGRIND_STACK_REGISTER client request has been clarified and
1297 documented. The convention is that start and end are respectively
1298 the lowest and highest addressable bytes of the stack.
1300 * ==================== FIXED BUGS ====================
1302 The following bugs have been fixed or resolved. Note that "n-i-bz"
1303 stands for "not in bugzilla" -- that is, a bug that was reported to us
1304 but never got a bugzilla entry. We encourage you to file bugs in
1305 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1306 than mailing the developers (or mailing lists) directly -- bugs that
1307 are not entered into bugzilla tend to get forgotten about or ignored.
1309 To see details of a given bug, visit
1310 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1311 where XXXXXX is the bug number as listed below.
1313 175819 Support for ipv6 socket reporting with --track-fds
1314 232510 make distcheck fails
1315 249435 Analyzing wine programs with callgrind triggers a crash
1316 278972 support for inlined function calls in stacktraces and suppression
1318 291310 FXSAVE instruction marks memory as undefined on amd64
1319 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
1320 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
1321 315199 vgcore file for threaded app does not show which thread crashed
1322 315952 tun/tap ioctls are not supported
1323 323178 Unhandled instruction: PLDW register (ARM)
1324 323179 Unhandled instruction: PLDW immediate (ARM)
1325 324050 Helgrind: SEGV because of unaligned stack when using movdqa
1326 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
1327 325124 [MIPSEL] Compilation error
1328 325477 Phase 4 support for IBM Power ISA 2.07
1329 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
1330 325628 Phase 5 support for IBM Power ISA 2.07
1331 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
1332 325751 Missing the two privileged Power PC Transactional Memory Instructions
1333 325816 Phase 6 support for IBM Power ISA 2.07
1334 325856 Make SGCheck fail gracefully on unsupported platforms
1335 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
1336 326436 DRD: False positive in libstdc++ std::list::push_back
1337 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
1338 326462 Refactor vgdb to isolate invoker stuff into separate module
1339 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
1340 326623 DRD: false positive conflict report in a field assignment
1341 326724 Valgrind does not compile on OSX 1.9 Mavericks
1342 326816 Intercept for __strncpy_sse2_unaligned missing?
1343 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
1344 326983 Clear direction flag after tests on amd64.
1345 327212 Do not prepend the current directory to absolute path names.
1346 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
1347 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
1348 327284 s390x: Fix translation of the risbg instruction
1349 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
1350 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
1351 327916 DW_TAG_typedef may have no name
1352 327943 s390x: add a redirection for the 'index' function
1353 328100 XABORT not implemented
1354 328205 Implement additional Xen hypercalls
1355 328454 add support Backtraces with ARM unwind tables (EXIDX)
1356 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
1357 328711 valgrind.1 manpage "memcheck options" section is badly generated
1358 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
1359 329612 Incorrect handling of AT_BASE for image execution
1360 329694 clang warns about using uninitialized variable
1361 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
1362 330228 mmap must align to VKI_SHMLBA on mips32
1363 330257 LLVM does not support `-mno-dynamic-no-pic` option
1364 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
1365 330459 --track-fds=yes doesn't track eventfds
1366 330469 Add clock_adjtime syscall support
1367 330594 Missing sysalls on PowerPC / uClibc
1368 330622 Add test to regression suite for POWER instruction: dcbzl
1369 330939 Support for AMD's syscall instruction on x86
1371 330941 Typo in PRE(poll) syscall wrapper
1372 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
1373 331254 Fix expected output for memcheck/tests/dw4
1374 331255 Fix race condition in test none/tests/coolo_sigaction
1375 331257 Fix type of jump buffer in test none/tests/faultstatus
1376 331305 configure uses bash specific syntax
1377 331337 s390x WARNING: unhandled syscall: 326 (dup3)
1378 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
1379 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
1380 331829 Unexpected ioctl opcode sign extension
1381 331830 ppc64: WARNING: unhandled syscall: 96/97
1382 331839 drd/tests/sem_open specifies invalid semaphore name
1383 331847 outcome of drd/tests/thread_name is nondeterministic
1384 332037 Valgrind cannot handle Thumb "add pc, reg"
1385 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
1386 consistency checks enabled
1387 332263 intercepts for pthread_rwlock_timedrdlock and
1388 pthread_rwlock_timedwrlock are incorrect
1389 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
1391 332276 Implement additional Xen hypercalls
1392 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
1393 332765 Fix ms_print to create temporary files in a proper directory
1394 333072 drd: Add semaphore annotations
1395 333145 Tests for missaligned PC+#imm access for arm
1396 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
1397 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
1398 333248 WARNING: unhandled syscall: unix:443
1399 333428 ldr.w pc [rD, #imm] instruction leads to assertion
1400 333501 cachegrind: assertion: Cache set count is not a power of two.
1403 333666 Recognize MPX instructions and bnd prefix.
1404 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
1405 333817 Valgrind reports the memory areas written to by the SG_IO
1407 334049 lzcnt fails silently (x86_32)
1408 334384 Valgrind does not have support Little Endian support for
1410 334585 recvmmsg unhandled (+patch) (arm)
1411 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
1412 334727 Build fails with -Werror=format-security
1413 334788 clarify doc about --log-file initial program directory
1414 334834 PPC64 Little Endian support, patch 2
1415 334836 PPC64 Little Endian support, patch 3 testcase fixes
1416 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
1417 335034 Unhandled ioctl: HCIGETDEVLIST
1418 335155 vgdb, fix error print statement.
1419 335262 arm64: movi 8bit version is not supported
1420 335263 arm64: dmb instruction is not implemented
1421 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
1422 335496 arm64: sbc/abc instructions are not implemented
1423 335554 arm64: unhandled instruction: abs
1424 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
1425 335735 arm64: unhandled instruction: cnt
1426 335736 arm64: unhandled instruction: uaddlv
1427 335848 arm64: unhandled instruction: {s,u}cvtf
1428 335902 arm64: unhandled instruction: sli
1429 335903 arm64: unhandled instruction: umull (vector)
1430 336055 arm64: unhandled instruction: mov (element)
1431 336062 arm64: unhandled instruction: shrn{,2}
1432 336139 mip64: [...] valgrind hangs and spins on a single core [...]
1433 336189 arm64: unhandled Instruction: mvn
1434 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
1435 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
1436 336772 Make moans about unknown ioctls more informative
1437 336957 Add a section about the Solaris/illumos port on the webpage
1438 337094 ifunc wrapper is broken on ppc64
1439 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
1440 337528 leak check heuristic for block prefixed by length as 64bit number
1441 337740 Implement additional Xen hypercalls
1442 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
1443 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
1444 337871 deprecate --db-attach
1445 338023 Add support for all V4L2/media ioctls
1446 338024 inlined functions are not shown if DW_AT_ranges is used
1447 338106 Add support for 'kcmp' syscall
1448 338115 DRD: computed conflict set differs from actual after fork
1449 338160 implement display of thread local storage in gdbsrv
1450 338205 configure.ac and check for -Wno-tautological-compare
1451 338300 coredumps are missing one byte of every segment
1452 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
1453 338499 --sim-hints parsing broken due to wrong order in tokens
1454 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
1455 338681 Unable to unwind through clone thread created on i386-linux
1456 338698 race condition between gdbsrv and vgdb on startup
1457 338703 helgrind on arm-linux gets false positives in dynamic loader
1458 338791 alt dwz files can be relative of debug/main file
1459 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
1460 338932 build V-trunk with gcc-trunk
1461 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
1462 345079 Fix build problems in VEX/useful/test_main.c
1463 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
1464 n-i-bz s390x: Fix memory corruption for multithreaded applications
1465 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
1466 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
1467 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
1468 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
1469 n-i-bz Add command-line option --default-suppressions
1470 n-i-bz Add support for BLKDISCARDZEROES ioctl
1471 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
1472 n-i-bz Add support for sys_pivot_root and sys_unshare
1474 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
1475 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
1476 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
1480 Release 3.9.0 (31 October 2013)
1481 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1482 3.9.0 is a feature release with many improvements and the usual
1483 collection of bug fixes.
1485 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
1486 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
1487 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
1488 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
1490 * ================== PLATFORM CHANGES =================
1492 * Support for MIPS64 LE and BE running Linux. Valgrind has been
1493 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
1495 * Support for MIPS DSP ASE on MIPS32 platforms.
1497 * Support for s390x Decimal Floating Point instructions on hosts that
1498 have the DFP facility installed.
1500 * Support for POWER8 (Power ISA 2.07) instructions
1502 * Support for Intel AVX2 instructions. This is available only on 64
1505 * Initial support for Intel Transactional Synchronization Extensions,
1508 * Initial support for Hardware Transactional Memory on POWER.
1510 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
1511 run large GUI apps tolerably well.
1513 * ==================== TOOL CHANGES ====================
1517 - Improvements in handling of vectorised code, leading to
1518 significantly fewer false error reports. You need to use the flag
1519 --partial-loads-ok=yes to get the benefits of these changes.
1521 - Better control over the leak checker. It is now possible to
1522 specify which leak kinds (definite/indirect/possible/reachable)
1523 should be displayed, which should be regarded as errors, and which
1524 should be suppressed by a given leak suppression. This is done
1525 using the options --show-leak-kinds=kind1,kind2,..,
1526 --errors-for-leak-kinds=kind1,kind2,.. and an optional
1527 "match-leak-kinds:" line in suppression entries, respectively.
1529 Note that generated leak suppressions contain this new line and
1530 are therefore more specific than in previous releases. To get the
1531 same behaviour as previous releases, remove the "match-leak-kinds:"
1532 line from generated suppressions before using them.
1534 - Reduced "possible leak" reports from the leak checker by the use
1535 of better heuristics. The available heuristics provide detection
1536 of valid interior pointers to std::stdstring, to new[] allocated
1537 arrays with elements having destructors and to interior pointers
1538 pointing to an inner part of a C++ object using multiple
1539 inheritance. They can be selected individually using the
1540 option --leak-check-heuristics=heur1,heur2,...
1542 - Better control of stacktrace acquisition for heap-allocated
1543 blocks. Using the --keep-stacktraces option, it is possible to
1544 control independently whether a stack trace is acquired for each
1545 allocation and deallocation. This can be used to create better
1546 "use after free" errors or to decrease Valgrind's resource
1547 consumption by recording less information.
1549 - Better reporting of leak suppression usage. The list of used
1550 suppressions (shown when the -v option is given) now shows, for
1551 each leak suppressions, how many blocks and bytes it suppressed
1552 during the last leak search.
1556 - False errors resulting from the use of statically initialised
1557 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
1560 - False errors resulting from the use of pthread_cond_waits that
1561 timeout, have been removed.
1563 * ==================== OTHER CHANGES ====================
1565 * Some attempt to tune Valgrind's space requirements to the expected
1566 capabilities of the target:
1568 - The default size of the translation cache has been reduced from 8
1569 sectors to 6 on Android platforms, since each sector occupies
1570 about 40MB when using Memcheck.
1572 - The default size of the translation cache has been increased to 16
1573 sectors on all other platforms, reflecting the fact that large
1574 applications require instrumentation and storage of huge amounts
1575 of code. For similar reasons, the number of memory mapped
1576 segments that can be tracked has been increased by a factor of 6.
1578 - In all cases, the maximum number of sectors in the translation
1579 cache can be controlled by the new flag --num-transtab-sectors.
