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.
22 * Majorly overhauled register allocator (add details)
24 * ================== PLATFORM CHANGES =================
26 * Preliminary support for macOS 10.13 has been added.
27 * mips: support for MIPS32/MIPS64 Revision 6 has been added.
28 * mips: support for MIPS SIMD architecture (MSA) has been added.
29 * mips: support for MIPS N32 ABI support has been added.
31 * ==================== TOOL CHANGES ====================
35 - Addition of a flag --delta-stacktrace=no|yes [yes on linux amd64/x86],
36 specifying how full history stack traces are computed. Using yes can speed
37 up Helgrind by 25% when using --history-level=full.
39 * ==================== OTHER CHANGES ====================
41 * A new configure option --enable-lto=yes allows to build Valgrind
42 with link time optimisation. If the toolchain supports it,
43 this produces a smaller/faster Valgrind (up to 10%).
44 Note that if you are doing some Valgrind development, --enable-lto=yes
45 significantly slows down the build.
47 * ==================== FIXED BUGS ====================
49 The following bugs have been fixed or resolved. Note that "n-i-bz"
50 stands for "not in bugzilla" -- that is, a bug that was reported to us
51 but never got a bugzilla entry. We encourage you to file bugs in
52 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
53 than mailing the developers (or mailing lists) directly -- bugs that
54 are not entered into bugzilla tend to get forgotten about or ignored.
56 To see details of a given bug, visit
57 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
58 where XXXXXX is the bug number as listed below.
60 79362 Debug info is lost for .so files when they are dlclose'd
61 208052 strlcpy error when n = 0
62 255603 exp-sgcheck Assertion '!already_present' failed
63 338252 building valgrind with -flto (link time optimisation) fails
64 345763 MIPS N32 ABI support
65 368913 WARNING: unhandled arm64-linux syscall: 117 (ptrace)
66 == 388664 unhandled arm64-linux syscall: 117 (ptrace)
67 372347 Replacement problem of the additional c++14/c++17 new/delete operators
68 376257 helgrind history full speed up using a cached stack
69 379373 Fix syscall param msg->desc.port.name points to uninitialised byte(s)
71 379748 Fix missing pselect syscall (OS X 10.11)
72 380397 s390x: __GI_strcspn() replacemenet needed
73 381162 possible array overrun in VEX register allocator
74 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
75 381274 powerpc too chatty even with --sigill-diagnostics=no
76 381289 epoll_pwait can have a NULL sigmask
77 381553 VEX register allocator v3
78 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
79 381769 Use ucontext_t instead of struct ucontext
80 381805 arm32 needs ld.so index hardwire for new glibc security fixes
81 382256 gz compiler flag test doesn't work for gold
82 382407 vg_perf needs "--terse" command line option
83 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
84 382563 MIPS MSA ASE support
85 382998 xml-socket doesn't work
86 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
87 383723 Fix missing kevent_qos syscall (macOS 10.11)
88 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
89 384096 Mention AddrCheck at Memcheck's command line option [..]
90 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
91 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
92 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
93 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
94 384337 performance improvements to VEX register allocator v2 and v3
95 384526 reduce number of spill insns generated by VEX register allocator v3
96 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
97 384631 Sanitise client args as printed with -v
98 384633 Add a simple progress-reporting facility
99 384987 VEX regalloc: allocate caller-save registers for short lived vregs
100 385055 PPC VEX temporary storage exhausted
101 385182 PPC64 is missing support for the DSCR
102 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
103 385207 PPC64, generate_store_FPRF() generates too many Iops
104 385208 PPC64, xxperm instruction exhausts temporary memory
105 385210 PPC64, vpermr instruction could exhaust temporary memory
106 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
107 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
108 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
109 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
110 385408 s390x: z13 vector "support" instructions not implemented
111 385412 s390x: new non-vector z13 instructions not implemented
112 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
113 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
114 385939 Optionally exit on the first error
115 386318 valgrind.org/info/tools.html is missing SGCheck
116 386425 running valgrind + wine on armv7l gives illegal opcode
117 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
118 387410 MIPSr6 support
119 387664 Memcheck: make expensive-definedness-checks be the default
120 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
121 387766 asm shifts cause false positive "Conditional jump or move depends
122 on uninitialised value"
123 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
124 388786 Support bpf syscall in amd64 Linux
125 388862 Add replacements for wmemchr and wcsnlen on Linux
126 389065 valgrind meets gcc flag -Wlogical-op
127 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
128 390471 suppression by specification of source-file line number
129 390723 make xtree dump files world wide readable, similar to log files
130 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
131 392118 unhandled amd64-linux syscall: 332 (statx)
132 392449 callgrind not clearing the number of calls properly
133 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
134 lxssp, stxssp and stxvl instructions.