1581 * Changes in how debug info (line numbers, etc) is read:
1583 - Valgrind no longer temporarily mmaps the entire object to read
1584 from it. Instead, reading is done through a small fixed sized
1585 buffer. This avoids virtual memory usage spikes when Valgrind
1586 reads debuginfo from large shared objects.
1588 - A new experimental remote debug info server. Valgrind can read
1589 debug info from a different machine (typically, a build host)
1590 where debuginfo objects are stored. This can save a lot of time
1591 and hassle when running Valgrind on resource-constrained targets
1592 (phones, tablets) when the full debuginfo objects are stored
1593 somewhere else. This is enabled by the --debuginfo-server=
1596 - Consistency checking between main and debug objects can be
1597 disabled using the --allow-mismatched-debuginfo option.
1599 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
1600 scanning can recover stack traces in some cases when the normal
1601 unwind mechanisms fail. Stack scanning is best described as "a
1602 nasty, dangerous and misleading hack" and so is disabled by default.
1603 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
1606 * Detection and merging of recursive stack frame cycles. When your
1607 program has recursive algorithms, this limits the memory used by
1608 Valgrind for recorded stack traces and avoids recording
1609 uninteresting repeated calls. This is controlled by the command
1610 line option --merge-recursive-frame and by the monitor command
1611 "v.set merge-recursive-frames".
1613 * File name and line numbers for used suppressions. The list of used
1614 suppressions (shown when the -v option is given) now shows, for each
1615 used suppression, the file name and line number where the suppression
1618 * New and modified GDB server monitor features:
1620 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
1621 that can be used to execute gdbserver monitor commands from the
1624 - A new monitor command, "v.info open_fds", that gives the list of
1625 open file descriptors and additional details.
1627 - An optional message in the "v.info n_errs_found" monitor command,
1628 for example "v.info n_errs_found test 1234 finished", allowing a
1629 comment string to be added to the process output, perhaps for the
1630 purpose of separating errors of different tests or test phases.
1632 - A new monitor command "v.info execontext" that shows information
1633 about the stack traces recorded by Valgrind.
1635 - A new monitor command "v.do expensive_sanity_check_general" to run
1636 some internal consistency checks.
1638 * New flag --sigill-diagnostics to control whether a diagnostic
1639 message is printed when the JIT encounters an instruction it can't
1640 translate. The actual behavior -- delivery of SIGILL to the
1641 application -- is unchanged.
1643 * The maximum amount of memory that Valgrind can use on 64 bit targets
1644 has been increased from 32GB to 64GB. This should make it possible
1645 to run applications on Memcheck that natively require up to about 35GB.
1647 * ==================== FIXED BUGS ====================
1649 The following bugs have been fixed or resolved. Note that "n-i-bz"
1650 stands for "not in bugzilla" -- that is, a bug that was reported to us
1651 but never got a bugzilla entry. We encourage you to file bugs in
1652 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1653 than mailing the developers (or mailing lists) directly -- bugs that
1654 are not entered into bugzilla tend to get forgotten about or ignored.
1656 To see details of a given bug, visit
1657 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1658 where XXXXXX is the bug number as listed below.
1660 123837 system call: 4th argument is optional, depending on cmd
1661 135425 memcheck should tell you where Freed blocks were Mallocd
1662 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
1663 207815 Adds some of the drm ioctls to syswrap-linux.c
1664 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
1665 252955 Impossible to compile with ccache
1666 253519 Memcheck reports auxv pointer accesses as invalid reads.
1667 263034 Crash when loading some PPC64 binaries
1668 269599 Increase deepest backtrace
1669 274695 s390x: Support "compare to/from logical" instructions (z196)
1670 275800 s390x: Autodetect cache info (part 2)
1671 280271 Valgrind reports possible memory leaks on still-reachable std::string
1672 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
1673 289578 Backtraces with ARM unwind tables (stack scan flags)
1674 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
1675 304832 ppc32: build failure
1676 305431 Use find_buildid shdr fallback for separate .debug files
1677 305728 Add support for AVX2 instructions
1678 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
1679 306035 s390x: Fix IR generation for LAAG and friends
1680 306054 s390x: Condition code computation for convert-to-int/logical
1681 306098 s390x: alternate opcode form for convert to/from fixed
1682 306587 Fix cache line detection from auxiliary vector for PPC.
1683 306783 Mips unhandled syscall : 4025 / 4079 / 4182
1684 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
1685 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
1686 307101 sys_capget second argument can be NULL
1687 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
1688 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
1689 307113 s390x: DFP support
1690 307141 valgrind does't work in mips-linux system
1691 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
1692 307285 x86_amd64 feature test for avx in test suite is wrong
1693 307290 memcheck overlap testcase needs memcpy version filter
1694 307463 Please add "&limit=0" to the "all open bugs" link
1695 307465 --show-possibly-lost=no should reduce the error count / exit code
1696 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
1697 307729 pkgconfig support broken valgrind.pc
1698 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
1699 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
1700 308089 Unhandled syscall on ppc64: prctl
1701 308135 PPC32 MPC8xx has 16 bytes cache size
1702 308321 testsuite memcheck filter interferes with gdb_filter
1704 308341 vgdb should report process exit (or fatal signal)
1705 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
1706 308495 Remove build dependency on installed Xen headers
1707 308573 Internal error on 64-bit instruction executed in 32-bit mode
1709 308627 pmovmskb validity bit propagation is imprecise
1710 308644 vgdb command for having the info for the track-fds option
1711 308711 give more info about aspacemgr and arenas in out_of_memory
1712 308717 ARM: implement fixed-point VCVT.F64.[SU]32
1713 308718 ARM implement SMLALBB family of instructions
1714 308886 Missing support for PTRACE_SET/GETREGSET
1715 308930 syscall name_to_handle_at (303 on amd64) not handled
1716 309229 V-bit tester does not report number of tests generated
1717 309323 print unrecognized instuction on MIPS
1718 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
1719 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
1720 309430 Self hosting ppc64 encounters a vassert error on operand type
1721 309600 valgrind is a bit confused about 0-sized sections
1722 309823 Generate errors for still reachable blocks
1723 309921 PCMPISTRI validity bit propagation is imprecise
1724 309922 none/tests/ppc64/test_dfp5 sometimes fails
1725 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
1726 310424 --read-var-info does not properly describe static variables
1727 310792 search additional path for debug symbols
1728 310931 s390x: Message-security assist (MSA) instruction extension [..]
1729 311100 PPC DFP implementation of the integer operands is inconsistent [..]
1730 311318 ARM: "128-bit constant is not implemented" error message
1731 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
1732 311690 V crashes because it redirects branches inside of a redirected function
1733 311880 x86_64: make regtest hangs at shell_valid1
1734 311922 WARNING: unhandled syscall: 170
1736 312171 ppc: insn selection for DFP
1737 312571 Rounding mode call wrong for the DFP Iops [..]
1738 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
1739 312913 Dangling pointers error should also report the alloc stack trace
1740 312980 Building on Mountain Lion generates some compiler warnings
1741 313267 Adding MIPS64/Linux port to Valgrind
1744 313811 Buffer overflow in assert_fail
1745 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
1746 314269 ppc: dead code in insn selection
1747 314718 ARM: implement integer divide instruction (sdiv and udiv)
1748 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
1749 315441 sendmsg syscall should ignore unset msghdr msg_flags
1750 315534 msgrcv inside a thread causes valgrind to hang (block)
1751 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
1752 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
1753 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
1754 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
1755 316144 valgrind.1 manpage contains unknown ??? strings [..]
1756 316145 callgrind command line options in manpage reference (unknown) [..]
1757 316145 callgrind command line options in manpage reference [..]
1758 316181 drd: Fixed a 4x slowdown for certain applications
1759 316503 Valgrind does not support SSE4 "movntdqa" instruction
1760 316535 Use of |signed int| instead of |size_t| in valgrind messages
1761 316696 fluidanimate program of parsec 2.1 stuck
1762 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
1763 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
1764 317186 "Impossible happens" when occurs VCVT instruction on ARM
1765 317318 Support for Threading Building Blocks "scalable_malloc"
1766 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
1767 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
1768 317463 bmi testcase IR SANITY CHECK FAILURE
1769 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
1770 318050 libmpiwrap fails to compile with out-of-source build
1771 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
1772 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
1773 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
1774 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
1775 318932 Add missing PPC64 and PPC32 system call support
1776 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
1777 319395 Crash with unhandled instruction on STRT (Thumb) instructions
1778 319494 VEX Makefile-gcc standalone build update after r2702
1779 319505 [MIPSEL] Crash: unhandled UNRAY operator.
1780 319858 disInstr(thumb): unhandled instruction on instruction STRBT
1781 319932 disInstr(thumb): unhandled instruction on instruction STRHT
1782 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
1783 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
1784 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
1785 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
1786 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
1787 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
1788 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
1789 320895 add fanotify support (patch included)
1790 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
1791 321065 Valgrind updates for Xen 4.3
1792 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
1793 321363 Unhandled instruction: SSAX (ARM + Thumb)
1794 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
1795 321466 Unhandled instruction: SHASX (ARM + Thumb)
1796 321467 Unhandled instruction: SHSAX (ARM + Thumb)
1797 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
1798 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
1799 321620 Unhandled instruction: UASX (ARM + Thumb)
1800 321621 Unhandled instruction: USAX (ARM + Thumb)
1801 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
1802 321693 Unhandled instruction: LDRSBT (Thumb)
1803 321694 Unhandled instruction: UQASX (ARM + Thumb)
1804 321696 Unhandled instruction: UQSAX (Thumb + ARM)
1805 321697 Unhandled instruction: UHASX (ARM + Thumb)
1806 321703 Unhandled instruction: UHSAX (ARM + Thumb)
1807 321704 Unhandled instruction: REVSH (ARM + Thumb)
1808 321730 Add cg_diff and cg_merge man pages
1809 321738 Add vgdb and valgrind-listener man pages
1811 321891 Unhandled instruction: LDRHT (Thumb)
1812 321960 pthread_create() then alloca() causing invalid stack write errors
1813 321969 ppc32 and ppc64 don't support [lf]setxattr
1814 322254 Show threadname together with tid if set by application
1815 322294 Add initial support for IBM Power ISA 2.07
1816 322368 Assertion failure in wqthread_hijack under OS X 10.8
1817 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
1818 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
1819 322851 0bXXX binary literal syntax is not standard
1820 323035 Unhandled instruction: LDRSHT(Thumb)
1821 323036 Unhandled instruction: SMMLS (ARM and Thumb)
1822 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
1823 323175 Unhandled instruction: SMLALD (ARM + Thumb)
1824 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
1825 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
1826 323437 Phase 2 support for IBM Power ISA 2.07
1827 323713 Support mmxext (integer sse) subset on i386 (athlon)
1828 323803 Transactional memory instructions are not supported for Power
1829 323893 SSE3 not available on amd cpus in valgrind
1830 323905 Probable false positive from Valgrind/drd on close()
1831 323912 valgrind.h header isn't compatible for mingw64
1832 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
1833 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
1834 324181 mmap does not handle MAP_32BIT
1835 324227 memcheck false positive leak when a thread calls exit+block [..]
1836 324421 Support for fanotify API on ARM architecture
1837 324514 gdbserver monitor cmd output behaviour consistency [..]
1838 324518 ppc64: Emulation of dcbt instructions does not handle [..]
1839 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
1840 324582 When access is made to freed memory, report both allocation [..]
1841 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
1842 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
1843 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
1844 324834 Unhandled instructions in Microsoft C run-time for x86_64
1845 324894 Phase 3 support for IBM Power ISA 2.07
1846 326091 drd: Avoid false race reports from optimized strlen() impls
1847 326113 valgrind libvex hwcaps error on AMD64
1848 n-i-bz Some wrong command line options could be ignored
1849 n-i-bz patch to allow fair-sched on android
1850 n-i-bz report error for vgdb snapshot requested before execution
1851 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
1853 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
1857 Release 3.8.1 (19 September 2012)
1858 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1859 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
1860 that occur moderately frequently in real use cases, adds support for
1861 some missing instructions on ARM, and fixes a deadlock condition on
1862 MacOSX. If you package or deliver 3.8.0 for others to use, you might
1863 want to consider upgrading to 3.8.1 instead.
1865 The following bugs have been fixed or resolved. Note that "n-i-bz"
1866 stands for "not in bugzilla" -- that is, a bug that was reported to us
1867 but never got a bugzilla entry. We encourage you to file bugs in
1868 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1869 than mailing the developers (or mailing lists) directly -- bugs that
1870 are not entered into bugzilla tend to get forgotten about or ignored.
1872 To see details of a given bug, visit
1873 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1874 where XXXXXX is the bug number as listed below.
1877 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
1878 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
1879 298281 wcslen causes false(?) uninitialised value warnings
1880 301281 valgrind hangs on OS X when the process calls system()
1881 304035 disInstr(arm): unhandled instruction 0xE1023053
1882 304867 implement MOVBE instruction in x86 mode
1883 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
1884 305042 amd64: implement 0F 7F encoding of movq between two registers
1885 305199 ARM: implement QDADD and QDSUB
1886 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
1887 305513 killed by fatal signal: SIGSEGV
1888 305690 DRD reporting invalid semaphore when sem_trywait fails
1889 305926 Invalid alignment checks for some AVX instructions
1890 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
1891 306310 3.8.0 release tarball missing some files
1892 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
1893 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
1894 n-i-bz shmat of a segment > 4Gb does not work
1895 n-i-bz simulate_control_c script wrong USR1 signal number on mips
1896 n-i-bz vgdb ptrace calls wrong on mips [...]
1897 n-i-bz Fixes for more MPI false positives
1898 n-i-bz exp-sgcheck's memcpy causes programs to segfault
1899 n-i-bz OSX build w/ clang: asserts at startup
1900 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
1901 n-i-bz fix a couple of union tag-vs-field mixups
1902 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
1904 The following bugs were fixed in 3.8.0 but not listed in this NEWS
1907 254088 Valgrind should know about UD2 instruction
1910 304754 NEWS blows TeX's little mind
1912 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
1916 Release 3.8.0 (10 August 2012)
1917 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1918 3.8.0 is a feature release with many improvements and the usual
1919 collection of bug fixes.
1921 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
1922 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
1923 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
1924 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
1925 There is initial support for MacOSX 10.8, but it is not usable for
1926 serious work at present.
1928 * ================== PLATFORM CHANGES =================
1930 * Support for MIPS32 platforms running Linux. Valgrind has been
1931 tested on MIPS32 and MIPS32r2 platforms running different Debian
1932 Squeeze and MeeGo distributions. Both little-endian and big-endian
1933 cores are supported. The tools Memcheck, Massif and Lackey have
1934 been tested and are known to work. See README.mips for more details.
1936 * Preliminary support for Android running on x86.
1938 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
1940 * Support for Intel AVX instructions and for AES instructions. This
1941 support is available only for 64 bit code.
1943 * Support for POWER Decimal Floating Point instructions.
1945 * ==================== TOOL CHANGES ====================
1947 * Non-libc malloc implementations are now supported. This is useful
1948 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
1949 Using the new option --soname-synonyms, such tools can be informed
1950 that the malloc implementation is either linked statically into the
1951 executable, or is present in some other shared library different
1952 from libc.so. This makes it possible to process statically linked
1953 programs, and programs using other malloc libraries, for example
1954 TCMalloc or JEMalloc.
1956 * For tools that provide their own replacement for malloc et al, the
1957 option --redzone-size=<number> allows users to specify the size of
1958 the padding blocks (redzones) added before and after each client
1959 allocated block. Smaller redzones decrease the memory needed by
1960 Valgrind. Bigger redzones increase the chance to detect blocks
1961 overrun or underrun. Prior to this change, the redzone size was
1962 hardwired to 16 bytes in Memcheck.
1966 - The leak_check GDB server monitor command now can
1967 control the maximum nr of loss records to output.
1969 - Reduction of memory use for applications allocating
1970 many blocks and/or having many partially defined bytes.
1972 - Addition of GDB server monitor command 'block_list' that lists
1973 the addresses/sizes of the blocks of a leak search loss record.
1975 - Addition of GDB server monitor command 'who_points_at' that lists
1976 the locations pointing at a block.
1978 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
1979 detect an invalid access of these redzones, by marking them
1980 noaccess. Similarly, if a redzone size is given for a memory
1981 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
1982 This still allows to find some bugs if the user has forgotten to
1983 mark the pool superblock noaccess.
1985 - Performance of memory leak check has been improved, especially in
1986 cases where there are many leaked blocks and/or many suppression
1987 rules used to suppress leak reports.
1989 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
1990 more precise analysis, which is important for LLVM/Clang
1991 generated code. This is at the cost of somewhat reduced
1992 performance. Note there is no change to analysis precision or
1993 costs on Linux targets.
1997 - Added even more facilities that can help finding the cause of a data
1998 race, namely the command-line option --ptrace-addr and the macro
1999 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2001 - Fixed a subtle bug that could cause false positive data race reports.
2003 * ==================== OTHER CHANGES ====================
2005 * The C++ demangler has been updated so as to work well with C++
2006 compiled by up to at least g++ 4.6.
2008 * Tool developers can make replacement/wrapping more flexible thanks
2009 to the new option --soname-synonyms. This was reported above, but
2010 in fact is very general and applies to all function
2011 replacement/wrapping, not just to malloc-family functions.
2013 * Round-robin scheduling of threads can be selected, using the new
2014 option --fair-sched= yes. Prior to this change, the pipe-based
2015 thread serialisation mechanism (which is still the default) could
2016 give very unfair scheduling. --fair-sched=yes improves
2017 responsiveness of interactive multithreaded applications, and
2018 improves repeatability of results from the thread checkers Helgrind
2021 * For tool developers: support to run Valgrind on Valgrind has been
2022 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2024 * gdbserver now shows the float shadow registers as integer
2025 rather than float values, as the shadow values are mostly
2026 used as bit patterns.
2028 * Increased limit for the --num-callers command line flag to 500.
2030 * Performance improvements for error matching when there are many
2031 suppression records in use.
2033 * Improved support for DWARF4 debugging information (bug 284184).
2035 * Initial support for DWZ compressed Dwarf debug info.
2037 * Improved control over the IR optimiser's handling of the tradeoff
2038 between performance and precision of exceptions. Specifically,
2039 --vex-iropt-precise-memory-exns has been removed and replaced by
2040 --vex-iropt-register-updates, with extended functionality. This
2041 allows the Valgrind gdbserver to always show up to date register
2044 * Modest performance gains through the use of translation chaining for
2047 * ==================== FIXED BUGS ====================
2049 The following bugs have been fixed or resolved. Note that "n-i-bz"
2050 stands for "not in bugzilla" -- that is, a bug that was reported to us
2051 but never got a bugzilla entry. We encourage you to file bugs in
2052 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2053 than mailing the developers (or mailing lists) directly -- bugs that
2054 are not entered into bugzilla tend to get forgotten about or ignored.
2056 To see details of a given bug, visit
2057 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2058 where XXXXXX is the bug number as listed below.
2060 197914 Building valgrind from svn now requires automake-1.10
2061 203877 increase to 16Mb maximum allowed alignment for memalign et al
2062 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
2063 247386 make perf does not run all performance tests
2064 270006 Valgrind scheduler unfair
2065 270777 Adding MIPS/Linux port to Valgrind
2066 270796 s390x: Removed broken support for the TS insn
2067 271438 Fix configure for proper SSE4.2 detection
2068 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
2069 273475 Add support for AVX instructions
2070 274078 improved configure logic for mpicc
2071 276993 fix mremap 'no thrash checks'
2072 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
2073 281482 memcheck incorrect byte allocation count in realloc() for silly argument
2074 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
2075 283413 Fix wrong sanity check
2076 283671 Robustize alignment computation in LibVEX_Alloc
2077 283961 Adding support for some HCI IOCTLs
2078 284124 parse_type_DIE: confused by: DWARF 4
2079 284864 == 273475 (Add support for AVX instructions)
2080 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
2081 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
2082 285725 == 273475 (Add support for AVX instructions)
2083 286261 add wrapper for linux I2C_RDWR ioctl
2084 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
2085 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
2086 286384 configure fails "checking for a supported version of gcc"
2087 286497 == 273475 (Add support for AVX instructions)
2088 286596 == 273475 (Add support for AVX instructions)
2089 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
2090 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
2091 287260 Incorrect conditional jump or move depends on uninitialised value(s)
2092 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
2093 287307 == 273475 (Add support for AVX instructions)
2094 287858 VG_(strerror): unknown error
2095 288298 (MacOSX) unhandled syscall shm_unlink
2096 288995 == 273475 (Add support for AVX instructions)
2097 289470 Loading of large Mach-O thin binaries fails.
2098 289656 == 273475 (Add support for AVX instructions)
2099 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
2100 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
2101 289839 s390x: Provide support for unicode conversion instructions
2102 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
2103 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
2104 290655 Add support for AESKEYGENASSIST instruction
2105 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
2106 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
2107 291253 ES register not initialised in valgrind simulation
2108 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
2109 291865 s390x: Support the "Compare Double and Swap" family of instructions
2110 292300 == 273475 (Add support for AVX instructions)
2111 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
2112 292493 == 273475 (Add support for AVX instructions)
2113 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
2114 292627 Missing support for some SCSI ioctls
2115 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
2116 292841 == 273475 (Add support for AVX instructions)
2117 292993 implement the getcpu syscall on amd64-linux
2118 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
2119 293088 Add some VEX sanity checks for ppc64 unhandled instructions
2120 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
2121 293754 PCMPxSTRx not implemented for 16-bit characters
2122 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
2123 293808 CLFLUSH not supported by latest VEX for amd64
2124 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
2125 294048 MPSADBW instruction not implemented
2126 294055 regtest none/tests/shell fails when locale is not set to C
2127 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
2128 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
2129 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
2130 294260 disInstr_AMD64: disInstr miscalculated next %rip
2131 294523 --partial-loads-ok=yes causes false negatives
2132 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
2133 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
2134 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
2135 295089 can not annotate source for both helgrind and drd
2136 295221 POWER Processor decimal floating point instruction support missing
2137 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
2138 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
2139 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
2140 295617 ARM - Add some missing syscalls
2141 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
2142 296229 Linux user input device ioctls missing wrappers
2143 296318 ELF Debug info improvements (more than one rx/rw mapping)
2144 296422 Add translation chaining support
2145 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
2146 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
2147 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
2148 297078 gdbserver signal handling problems [..]