135 393023 callgrind_control risks using the wrong vgdb
136 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
137 393099 posix_memalign() invalid write if alignment == 0
138 395709 PPC64 is missing support for the xvnegsp instruction
139 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
141 396887 arch_prctl should return EINVAL on unknown option
142 397012 glibc ld.so uses arch_prctl on i386
144 n-i-bz Fix missing workq_ops operations (macOS)
145 n-i-bz fix bug in strspn replacement
146 n-i-bz Add support for the Linux BLKFLSBUF ioctl
147 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
148 n-i-bz Fix possible stack trashing by semctl syscall wrapping
149 n-i-bz Add support for the Linux membarrier() system call
151 Release 3.13.0 (15 June 2017)
152 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
154 3.13.0 is a feature release with many improvements and the usual collection of
157 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
158 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
159 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
160 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
162 * ==================== CORE CHANGES ===================
164 * The translation cache size has been increased to keep up with the demands of
165 large applications. The maximum number of sectors has increased from 24 to
166 48. The default number of sectors has increased from 16 to 32 on all
167 targets except Android, where the increase is from 6 to 12.
169 * The amount of memory that Valgrind can use has been increased from 64GB to
170 128GB. In particular this means your application can allocate up to about
171 60GB when running on Memcheck.
173 * Valgrind's default load address has been changed from 0x3800'0000 to
174 0x5800'0000, so as to make it possible to load larger executables. This
175 should make it possible to load executables of size at least 1200MB.
177 * A massive spaceleak caused by reading compressed debuginfo files has been
178 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
181 * The C++ demangler has been updated.
183 * Support for demangling Rust symbols has been added.
185 * A new representation of stack traces, the "XTree", has been added. An XTree
186 is a tree of stacktraces with data associated with the stacktraces. This is
187 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
188 consumption of your program. Reporting is controlled by the new options
189 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
191 A report can also be produced on demand using the gdbserver monitor command
192 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
193 format' and 'massif format. The existing visualisers for these formats (e.g.
194 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
195 analyse these reports.
197 Memcheck can also produce XTree leak reports using the Callgrind file
198 format. For more details, see the user manual.
200 * ================== PLATFORM CHANGES =================
202 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
204 * amd64: fixes for JIT failure problems on long AVX2 code blocks
206 * amd64 and x86: support for CET prefixes has been added
208 * arm32: a few missing ARMv8 instructions have been implemented
210 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
211 Store-Conditional instructions has been added. This is to deal with
212 processor implementations that implement the LL/SC specifications strictly
213 and as a result cause Valgrind to hang in certain situations. The
214 alternative implementation is automatically enabled at startup, as required.
215 You can use the option --sim-hints=fallback-llsc to force-enable it if you
218 * Support for OSX 10.12 has been improved.
220 * On Linux, clone handling has been improved to honour CLONE_VFORK that
221 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
222 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
223 CLONE_VM exact semantics will (still) not work.
225 * The TileGX/Linux port has been removed because it appears to be both unused
228 * ==================== TOOL CHANGES ====================
232 - Memcheck should give fewer false positives when running optimised
233 Clang/LLVM generated code.
235 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
237 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
238 to produce the end of execution leak report in a xtree callgrind format
241 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
242 the leak report in an xtree file.