2149 297147 drd false positives on newly allocated memory
2150 297329 disallow decoding of IBM Power DFP insns on some machines
2151 297497 POWER Processor decimal floating point instruction support missing
2152 297701 Another alias for strncasecmp_l in libc-2.13.so
2153 297911 'invalid write' not reported when using APIs for custom mem allocators.
2154 297976 s390x: revisit EX implementation
2155 297991 Valgrind interferes with mmap()+ftell()
2156 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
2157 297993 Fix compilation of valgrind with gcc -g3.
2158 298080 POWER Processor DFP support missing, part 3
2159 298227 == 273475 (Add support for AVX instructions)
2160 298335 == 273475 (Add support for AVX instructions)
2161 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
2162 298394 s390x: Don't bail out on an unknown machine model. [..]
2163 298421 accept4() syscall (366) support is missing for ARM
2164 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
2165 298732 valgrind installation problem in ubuntu with kernel version 3.x
2166 298862 POWER Processor DFP instruction support missing, part 4
2167 298864 DWARF reader mis-parses DW_FORM_ref_addr
2168 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
2169 299053 Support DWARF4 DW_AT_high_pc constant form
2170 299104 == 273475 (Add support for AVX instructions)
2171 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
2172 299629 dup3() syscall (358) support is missing for ARM
2173 299694 POWER Processor DFP instruction support missing, part 5
2174 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
2175 299803 == 273475 (Add support for AVX instructions)
2176 299804 == 273475 (Add support for AVX instructions)
2177 299805 == 273475 (Add support for AVX instructions)
2178 300140 ARM - Missing (T1) SMMUL
2179 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
2180 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
2181 300414 FCOM and FCOMP unimplemented for amd64 guest
2182 301204 infinite loop in canonicaliseSymtab with ifunc symbol
2183 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
2184 301265 add x86 support to Android build
2185 301984 configure script doesn't detect certain versions of clang
2186 302205 Fix compiler warnings for POWER VEX code and POWER test cases
2187 302287 Unhandled movbe instruction on Atom processors
2188 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
2189 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
2190 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
2191 302656 == 273475 (Add support for AVX instructions)
2192 302709 valgrind for ARM needs extra tls support for android emulator [..]
2193 302827 add wrapper for CDROM_GET_CAPABILITY
2194 302901 Valgrind crashes with dwz optimized debuginfo
2195 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
2196 303116 Add support for the POWER instruction popcntb
2197 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
2198 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
2199 303466 == 273475 (Add support for AVX instructions)
2200 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
2201 303963 strstr() function produces wrong results under valgrind callgrind
2202 304054 CALL_FN_xx macros need to enforce stack alignment
2203 304561 tee system call not supported
2204 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
2205 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
2206 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
2207 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
2208 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
2209 n-i-bz s390x: Shadow registers can now be examined using vgdb
2211 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
2212 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
2216 Release 3.7.0 (5 November 2011)
2217 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2218 3.7.0 is a feature release with many significant improvements and the
2219 usual collection of bug fixes.
2221 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2222 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
2223 Support for recent distros and toolchain components (glibc 2.14, gcc
2224 4.6, MacOSX 10.7) has been added.
2226 * ================== PLATFORM CHANGES =================
2228 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
2229 analyse 64-bit programs running on z/Architecture. Most user space
2230 instructions up to and including z10 are supported. Valgrind has
2231 been tested extensively on z9, z10, and z196 machines running SLES
2232 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
2233 known to work well. Callgrind, Helgrind, and DRD work reasonably
2234 well on z9 and later models. See README.s390 for more details.
2236 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
2237 64-bit processes are supported. Some complex threaded applications
2238 (Firefox) are observed to hang when run as 32 bit applications,
2239 whereas 64-bit versions run OK. The cause is unknown. Memcheck
2240 will likely report some false errors. In general, expect some rough
2241 spots. This release also supports MacOSX 10.6, but drops support
2244 * Preliminary support for Android (on ARM). Valgrind can now run
2245 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
2246 README.android for more details, plus instructions on how to get
2249 * Support for the IBM Power ISA 2.06 (Power7 instructions)
2251 * General correctness and performance improvements for ARM/Linux, and,
2252 by extension, ARM/Android.
2254 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
2255 instruction set support is under development but is not available in
2258 * Support for AIX5 has been removed.
2260 * ==================== TOOL CHANGES ====================
2262 * Memcheck: some incremental changes:
2264 - reduction of memory use in some circumstances
2266 - improved handling of freed memory, which in some circumstances
2267 can cause detection of use-after-free that would previously have
2270 - fix of a longstanding bug that could cause false negatives (missed
2271 errors) in programs doing vector saturated narrowing instructions.
2273 * Helgrind: performance improvements and major memory use reductions,
2274 particularly for large, long running applications which perform many
2275 synchronisation (lock, unlock, etc) events. Plus many smaller
2278 - display of locksets for both threads involved in a race
2280 - general improvements in formatting/clarity of error messages
2282 - addition of facilities and documentation regarding annotation
2283 of thread safe reference counted C++ classes
2285 - new flag --check-stack-refs=no|yes [yes], to disable race checking
2286 on thread stacks (a performance hack)
2288 - new flag --free-is-write=no|yes [no], to enable detection of races
2289 where one thread accesses heap memory but another one frees it,
2290 without any coordinating synchronisation event
2292 * DRD: enabled XML output; added support for delayed thread deletion
2293 in order to detect races that occur close to the end of a thread
2294 (--join-list-vol); fixed a memory leak triggered by repeated client
2295 memory allocatation and deallocation; improved Darwin support.
2297 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
2299 * exp-sgcheck: this tool has been reduced in scope so as to improve
2300 performance and remove checking that Memcheck does better.
2301 Specifically, the ability to check for overruns for stack and global
2302 arrays is unchanged, but the ability to check for overruns of heap
2303 blocks has been removed. The tool has accordingly been renamed to
2304 exp-sgcheck ("Stack and Global Array Checking").
2306 * ==================== OTHER CHANGES ====================
2308 * GDB server: Valgrind now has an embedded GDB server. That means it
2309 is possible to control a Valgrind run from GDB, doing all the usual
2310 things that GDB can do (single stepping, breakpoints, examining
2311 data, etc). Tool-specific functionality is also available. For
2312 example, it is possible to query the definedness state of variables
2313 or memory from within GDB when running Memcheck; arbitrarily large
2314 memory watchpoints are supported, etc. To use the GDB server, start
2315 Valgrind with the flag --vgdb-error=0 and follow the on-screen
2318 * Improved support for unfriendly self-modifying code: a new option
2319 --smc-check=all-non-file is available. This adds the relevant
2320 consistency checks only to code that originates in non-file-backed
2321 mappings. In effect this confines the consistency checking only to
2322 code that is or might be JIT generated, and avoids checks on code
2323 that must have been compiled ahead of time. This significantly
2324 improves performance on applications that generate code at run time.
2326 * It is now possible to build a working Valgrind using Clang-2.9 on
2329 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
2330 These enable and disable error reporting on a per-thread, and
2331 nestable, basis. This is useful for hiding errors in particularly
2332 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
2333 now uses this facility.
2335 * Added the --mod-funcname option to cg_diff.
2337 * ==================== FIXED BUGS ====================
2339 The following bugs have been fixed or resolved. Note that "n-i-bz"
2340 stands for "not in bugzilla" -- that is, a bug that was reported to us
2341 but never got a bugzilla entry. We encourage you to file bugs in
2342 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2343 mailing the developers (or mailing lists) directly -- bugs that are
2344 not entered into bugzilla tend to get forgotten about or ignored.
2346 To see details of a given bug, visit
2347 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2348 where XXXXXX is the bug number as listed below.
2350 79311 malloc silly arg warning does not give stack trace
2351 210935 port valgrind.h (not valgrind) to win32 to support client requests
2352 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
2353 243404 Port to zSeries
2354 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
2355 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
2356 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
2357 253206 Some fixes for the faultstatus testcase
2358 255223 capget testcase fails when running as root
2359 256703 xlc_dbl_u32.c testcase broken
2360 256726 Helgrind tests have broken inline asm
2361 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
2362 264800 testcase compile failure on zseries
2363 265762 make public VEX headers compilable by G++ 3.x
2364 265771 assertion in jumps.c (r11523) fails with glibc-2.3
2365 266753 configure script does not give the user the option to not use QtCore
2366 266931 gen_insn_test.pl is broken
2367 266961 ld-linux.so.2 i?86-linux strlen issues
2368 266990 setns instruction causes false positive
2369 267020 Make directory for temporary files configurable at run-time.
2370 267342 == 267997 (segmentation fault on Mac OS 10.6)
2371 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
2372 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
2373 267488 regtest: darwin support for 64-bit build
2374 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
2375 267630 Add support for IBM Power ISA 2.06 -- stage 1
2376 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
2377 267819 Add client request for informing the core about reallocation
2378 267925 laog data structure quadratic for a single sequence of lock
2379 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
2380 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
2381 268513 missed optimizations in fold_Expr
2382 268619 s390x: fpr - gpr transfer facility
2383 268620 s390x: reconsider "long displacement" requirement
2384 268621 s390x: improve IR generation for XC
2385 268715 s390x: FLOGR is not universally available
2386 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
2387 268930 s390x: MHY is not universally available
2388 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
2389 269079 Support ptrace system call on ARM
2390 269144 missing "Bad option" error message
2391 269209 conditional load and store facility (z196)
2392 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
2393 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
2394 269736 s390x: minor code generation tweaks
2395 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
2396 269863 s390x: remove unused function parameters
2397 269864 s390x: tweak s390_emit_load_cc
2398 269884 == 250101 (overhead for huge blocks exhausts space too soon)
2399 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
2400 270115 s390x: rewrite some testcases
2401 270309 == 267997 (valgrind crash on startup)
2402 270320 add support for Linux FIOQSIZE ioctl() call
2403 270326 segfault while trying to sanitize the environment passed to execle
2404 270794 IBM POWER7 support patch causes regression in none/tests
2405 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
2406 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
2407 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
2408 270959 s390x: invalid use of R0 as base register
2409 271042 VSX configure check fails when it should not
2410 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
2411 271259 s390x: fix code confusion
2412 271337 == 267997 (Valgrind segfaults on MacOS X)
2413 271385 s390x: Implement Ist_MBE
2414 271501 s390x: misc cleanups
2415 271504 s390x: promote likely and unlikely
2416 271579 ppc: using wrong enum type
2417 271615 unhandled instruction "popcnt" (arch=amd10h)
2418 271730 Fix bug when checking ioctls: duplicate check
2419 271776 s390x: provide STFLE instruction support
2420 271779 s390x: provide clock instructions like STCK
2421 271799 Darwin: ioctls without an arg report a memory error
2422 271820 arm: fix type confusion
2423 271917 pthread_cond_timedwait failure leads to not-locked false positive
2424 272067 s390x: fix DISP20 macro
2425 272615 A typo in debug output in mc_leakcheck.c
2426 272661 callgrind_annotate chokes when run from paths containing regex chars
2427 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
2428 272955 Unhandled syscall error for pwrite64 on ppc64 arch
2429 272967 make documentation build-system more robust
2430 272986 Fix gcc-4.6 warnings with valgrind.h
2431 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
2432 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
2433 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
2434 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
2435 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
2436 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
2437 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
2438 273778 exp-ptrcheck: unhandled sysno == 259
2439 274089 exp-ptrcheck: unhandled sysno == 208
2440 274378 s390x: Various dispatcher tweaks
2441 274447 WARNING: unhandled syscall: 340
2442 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
2443 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
2444 274926 valgrind does not build against linux-3
2445 275148 configure FAIL with glibc-2.14
2446 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
2447 275168 Make Valgrind work for MacOSX 10.7 Lion
2448 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
2449 275278 valgrind does not build on Linux kernel 3.0.* due to silly
2450 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
2451 275308 Fix implementation for ppc64 fres instruc
2452 275339 s390x: fix testcase compile warnings
2453 275517 s390x: Provide support for CKSM instruction
2454 275710 s390x: get rid of redundant address mode calculation
2455 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
2456 275852 == 250101 (valgrind uses all swap space and is killed)
2457 276784 Add support for IBM Power ISA 2.06 -- stage 3
2458 276987 gdbsrv: fix tests following recent commits
2459 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
2460 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
2461 277471 Unhandled syscall: 340
2462 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
2463 277653 ARM: support Thumb2 PLD instruction
2464 277663 ARM: NEON float VMUL by scalar incorrect
2465 277689 ARM: tests for VSTn with register post-index are broken
2466 277694 ARM: BLX LR instruction broken in ARM mode
2467 277780 ARM: VMOV.F32 (immediate) instruction is broken
2468 278057 fuse filesystem syscall deadlocks
2469 278078 Unimplemented syscall 280 on ppc32
2470 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
2471 278454 VALGRIND_STACK_DEREGISTER has wrong output type
2472 278502 == 275284 (Valgrind confuses memcpy() and memmove())
2473 278892 gdbsrv: factorize gdb version handling, fix doc and typos
2474 279027 Support for MVCL and CLCL instruction
2475 279027 s390x: Provide support for CLCL and MVCL instructions
2476 279062 Remove a redundant check in the insn selector for ppc.
2477 279071 JDK creates PTEST with redundant REX.W prefix
2478 279212 gdbsrv: add monitor cmd v.info scheduler.
2479 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
2480 279698 memcheck discards valid-bits for packuswb
2481 279795 memcheck reports uninitialised values for mincore on amd64
2482 279994 Add support for IBM Power ISA 2.06 -- stage 3
2483 280083 mempolicy syscall check errors
2484 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
2485 280710 s390x: config files for nightly builds
2486 280757 /tmp dir still used by valgrind even if TMPDIR is specified
2487 280965 Valgrind breaks fcntl locks when program does mmap
2488 281138 WARNING: unhandled syscall: 340
2489 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
2490 281304 == 275168 (Darwin: dyld "cannot load inserted library")
2491 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
2492 281468 s390x: handle do_clone and gcc clones in call traces
2493 281488 ARM: VFP register corruption
2494 281828 == 275284 (false memmove warning: "Source and destination overlap")
2495 281883 s390x: Fix system call wrapper for "clone".
2496 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
2497 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
2498 282238 SLES10: make check fails
2499 282979 strcasestr needs replacement with recent(>=2.12) glibc
2500 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
2501 283243 Regression in ppc64 memcheck tests
2502 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
2503 283427 re-connect epoll_pwait syscall on ARM linux
2504 283600 gdbsrv: android: port vgdb.c
2505 283709 none/tests/faultstatus needs to account for page size
2506 284305 filter_gdb needs enhancement to work on ppc64
2507 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
2508 284472 Thumb2 ROR.W encoding T2 not implemented
2509 284621 XML-escape process command line in XML output
2510 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
2511 that have non-power-of-2 sizes (also AMDs)
2512 n-i-bz don't be spooked by libraries mashed by elfhack
2513 n-i-bz don't be spooked by libxul.so linked with gold
2514 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
2516 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
2517 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
2518 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
2522 Release 3.6.1 (16 February 2011)
2523 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2524 3.6.1 is a bug fix release. It adds support for some SSE4
2525 instructions that were omitted in 3.6.0 due to lack of time. Initial
2526 support for glibc-2.13 has been added. A number of bugs causing
2527 crashing or assertion failures have been fixed.
2529 The following bugs have been fixed or resolved. Note that "n-i-bz"
2530 stands for "not in bugzilla" -- that is, a bug that was reported to us
2531 but never got a bugzilla entry. We encourage you to file bugs in
2532 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2533 mailing the developers (or mailing lists) directly -- bugs that are
2534 not entered into bugzilla tend to get forgotten about or ignored.
2536 To see details of a given bug, visit
2537 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2538 where XXXXXX is the bug number as listed below.
2540 188572 Valgrind on Mac should suppress setenv() mem leak
2541 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
2542 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
2543 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
2544 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
2545 254420 memory pool tracking broken
2546 254957 Test code failing to compile due to changes in memcheck.h
2547 255009 helgrind/drd: crash on chmod with invalid parameter
2548 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
2549 255355 helgrind/drd: crash on threaded programs doing fork
2551 255418 (SSE4.x) rint call compiled with ICC
2552 255822 --gen-suppressions can create invalid files: "too many callers [...]"
2553 255888 closing valgrindoutput tag outputted to log-stream on error
2554 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
2555 255966 Slowness when using mempool annotations
2556 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
2557 256600 super-optimized strcasecmp() false positive
2558 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
2559 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
2560 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
2561 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
2562 257276 Missing case in memcheck --track-origins=yes
2563 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
2564 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
2565 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
2566 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
2567 263099 callgrind_annotate counts Ir improperly [...]
2568 263877 undefined coprocessor instruction on ARMv7
2569 265964 configure FAIL with glibc-2.13
2570 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
2571 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
2572 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
2573 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
2574 n-i-bz DRD: disable --free-is-write due to implementation difficulties
2576 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
2580 Release 3.6.0 (21 October 2010)
2581 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2582 3.6.0 is a feature release with many significant improvements and the
2583 usual collection of bug fixes.
2585 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2586 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
2587 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
2589 -------------------------
2591 Here are some highlights. Details are shown further down:
2593 * Support for ARM/Linux.
2595 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
2597 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
2599 * Support for the SSE4.2 instruction set.
2601 * Enhancements to the Callgrind profiler, including the ability to
2602 handle CPUs with three levels of cache.
2604 * A new experimental heap profiler, DHAT.
2606 * A huge number of bug fixes and small enhancements.
2608 -------------------------
2610 Here are details of the above changes, together with descriptions of
2611 many other changes, and a list of fixed bugs.
2613 * ================== PLATFORM CHANGES =================
2615 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
2616 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
2617 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
2619 This requires a CPU capable of running the ARMv7-A instruction set
2620 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
2621 of the user space instruction set, including ARM and Thumb integer
2622 code, VFPv3, NEON and V6 media instructions. The Memcheck,
2623 Cachegrind and Massif tools work properly; other tools work to
2626 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
2627 with support for recent releases of the underlying toolchain
2628 components, notably gcc-4.5 and glibc-2.12.
2630 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
2631 support also works much better on OS X 10.5, and is as solid as
2634 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
2635 64-bit mode. In 32-bit mode, support is only available up to and
2636 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
2637 supported in 64-bit mode, and 32-bit mode does in fact support the
2638 bare minimum SSE4 instructions to needed to run programs on Mac OS X
2639 10.6 on 32-bit targets.
2641 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
2642 and including version 2.05 is supported.
2644 * ==================== TOOL CHANGES ====================
2646 * Cachegrind has a new processing script, cg_diff, which finds the
2647 difference between two profiles. It's very useful for evaluating
2648 the performance effects of a change in a program.
2650 Related to this change, the meaning of cg_annotate's (rarely-used)
2651 --threshold option has changed; this is unlikely to affect many
2652 people, if you do use it please see the user manual for details.
2654 * Callgrind now can do branch prediction simulation, similar to
2655 Cachegrind. In addition, it optionally can count the number of
2656 executed global bus events. Both can be used for a better
2657 approximation of a "Cycle Estimation" as derived event (you need to
2658 update the event formula in KCachegrind yourself).
2660 * Cachegrind and Callgrind now refer to the LL (last-level) cache
2661 rather than the L2 cache. This is to accommodate machines with
2662 three levels of caches -- if Cachegrind/Callgrind auto-detects the
2663 cache configuration of such a machine it will run the simulation as
2664 if the L2 cache isn't present. This means the results are less
2665 likely to match the true result for the machine, but
2666 Cachegrind/Callgrind's results are already only approximate, and
2667 should not be considered authoritative. The results are still
2668 useful for giving a general idea about a program's locality.
2670 * Massif has a new option, --pages-as-heap, which is disabled by
2671 default. When enabled, instead of tracking allocations at the level
2672 of heap blocks (as allocated with malloc/new/new[]), it instead
2673 tracks memory allocations at the level of memory pages (as mapped by
2674 mmap, brk, etc). Each mapped page is treated as its own block.
2675 Interpreting the page-level output is harder than the heap-level
2676 output, but this option is useful if you want to account for every
2677 byte of memory used by a program.
2679 * DRD has two new command-line options: --free-is-write and
2680 --trace-alloc. The former allows to detect reading from already freed
2681 memory, and the latter allows tracing of all memory allocations and
2684 * DRD has several new annotations. Custom barrier implementations can
2685 now be annotated, as well as benign races on static variables.
2687 * DRD's happens before / happens after annotations have been made more
2688 powerful, so that they can now also be used to annotate e.g. a smart
2689 pointer implementation.
2691 * Helgrind's annotation set has also been drastically improved, so as
2692 to provide to users a general set of annotations to describe locks,
2693 semaphores, barriers and condition variables. Annotations to
2694 describe thread-safe reference counted heap objects have also been
2697 * Memcheck has a new command-line option, --show-possibly-lost, which
2698 is enabled by default. When disabled, the leak detector will not
2699 show possibly-lost blocks.
2701 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
2702 has been added. DHAT keeps track of allocated heap blocks, and also
2703 inspects every memory reference to see which block (if any) is being
2704 accessed. This gives a lot of insight into block lifetimes,
2705 utilisation, turnover, liveness, and the location of hot and cold
2706 fields. You can use DHAT to do hot-field profiling.
2708 * ==================== OTHER CHANGES ====================
2710 * Improved support for unfriendly self-modifying code: the extra
2711 overhead incurred by --smc-check=all has been reduced by
2712 approximately a factor of 5 as compared with 3.5.0.
2714 * Ability to show directory names for source files in error messages.
2715 This is combined with a flexible mechanism for specifying which
2716 parts of the paths should be shown. This is enabled by the new flag
2719 * A new flag, --require-text-symbol, which will stop the run if a
2720 specified symbol is not found it a given shared object when it is
2721 loaded into the process. This makes advanced working with function
2722 intercepting and wrapping safer and more reliable.
2724 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
2725 and control of Valgrind is now available for the tools Memcheck and
2726 Helgrind. XML output from Valgrind is available for Memcheck,
2727 Helgrind and exp-Ptrcheck.
2729 * More reliable stack unwinding on amd64-linux, particularly in the
2730 presence of function wrappers, and with gcc-4.5 compiled code.