246 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
248 - For some workloads (typically, for big applications), Massif memory
249 consumption and CPU consumption has decreased significantly.
253 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
255 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
256 for Ada gnat compiled applications.
258 * ==================== OTHER CHANGES ====================
260 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
261 append the inner guest stacktrace to the inner host stacktrace. This helps
262 to investigate the errors reported by the outer, when they are caused by the
263 inner guest program (such as an inner regtest). See README_DEVELOPERS for
266 * To allow fast detection of callgrind files by desktop environments and file
267 managers, the format was extended to have an optional first line that
268 uniquely identifies the format ("# callgrind format"). Callgrind creates
269 this line now, as does the new xtree functionality.
271 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
272 have a new %n format letter that is replaced by a sequence number.
274 * "--version -v" now shows the SVN revision numbers from which Valgrind was
277 * ==================== FIXED BUGS ====================
279 The following bugs have been fixed or resolved. Note that "n-i-bz"
280 stands for "not in bugzilla" -- that is, a bug that was reported to us
281 but never got a bugzilla entry. We encourage you to file bugs in
282 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
283 than mailing the developers (or mailing lists) directly -- bugs that
284 are not entered into bugzilla tend to get forgotten about or ignored.
286 To see details of a given bug, visit
287 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
288 where XXXXXX is the bug number as listed below.
290 162848 --log-file output isn't split when a program forks
291 340777 Illegal instruction on mips (ar71xx)
292 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
293 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
294 to a different stack.
295 344139 x86 stack-seg overrides, needed by the Wine people
296 344524 store conditional of guest applications always fail - observed on
298 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
299 352395 Please provide SVN revision info in --version -v
300 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
301 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
302 INVALID_POSIX_THREADID' failed
303 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
304 barrier implementation
305 358697 valgrind.h: Some code remains even when defining NVALGRIND
306 359202 Add musl libc configure/compile
307 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
308 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
309 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
310 362223 assertion failed when .valgrindrc is a directory instead of a file
311 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
312 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
313 368507 can't malloc chunks larger than about 34GB
314 368529 Android arm target link error, missing atexit and pthread_atfork
315 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
316 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
317 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
318 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
319 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
320 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
321 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
322 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
323 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
324 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
325 370028 Reduce the number of compiler warnings on MIPS platforms
326 370635 arm64 missing syscall getcpu
327 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
328 371227 Clean AArch64 syscall table
329 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
330 371471 Valgrind complains about non legit memory leaks on placement new (C++)
331 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
332 371503 disInstr(arm64): unhandled instruction 0xF89F0000
333 371869 support '%' in symbol Z-encoding
334 371916 execution tree xtree concept
335 372120 c++ demangler demangles symbols which are not c++
336 372185 Support of valgrind on ARMv8 with 32 bit executable
337 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
338 372195 Power PC, xxsel instruction is not always recognized.