2732 * Modest scalability (performance improvements) for massive
2733 long-running applications, particularly for those with huge amounts
2736 * Support for analyzing programs running under Wine with has been
2737 improved. The header files <valgrind/valgrind.h>,
2738 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
2739 Windows-programs compiled with MinGW or one of the Microsoft Visual
2742 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
2743 The 32-bit simulator was not affected. This did not occur often,
2744 but when it did would usually crash the program under test.
2747 * A large number of bugs were fixed. These are shown below.
2749 * A number of bugs were investigated, and were candidates for fixing,
2750 but are not fixed in 3.6.0, due to lack of developer time. They may
2751 get fixed in later releases. They are:
2753 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
2754 212419 false positive "lock order violated" (A+B vs A)
2755 213685 Undefined value propagates past dependency breaking instruction
2756 216837 Incorrect instrumentation of NSOperationQueue on Darwin
2757 237920 valgrind segfault on fork failure
2758 242137 support for code compiled by LLVM-2.8
2759 242423 Another unknown Intel cache config value
2760 243232 Inconsistent Lock Orderings report with trylock
2761 243483 ppc: callgrind triggers VEX assertion failure
2762 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
2763 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
2765 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
2766 249435 Analyzing wine programs with callgrind triggers a crash
2767 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
2768 250065 Handling large allocations
2769 250101 huge "free" memory usage due to m_mallocfree.c
2770 "superblocks fragmentation"
2771 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
2772 252091 Callgrind on ARM does not detect function returns correctly
2773 252600 [PATCH] Allow lhs to be a pointer for shl/shr
2774 254420 memory pool tracking broken
2775 n-i-bz support for adding symbols for JIT generated code
2778 The following bugs have been fixed or resolved. Note that "n-i-bz"
2779 stands for "not in bugzilla" -- that is, a bug that was reported to us
2780 but never got a bugzilla entry. We encourage you to file bugs in
2781 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2782 mailing the developers (or mailing lists) directly -- bugs that are
2783 not entered into bugzilla tend to get forgotten about or ignored.
2785 To see details of a given bug, visit
2786 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2787 where XXXXXX is the bug number as listed below.
2789 135264 dcbzl instruction missing
2791 153699 Valgrind should report unaligned reads with movdqa
2793 190429 Valgrind reports lost of errors in ld.so
2794 with x86_64 2.9.90 glibc
2795 197266 valgrind appears to choke on the xmms instruction
2797 197988 Crash when demangling very large symbol names
2798 202315 unhandled syscall: 332 (inotify_init1)
2799 203256 Add page-level profiling to Massif
2800 205093 dsymutil=yes needs quotes, locking (partial fix)
2801 205241 Snow Leopard 10.6 support (partial fix)
2802 206600 Leak checker fails to upgrade indirect blocks when their
2803 parent becomes reachable
2804 210935 port valgrind.h (not valgrind) to win32 so apps run under
2805 wine can make client requests
2806 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
2807 within Linux ip-stack checksum functions
2808 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
2810 213685 Undefined value propagates past dependency breaking instruction
2812 215914 Valgrind inserts bogus empty environment variable
2814 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
2815 222545 shmat fails under valgind on some arm targets
2816 222560 ARM NEON support
2819 232509 Docs build fails with formatting inside <title></title> elements
2821 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
2822 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
2823 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
2824 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
2825 237485 symlink (syscall 57) is not supported on Mac OS
2826 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
2828 238208 is_just_below_ESP doesn't take into account red-zone
2829 238345 valgrind passes wrong $0 when executing a shell script
2830 238679 mq_timedreceive syscall doesn't flag the reception buffer
2832 238696 fcntl command F_DUPFD_CLOEXEC not supported
2833 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
2834 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
2835 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
2837 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
2843 242606 unhandled syscall: setegid (in Ptrcheck)
2844 242814 Helgrind "Impossible has happened" during
2845 QApplication::initInstance();
2846 243064 Valgrind attempting to read debug information from iso
2847 243270 Make stack unwinding in Valgrind wrappers more reliable
2848 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
2849 sysno = 277 (mq_open)
2850 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
2851 244493 ARM VFP d16-d31 registers support
2852 244670 add support for audit_session_self syscall on Mac OS 10.6
2853 244921 The xml report of helgrind tool is not well format
2854 244923 In the xml report file, the <preamble> not escape the
2855 xml char, eg '<','&','>'
2856 245535 print full path names in plain text reports
2857 245925 x86-64 red zone handling problem
2858 246258 Valgrind not catching integer underruns + new [] s
2859 246311 reg/reg cmpxchg doesn't work on amd64
2860 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
2861 246888 Improve Makefile.vex.am
2862 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
2863 to [f]chmod_extended
2864 247526 IBM POWER6 (ISA 2.05) support is incomplete
2865 247561 Some leak testcases fails due to reachable addresses in
2867 247875 sizeofIRType to handle Ity_I128
2868 247894 [PATCH] unhandled syscall sys_readahead
2869 247980 Doesn't honor CFLAGS passed to configure
2870 248373 darwin10.supp is empty in the trunk
2871 248822 Linux FIBMAP ioctl has int parameter instead of long
2872 248893 [PATCH] make readdwarf.c big endianess safe to enable
2873 unwinding on big endian systems
2874 249224 Syscall 336 not supported (SYS_proc_info)
2876 249775 Incorrect scheme for detecting NEON capabilities of host CPU
2877 249943 jni JVM init fails when using valgrind
2878 249991 Valgrind incorrectly declares AESKEYGENASSIST support
2880 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
2881 250799 frexp$fenv_access_off function generates SIGILL
2882 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
2883 251251 support pclmulqdq insn
2884 251362 valgrind: ARM: attach to debugger either fails or provokes
2886 251674 Unhandled syscall 294
2889 254257 Add support for debugfiles found by build-id
2890 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
2891 254646 Wrapped functions cause stack misalignment on OS X
2892 (and possibly Linux)
2893 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
2895 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
2899 Release 3.5.0 (19 August 2009)
2900 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2901 3.5.0 is a feature release with many significant improvements and the
2902 usual collection of bug fixes. The main improvement is that Valgrind
2903 now works on Mac OS X.
2905 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
2906 and X86/Darwin. Support for recent distros and toolchain components
2907 (glibc 2.10, gcc 4.5) has been added.
2909 -------------------------
2911 Here is a short summary of the changes. Details are shown further
2914 * Support for Mac OS X (10.5.x).
2916 * Improvements and simplifications to Memcheck's leak checker.
2918 * Clarification and simplifications in various aspects of Valgrind's
2921 * XML output for Helgrind and Ptrcheck.
2923 * Performance and stability improvements for Helgrind and DRD.
2925 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
2927 * A new experimental tool, BBV, useful for computer architecture
2930 * Improved Wine support, including ability to read Windows PDB
2933 -------------------------
2935 Here are details of the above changes, followed by descriptions of
2936 many other minor changes, and a list of fixed bugs.
2939 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
2940 called "Darwin" because that is the name of the OS core, which is the
2941 level that Valgrind works at.)
2945 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
2946 because it would require work and 10.4 is only becoming less common.
2948 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
2949 fairly well. For 10.5.x, 32-bit programs are the default even on
2950 64-bit machines, so it handles most current programs.
2952 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
2953 officially supported, but simple programs at least will probably work.
2954 However, start-up is slow.
2956 - PowerPC machines are not supported.
2958 Things that don't work:
2960 - The Ptrcheck tool.
2962 - Objective-C garbage collection.
2966 - If you have Rogue Amoeba's "Instant Hijack" program installed,
2967 Valgrind will fail with a SIGTRAP at start-up. See
2968 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
2973 - You will likely find --dsymutil=yes a useful option, as error
2974 messages may be imprecise without it.
2976 - Mac OS X support is new and therefore will be less robust than the
2977 Linux support. Please report any bugs you find.
2979 - Threaded programs may run more slowly than on Linux.
2981 Many thanks to Greg Parker for developing this port over several years.
2984 * Memcheck's leak checker has been improved.
2986 - The results for --leak-check=summary now match the summary results
2987 for --leak-check=full. Previously they could differ because
2988 --leak-check=summary counted "indirectly lost" blocks and
2989 "suppressed" blocks as "definitely lost".
2991 - Blocks that are only reachable via at least one interior-pointer,
2992 but are directly pointed to by a start-pointer, were previously
2993 marked as "still reachable". They are now correctly marked as
2996 - The default value for the --leak-resolution option has been
2997 changed from "low" to "high". In general, this means that more
2998 leak reports will be produced, but each leak report will describe
2999 fewer leaked blocks.
3001 - With --leak-check=full, "definitely lost" and "possibly lost"
3002 leaks are now considered as proper errors, ie. they are counted
3003 for the "ERROR SUMMARY" and affect the behaviour of
3004 --error-exitcode. These leaks are not counted as errors if
3005 --leak-check=summary is specified, however.
3007 - Documentation for the leak checker has been improved.
3010 * Various aspects of Valgrind's text output have changed.
3012 - Valgrind's start-up message has changed. It is shorter but also
3013 includes the command being run, which makes it easier to use
3014 --trace-children=yes. An example:
3016 - Valgrind's shut-down messages have also changed. This is most
3017 noticeable with Memcheck, where the leak summary now occurs before
3018 the error summary. This change was necessary to allow leaks to be
3019 counted as proper errors (see the description of the leak checker
3020 changes above for more details). This was also necessary to fix a
3021 longstanding bug in which uses of suppressions against leaks were
3022 not "counted", leading to difficulties in maintaining suppression
3023 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3025 - Behavior of -v has changed. In previous versions, -v printed out
3026 a mixture of marginally-user-useful information, and tool/core
3027 statistics. The statistics printing has now been moved to its own
3028 flag, --stats=yes. This means -v is less verbose and more likely
3029 to convey useful end-user information.
3031 - The format of some (non-XML) stack trace entries has changed a
3032 little. Previously there were six possible forms:
3034 0x80483BF: really (a.c:20)
3035 0x80483BF: really (in /foo/a.out)
3037 0x80483BF: (within /foo/a.out)
3038 0x80483BF: ??? (a.c:20)
3041 The third and fourth of these forms have been made more consistent
3042 with the others. The six possible forms are now:
3044 0x80483BF: really (a.c:20)
3045 0x80483BF: really (in /foo/a.out)
3046 0x80483BF: really (in ???)
3047 0x80483BF: ??? (in /foo/a.out)
3048 0x80483BF: ??? (a.c:20)
3051 Stack traces produced when --xml=yes is specified are different
3055 * Helgrind and Ptrcheck now support XML output, so they can be used
3056 from GUI tools. Also, the XML output mechanism has been
3059 - The XML format has been overhauled and generalised, so it is more
3060 suitable for error reporting tools in general. The Memcheck
3061 specific aspects of it have been removed. The new format, which
3062 is an evolution of the old format, is described in
3063 docs/internals/xml-output-protocol4.txt.
3065 - Memcheck has been updated to use the new format.
3067 - Helgrind and Ptrcheck are now able to emit output in this format.
3069 - The XML output mechanism has been overhauled. XML is now output
3070 to its own file descriptor, which means that:
3072 * Valgrind can output text and XML independently.
3074 * The longstanding problem of XML output being corrupted by
3075 unexpected un-tagged text messages is solved.
3077 As before, the destination for text output is specified using
3078 --log-file=, --log-fd= or --log-socket=.
3080 As before, XML output for a tool is enabled using --xml=yes.
3082 Because there's a new XML output channel, the XML output
3083 destination is now specified by --xml-file=, --xml-fd= or
3086 Initial feedback has shown this causes some confusion. To
3087 clarify, the two envisaged usage scenarios are:
3089 (1) Normal text output. In this case, do not specify --xml=yes
3090 nor any of --xml-file=, --xml-fd= or --xml-socket=.
3092 (2) XML output. In this case, specify --xml=yes, and one of
3093 --xml-file=, --xml-fd= or --xml-socket= to select the XML
3094 destination, one of --log-file=, --log-fd= or --log-socket=
3095 to select the destination for any remaining text messages,
3096 and, importantly, -q.
3098 -q makes Valgrind completely silent on the text channel,
3099 except in the case of critical failures, such as Valgrind
3100 itself segfaulting, or failing to read debugging information.
3101 Hence, in this scenario, it suffices to check whether or not
3102 any output appeared on the text channel. If yes, then it is
3103 likely to be a critical error which should be brought to the
3104 attention of the user. If no (the text channel produced no
3105 output) then it can be assumed that the run was successful.
3107 This allows GUIs to make the critical distinction they need to
3108 make (did the run fail or not?) without having to search or
3109 filter the text output channel in any way.
3111 It is also recommended to use --child-silent-after-fork=yes in
3115 * Improvements and changes in Helgrind:
3117 - XML output, as described above
3119 - Checks for consistent association between pthread condition
3120 variables and their associated mutexes are now performed.
3122 - pthread_spinlock functions are supported.
3124 - Modest performance improvements.
3126 - Initial (skeletal) support for describing the behaviour of
3127 non-POSIX synchronisation objects through ThreadSanitizer
3128 compatible ANNOTATE_* macros.
3130 - More controllable tradeoffs between performance and the level of
3131 detail of "previous" accesses in a race. There are now three
3134 * --history-level=full. This is the default, and was also the
3135 default in 3.4.x. It shows both stacks involved in a race, but
3136 requires a lot of memory and can be very slow in programs that
3137 do many inter-thread synchronisation events.
3139 * --history-level=none. This only shows the later stack involved
3140 in a race. This can be much faster than --history-level=full,
3141 but makes it much more difficult to find the other access
3142 involved in the race.
3144 The new intermediate setting is
3146 * --history-level=approx
3148 For the earlier (other) access, two stacks are presented. The
3149 earlier access is guaranteed to be somewhere in between the two
3150 program points denoted by those stacks. This is not as useful
3151 as showing the exact stack for the previous access (as per
3152 --history-level=full), but it is better than nothing, and it's
3153 almost as fast as --history-level=none.
3156 * New features and improvements in DRD:
3158 - The error messages printed by DRD are now easier to interpret.
3159 Instead of using two different numbers to identify each thread
3160 (Valgrind thread ID and DRD thread ID), DRD does now identify
3161 threads via a single number (the DRD thread ID). Furthermore
3162 "first observed at" information is now printed for all error
3163 messages related to synchronization objects.
3165 - Added support for named semaphores (sem_open() and sem_close()).
3167 - Race conditions between pthread_barrier_wait() and
3168 pthread_barrier_destroy() calls are now reported.
3170 - Added support for custom allocators through the macros
3171 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
3172 in <valgrind/valgrind.h>). An alternative for these two macros is
3173 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
3176 - Added support for annotating non-POSIX synchronization objects
3177 through several new ANNOTATE_*() macros.
3179 - OpenMP: added support for the OpenMP runtime (libgomp) included
3180 with gcc versions 4.3.0 and 4.4.0.
3184 - Added two new command-line options (--first-race-only and
3185 --segment-merging-interval).
3188 * Genuinely atomic support for x86/amd64/ppc atomic instructions
3190 Valgrind will now preserve (memory-access) atomicity of LOCK-
3191 prefixed x86/amd64 instructions, and any others implying a global
3192 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
3194 This means that Valgrinded processes will "play nicely" in
3195 situations where communication with other processes, or the kernel,
3196 is done through shared memory and coordinated with such atomic
3197 instructions. Prior to this change, such arrangements usually
3198 resulted in hangs, races or other synchronisation failures, because
3199 Valgrind did not honour atomicity of such instructions.
3202 * A new experimental tool, BBV, has been added. BBV generates basic
3203 block vectors for use with the SimPoint analysis tool, which allows
3204 a program's overall behaviour to be approximated by running only a
3205 fraction of it. This is useful for computer architecture
3206 researchers. You can run BBV by specifying --tool=exp-bbv (the
3207 "exp-" prefix is short for "experimental"). BBV was written by
3211 * Modestly improved support for running Windows applications under
3212 Wine. In particular, initial support for reading Windows .PDB debug
3213 information has been added.
3216 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
3217 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
3221 * The Valgrind client requests VALGRIND_PRINTF and
3222 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
3223 the string was always printed immediately on its own line. Now, the
3224 string will be added to a buffer but not printed until a newline is
3225 encountered, or other Valgrind output is printed (note that for
3226 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
3227 "other Valgrind output"). This allows you to use multiple
3228 VALGRIND_PRINTF calls to build up a single output line, and also to
3229 print multiple output lines with a single request (by embedding
3230 multiple newlines in the string).
3233 * The graphs drawn by Massif's ms_print program have changed slightly:
3235 - The half-height chars '.' and ',' are no longer drawn, because
3236 they are confusing. The --y option can be used if the default
3237 y-resolution is not high enough.
3239 - Horizontal lines are now drawn after the top of a snapshot if
3240 there is a gap until the next snapshot. This makes it clear that
3241 the memory usage has not dropped to zero between snapshots.
3244 * Something that happened in 3.4.0, but wasn't clearly announced: the
3245 option --read-var-info=yes can be used by some tools (Memcheck,
3246 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
3247 variable type and location information. This makes those tools
3248 start up more slowly and increases memory consumption, but
3249 descriptions of data addresses in error messages become more
3253 * exp-Omega, an experimental instantaneous leak-detecting tool, was
3254 disabled in 3.4.0 due to a lack of interest and maintenance,
3255 although the source code was still in the distribution. The source
3256 code has now been removed from the distribution. For anyone
3257 interested, the removal occurred in SVN revision r10247.
3260 * Some changes have been made to the build system.
3262 - VEX/ is now integrated properly into the build system. This means
3263 that dependency tracking within VEX/ now works properly, "make
3264 install" will work without requiring "make" before it, and
3265 parallel builds (ie. 'make -j') now work (previously a
3266 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
3267 was effectively ignored).
3269 - The --with-vex configure option has been removed. It was of
3270 little use and removing it simplified the build system.
3272 - The location of some install files has changed. This should not
3273 affect most users. Those who might be affected:
3275 * For people who use Valgrind with MPI programs, the installed
3276 libmpiwrap.so library has moved from
3277 $(INSTALL)/<platform>/libmpiwrap.so to
3278 $(INSTALL)/libmpiwrap-<platform>.so.
3280 * For people who distribute standalone Valgrind tools, the
3281 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
3282 have moved to $(INSTALL)/libcoregrind-<platform>.a.
3284 These changes simplify the build system.
3286 - Previously, all the distributed suppression (*.supp) files were
3287 installed. Now, only default.supp is installed. This should not
3288 affect users as the other installed suppression files were not
3289 read; the fact that they were installed was a mistake.
3292 * KNOWN LIMITATIONS:
3294 - Memcheck is unusable with the Intel compiler suite version 11.1,
3295 when it generates code for SSE2-and-above capable targets. This
3296 is because of icc's use of highly optimised inlined strlen
3297 implementations. It causes Memcheck to report huge numbers of
3298 false errors even in simple programs. Helgrind and DRD may also
3301 Versions 11.0 and earlier may be OK, but this has not been
3305 The following bugs have been fixed or resolved. Note that "n-i-bz"
3306 stands for "not in bugzilla" -- that is, a bug that was reported to us
3307 but never got a bugzilla entry. We encourage you to file bugs in
3308 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3309 mailing the developers (or mailing lists) directly -- bugs that are
3310 not entered into bugzilla tend to get forgotten about or ignored.
3312 To see details of a given bug, visit
3313 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3314 where XXXXXX is the bug number as listed below.
3316 84303 How about a LockCheck tool?
3317 91633 dereference of null ptr in vgPlain_st_basetype
3318 97452 Valgrind doesn't report any pthreads problems
3319 100628 leak-check gets assertion failure when using
3320 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
3321 108528 NPTL pthread cleanup handlers not called
3322 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
3323 110128 mallinfo is not implemented...
3324 110770 VEX: Generated files not always updated when making valgrind
3325 111102 Memcheck: problems with large (memory footprint) applications
3326 115673 Vex's decoder should never assert
3327 117564 False positive: Syscall param clone(child_tidptr) contains
3328 uninitialised byte(s)
3329 119404 executing ssh from inside valgrind fails
3330 133679 Callgrind does not write path names to sources with dwarf debug
3332 135847 configure.in problem with non gnu compilers (and possible fix)
3333 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
3334 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
3335 136230 memcheck reports "possibly lost", should be "still reachable"
3336 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
3337 137904 Valgrind reports a memory leak when using POSIX threads,
3339 139076 valgrind VT_GETSTATE error
3340 142228 complaint of elf_dynamic_do_rela in trivial usage
3341 145347 spurious warning with USBDEVFS_REAPURB
3342 148441 (wine) can't find memory leak in Wine, win32 binary
3344 148742 Leak-check fails assert on exit
3345 149878 add (proper) check for calloc integer overflow
3346 150606 Call graph is broken when using callgrind control
3347 152393 leak errors produce an exit code of 0. I need some way to
3348 cause leak errors to result in a nonzero exit code.
3349 157154 documentation (leak-resolution doc speaks about num-callers
3350 def=4) + what is a loss record
3351 159501 incorrect handling of ALSA ioctls
3352 162020 Valgrinding an empty/zero-byte file crashes valgrind
3353 162482 ppc: Valgrind crashes while reading stabs information
3354 162718 x86: avoid segment selector 0 in sys_set_thread_area()
3355 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
3356 163560 VEX/test_main.c is missing from valgrind-3.3.1
3357 164353 malloc_usable_size() doesn't return a usable size
3358 165468 Inconsistent formatting in memcheck manual -- please fix
3359 169505 main.c:286 (endOfInstr):
3360 Assertion 'ii->cost_offset == *cost_offset' failed
3361 177206 Generate default.supp during compile instead of configure
3362 177209 Configure valt_load_address based on arch+os
3363 177305 eventfd / syscall 323 patch lost
3364 179731 Tests fail to build because of inlining of non-local asm labels
3365 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
3366 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
3367 181594 Bogus warning for empty text segment
3368 181707 dwarf doesn't require enumerations to have name
3369 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3370 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3371 Assertion '!already_present' failed.
3372 185359 exp-ptrcheck: unhandled syscall getresuid()
3373 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
3374 185816 Valgrind is unable to handle debug info for files with split
3375 debug info that are prelinked afterwards
3376 185980 [darwin] unhandled syscall: sem_open
3377 186238 bbToIR_AMD64: disInstr miscalculated next %rip
3378 186507 exp-ptrcheck unhandled syscalls prctl, etc.