339 372504 Hanging on exit_group
340 372600 process loops forever when fatal signals are arriving quickly
341 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
342 373046 Stacks registered by core are never deregistered
343 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
344 373086 Implement additional Xen hypercalls
345 373192 Calling posix_spawn in glibc 2.24 completely broken
346 373488 Support for fanotify API on ARM64 architecture
347 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
348 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
349 373938 const IRExpr arguments for matchIRExpr()
350 374719 some spelling fixes
351 374963 increase valgrind's load address to prevent mmap failure
352 375514 valgrind_get_tls_addr() does not work in case of static TLS
353 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
354 for ML_(find_rx_mapping)()
355 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
356 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
357 == 377159 "vex: the `impossible' happened" still present
358 == 375150 Assertion 'tres.status == VexTransOK' failed
359 == 378068 valgrind crashes on AVX2 function in FFmpeg
360 376142 Segfaults on MIPS Cavium Octeon boards
361 376279 disInstr(arm64): unhandled instruction 0xD50320FF
362 376455 Solaris: unhandled syscall lgrpsys(180)
363 376518 Solaris: unhandled fast trap getlgrp(6)
364 376611 ppc64 and arm64 don't know about prlimit64 syscall
365 376729 PPC64, remove R2 from the clobber list
367 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
368 to be wrongly marked as addressable
369 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
370 PIE enabled by default
371 377376 memcheck/tests/linux/getregset fails with glibc2.24
372 377427 PPC64, lxv instruction failing on odd destination register
373 377478 PPC64: ISA 3.0 setup fixes
374 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
375 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
376 and 2 args for FUTEX_TRYLOCK_PI
377 377717 Fix massive space leak when reading compressed debuginfo sections
378 377891 Update Xen 4.6 domctl wrappers
379 377930 fcntl syscall wrapper is missing flock structure check
380 378524 libvexmultiarch_test regression on s390x and ppc64
381 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
382 378673 Update libiberty demangler
383 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
384 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
385 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
386 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
387 (task_register_dyld_image_infos)
388 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
389 (task_register_dyld_shared_cache_image_info)
390 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
391 379473 MIPS: add support for rdhwr cycle counter register
392 379504 remove TileGX/Linux port
393 379525 Support more x86 nop opcodes
394 379838 disAMode(x86): not an addr!
395 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
396 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
397 379895 clock_gettime does not execute POST syscall wrapper
398 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
399 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
400 380200 xtree generated callgrind files refer to files without directory name
401 380202 Assertion failure for cache line size (cls == 64) on aarch64.
402 380397 s390x: __GI_strcspn() replacement needed
403 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
405 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
406 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
407 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
411 Release 3.12.0 (20 October 2016)
412 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
414 3.12.0 is a feature release with many improvements and the usual
415 collection of bug fixes.
417 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
418 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
419 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
420 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
421 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
422 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
424 * ================== PLATFORM CHANGES =================
426 * POWER: Support for ISA 3.0 has been added
428 * mips: support for O32 FPXX ABI has been added.
429 * mips: improved recognition of different processors
430 * mips: determination of page size now done at run time
432 * amd64: Partial support for AMD FMA4 instructions.
434 * arm, arm64: Support for v8 crypto and CRC instructions.
436 * Improvements and robustification of the Solaris port.
438 * Preliminary support for MacOS 10.12 (Sierra) has been added.
440 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
441 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
442 where possible. Valgrind's support for 32-bit x86 has stagnated in
443 recent years and has fallen far behind that for 64-bit x86
444 instructions. By contrast 64-bit x86 is well supported, up to and
447 * ==================== TOOL CHANGES ====================
451 - Added meta mempool support for describing a custom allocator which:
452 - Auto-frees all chunks assuming that destroying a pool destroys all
454 - Uses itself to allocate other memory blocks
456 - New flag --ignore-range-below-sp to ignore memory accesses below
457 the stack pointer, if you really have to. The related flag
458 --workaround-gcc296-bugs=yes is now deprecated. Use
459 --ignore-range-below-sp=1024-1 as a replacement.
463 - Improved thread startup time significantly on non-Linux platforms.
467 - Added collection of the metric "tot-blocks-allocd"
469 * ==================== OTHER CHANGES ====================
471 * Replacement/wrapping of malloc/new related functions is now done not just
472 for system libraries by default, but for any globally defined malloc/new
473 related function (both in shared libraries and statically linked alternative
474 malloc implementations). The dynamic (runtime) linker is excluded, though.
475 To only intercept malloc/new related functions in
476 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
477 "nouserintercepts" can be any non-existing library name).
478 This new functionality is not implemented for MacOS X.
480 * The maximum number of callers in a suppression entry is now equal to
481 the maximum size for --num-callers (500).