3379 186790 Suppression pattern used for leaks are not reported
3380 186796 Symbols with length>200 in suppression files are ignored
3381 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
3382 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
3383 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
3384 188046 bashisms in the configure script
3385 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
3386 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
3387 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
3388 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
3389 assertion fail "!lock->heldBy"
3390 188427 Add support for epoll_create1 (with patch)
3391 188530 Support for SIOCGSTAMPNS
3392 188560 Include valgrind.spec in the tarball
3393 188572 Valgrind on Mac should suppress setenv() mem leak
3394 189054 Valgrind fails to build because of duplicate non-local asm labels
3395 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
3396 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
3397 189763 drd assertion failure: s_threadinfo[tid].is_recording
3398 190219 unhandled syscall: 328 (x86-linux)
3399 190391 dup of 181394; see above
3400 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
3401 190820 No debug information on powerpc-linux
3402 191095 PATCH: Improve usbdevfs ioctl handling
3403 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
3405 191189 --xml=yes should obey --gen-suppressions=all
3406 191192 syslog() needs a suppression on macosx
3407 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
3408 191761 getrlimit on MacOSX
3409 191992 multiple --fn-skip only works sometimes; dependent on order
3410 192634 V. reports "aspacem sync_check_mapping_callback:
3411 segment mismatch" on Darwin
3412 192954 __extension__ missing on 2 client requests
3413 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
3414 194474 "INSTALL" file has different build instructions than "README"
3415 194671 Unhandled syscall (sem_wait?) from mac valgrind
3416 195069 memcheck: reports leak (memory still reachable) for
3418 195169 drd: (vgDrd_barrier_post_wait):
3419 Assertion 'r->sg[p->post_iteration]' failed.
3420 195268 valgrind --log-file doesn't accept ~/...
3421 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
3422 195860 WARNING: unhandled syscall: unix:223
3423 196528 need a error suppression for pthread_rwlock_init under os x?
3424 197227 Support aio_* syscalls on Darwin
3425 197456 valgrind should reject --suppressions=(directory)
3426 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
3427 197591 unhandled syscall 27 (mincore)
3428 197793 Merge DCAS branch to the trunk == 85756, 142103
3429 197794 Avoid duplicate filenames in Vex
3430 197898 make check fails on current SVN
3431 197901 make check fails also under exp-ptrcheck in current SVN
3432 197929 Make --leak-resolution=high the default
3433 197930 Reduce spacing between leak reports
3434 197933 Print command line of client at start-up, and shorten preamble
3435 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
3436 198395 add BBV to the distribution as an experimental tool
3437 198624 Missing syscalls on Darwin: 82, 167, 281, 347
3438 198649 callgrind_annotate doesn't cumulate counters
3439 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
3440 199977 Valgrind complains about an unrecognized instruction in the
3441 atomic_incs test program
3442 200029 valgrind isn't able to read Fedora 12 debuginfo
3443 200760 darwin unhandled syscall: unix:284
3444 200827 DRD doesn't work on Mac OS X
3445 200990 VG_(read_millisecond_timer)() does not work correctly
3446 201016 Valgrind does not support pthread_kill() on Mac OS
3447 201169 Document --read-var-info
3448 201323 Pre-3.5.0 performance sanity checking
3449 201384 Review user manual for the 3.5.0 release
3450 201585 mfpvr not implemented on ppc
3451 201708 tests failing because x86 direction flag is left set
3452 201757 Valgrind doesn't handle any recent sys_futex additions
3453 204377 64-bit valgrind can not start a shell script
3454 (with #!/path/to/shell) if the shell is a 32-bit executable
3455 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
3456 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
3457 about memory allocation events with memory access tracing enabled
3458 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
3460 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
3464 Release 3.4.1 (28 February 2009)
3465 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3466 3.4.1 is a bug-fix release that fixes some regressions and assertion
3467 failures in debug info reading in 3.4.0, most notably incorrect stack
3468 traces on amd64-linux on older (glibc-2.3 based) systems. Various
3469 other debug info problems are also fixed. A number of bugs in the
3470 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
3472 In view of the fact that 3.4.0 contains user-visible regressions
3473 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
3474 encouraged to ship 3.4.1 in preference to 3.4.0.
3476 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
3477 bugzilla" -- that is, a bug that was reported to us but never got a
3478 bugzilla entry. We encourage you to file bugs in bugzilla
3479 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
3480 developers (or mailing lists) directly -- bugs that are not entered
3481 into bugzilla tend to get forgotten about or ignored.
3483 n-i-bz Fix various bugs reading icc-11 generated debug info
3484 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
3485 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
3486 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
3487 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
3488 179618 exp-ptrcheck crashed / exit prematurely
3489 179624 helgrind: false positive races with pthread_create and
3490 recv/open/close/read
3491 134207 pkg-config output contains @VG_PLATFORM@
3492 176926 floating point exception at valgrind startup with PPC 440EPX
3493 181594 Bogus warning for empty text segment
3494 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
3495 181707 Dwarf3 doesn't require enumerations to have name
3496 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3497 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3498 Assertion '!already_present' failed.
3499 185359 exp-ptrcheck unhandled syscall getresuid()
3501 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
3502 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
3506 Release 3.4.0 (2 January 2009)
3507 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3508 3.4.0 is a feature release with many significant improvements and the
3509 usual collection of bug fixes. This release supports X86/Linux,
3510 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
3511 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
3513 3.4.0 brings some significant tool improvements. Memcheck can now
3514 report the origin of uninitialised values, the thread checkers
3515 Helgrind and DRD are much improved, and we have a new experimental
3516 tool, exp-Ptrcheck, which is able to detect overruns of stack and
3517 global arrays. In detail:
3519 * Memcheck is now able to track the origin of uninitialised values.
3520 When it reports an uninitialised value error, it will try to show
3521 the origin of the value, as either a heap or stack allocation.
3522 Origin tracking is expensive and so is not enabled by default. To
3523 use it, specify --track-origins=yes. Memcheck's speed will be
3524 essentially halved, and memory usage will be significantly
3525 increased. Nevertheless it can drastically reduce the effort
3526 required to identify the root cause of uninitialised value errors,
3527 and so is often a programmer productivity win, despite running more
3530 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
3531 3.4.0, will be released shortly.
3533 * Helgrind's race detection algorithm has been completely redesigned
3534 and reimplemented, to address usability and scalability concerns:
3536 - The new algorithm has a lower false-error rate: it is much less
3537 likely to report races that do not really exist.
3539 - Helgrind will display full call stacks for both accesses involved
3540 in a race. This makes it easier to identify the root causes of
3543 - Limitations on the size of program that can run have been removed.
3545 - Performance has been modestly improved, although that is very
3548 - Direct support for Qt4 threading has been added.
3550 - pthread_barriers are now directly supported.
3552 - Helgrind works well on all supported Linux targets.
3554 * The DRD thread debugging tool has seen major improvements:
3556 - Greatly improved performance and significantly reduced memory
3559 - Support for several major threading libraries (Boost.Thread, Qt4,
3560 glib, OpenMP) has been added.
3562 - Support for atomic instructions, POSIX semaphores, barriers and
3563 reader-writer locks has been added.
3565 - Works now on PowerPC CPUs too.
3567 - Added support for printing thread stack usage at thread exit time.
3569 - Added support for debugging lock contention.
3571 - Added a manual for Drd.
3573 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
3574 checks for misuses of pointers. In that sense it is a bit like
3575 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
3576 detect overruns of stack and global arrays, it can detect
3577 arbitrarily far out-of-bounds accesses to heap blocks, and it can
3578 detect accesses heap blocks that have been freed a very long time
3579 ago (millions of blocks in the past).
3581 Ptrcheck currently works only on x86-linux and amd64-linux. To use
3582 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
3583 of the main Valgrind documentation. As this is an experimental
3584 tool, we would be particularly interested in hearing about your
3585 experiences with it.
3587 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
3588 longer built by default, although the code remains in the repository
3589 and the tarball. This is due to three factors: a perceived lack of
3590 users, a lack of maintenance, and concerns that it may not be
3591 possible to achieve reliable operation using the existing design.
3593 * As usual, support for the latest Linux distros and toolchain
3594 components has been added. It should work well on Fedora Core 10,
3595 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
3596 state) is supported, as is glibc-2.9. The C++ demangler has been
3597 updated so as to work well with C++ compiled by even the most recent
3600 * You can now use frame-level wildcards in suppressions. This was a
3601 frequently-requested enhancement. A line "..." in a suppression now
3602 matches zero or more frames. This makes it easier to write
3603 suppressions which are precise yet insensitive to changes in
3606 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
3608 * Very basic support for IBM Power6 has been added (64-bit processes only).
3610 * Valgrind is now cross-compilable. For example, it is possible to
3611 cross compile Valgrind on an x86/amd64-linux host, so that it runs
3612 on a ppc32/64-linux target.
3614 * You can set the main thread's stack size at startup using the
3615 new --main-stacksize= flag (subject of course to ulimit settings).
3616 This is useful for running apps that need a lot of stack space.
3618 * The limitation that you can't use --trace-children=yes together
3619 with --db-attach=yes has been removed.
3621 * The following bugs have been fixed. Note that "n-i-bz" stands for
3622 "not in bugzilla" -- that is, a bug that was reported to us but
3623 never got a bugzilla entry. We encourage you to file bugs in
3624 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3625 mailing the developers (or mailing lists) directly.
3627 n-i-bz Make return types for some client requests 64-bit clean
3628 n-i-bz glibc 2.9 support
3629 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
3630 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
3631 n-i-bz Building in an env without gdb gives bogus gdb attach
3632 92456 Tracing the origin of uninitialised memory
3633 106497 Valgrind does not demangle some C++ template symbols
3635 151612 Suppression with "..." (frame-level wildcards in .supp files)
3636 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
3637 159285 unhandled syscall:25 (stime, on x86-linux)
3638 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
3639 160954 ppc build of valgrind crashes with illegal instruction (isel)
3640 160956 mallinfo implementation, w/ patch
3641 162092 Valgrind fails to start gnome-system-monitor
3642 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
3643 163794 assertion failure with "--track-origins=yes"
3644 163933 sigcontext.err and .trapno must be set together
3645 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
3646 164476 Missing kernel module loading system calls
3647 164669 SVN regression: mmap() drops posix file locks
3648 166581 Callgrind output corruption when program forks
3649 167288 Patch file for missing system calls on Cell BE
3650 168943 unsupported scas instruction pentium
3651 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
3652 172417 x86->IR: 0x82 ...
3653 172563 amd64->IR: 0xD9 0xF5 - fprem1
3654 173099 .lds linker script generation error
3655 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
3656 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
3658 174908 --log-file value not expanded correctly for core file
3659 175044 Add lookup_dcookie for amd64
3660 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
3662 Developer-visible changes:
3664 * Valgrind's debug-info reading machinery has been majorly overhauled.
3665 It can now correctly establish the addresses for ELF data symbols,
3666 which is something that has never worked properly before now.
3668 Also, Valgrind can now read DWARF3 type and location information for
3669 stack and global variables. This makes it possible to use the
3670 framework to build tools that rely on knowing the type and locations
3671 of stack and global variables, for example exp-Ptrcheck.
3673 Reading of such information is disabled by default, because most
3674 tools don't need it, and because it is expensive in space and time.
3675 However, you can force Valgrind to read it, using the
3676 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
3677 make use of such information, if present, to provide source-level
3678 descriptions of data addresses in the error messages they create.
3680 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
3681 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).