482 Note that --gen-suppressions=yes|all similarly generates suppressions
483 containing up to --num-callers frames.
485 * New and modified GDB server monitor features:
487 - Valgrind's gdbserver now accepts the command 'catch syscall'.
488 Note that you must have GDB >= 7.11 to use 'catch syscall' with
491 * New option --run-cxx-freeres=<yes|no> can be used to change whether
492 __gnu_cxx::__freeres() cleanup function is called or not. Default is
495 * Valgrind is able to read compressed debuginfo sections in two formats:
496 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
497 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
499 * Modest JIT-cost improvements: the cost of instrumenting code blocks
500 for the most common use case (x86_64-linux, Memcheck) has been
503 * Improved performance for programs that do a lot of discarding of
504 instruction address ranges of 8KB or less.
506 * The C++ symbol demangler has been updated.
508 * More robustness against invalid syscall parameters on Linux.
510 * ==================== FIXED BUGS ====================
512 The following bugs have been fixed or resolved. Note that "n-i-bz"
513 stands for "not in bugzilla" -- that is, a bug that was reported to us
514 but never got a bugzilla entry. We encourage you to file bugs in
515 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
516 than mailing the developers (or mailing lists) directly -- bugs that
517 are not entered into bugzilla tend to get forgotten about or ignored.
519 To see details of a given bug, visit
520 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
521 where XXXXXX is the bug number as listed below.
523 191069 Exiting due to signal not reported in XML output
524 199468 Suppressions: stack size limited to 25
525 while --num-callers allows more frames
526 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
527 278744 cvtps2pd with redundant RexW
528 303877 valgrind doesn't support compressed debuginfo sections.
529 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
530 348345 Assertion fails for negative lineno
531 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
532 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
533 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
534 351804 Crash on generating suppressions for "printf" call on OS X 10.10
535 352197 mips: mmap2() not wrapped correctly for page size > 4096
536 353083 arm64 doesn't implement various xattr system calls
537 353084 arm64 doesn't support sigpending system call
538 353137 www: update info for Supported Platforms
539 353138 www: update "The Valgrind Developers" page
540 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
543 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
544 353398 WARNING: unhandled amd64-solaris syscall: 207
545 353660 XML in auxwhat tag not escaping reserved symbols properly
546 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
547 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
548 353802 ELF debug info reader confused with multiple .rodata sections
549 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
550 353917 unhandled amd64-solaris syscall fchdir(120)
551 353920 unhandled amd64-solaris syscall: 170
552 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
553 354392 unhandled amd64-solaris syscall: 171
554 354797 Vbit test does not include Iops for Power 8 instruction support
555 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
559 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
560 355188 valgrind should intercept all malloc related global functions
561 355454 do not intercept malloc related symbols from the runtime linker
562 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
563 356044 Dwarf line info reader misinterprets is_stmt register
564 356112 mips: replace addi with addiu
565 356393 valgrind (vex) crashes because isZeroU happened
568 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
569 356678 arm64-linux: unhandled syscall 232 (mincore)
570 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
571 356823 Unsupported ARM instruction: stlex
572 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
573 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
574 357673 crash if I try to run valgrind with a binary link with libcurl
575 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
576 357871 pthread_spin_destroy not properly wrapped
577 357887 Calls to VG_(fclose) do not close the file descriptor
578 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
579 358030 support direct socket calls on x86 32bit (new in linux 4.3)
580 358478 drd/tests/std_thread.cpp doesn't build with GCC6
581 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
582 359181 Buffer Overflow during Demangling
583 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
584 359289 s390x: popcnt (B9E1) not implemented
585 359472 The Power PC vsubuqm instruction doesn't always give the correct result
586 359503 Add missing syscalls for aarch64 (arm64)
587 359645 "You need libc6-dbg" help message could be more helpful
588 359703 s390: wire up separate socketcalls system calls
589 359724 getsockname might crash - deref_UInt should call safe_to_deref
590 359733 amd64 implement ld.so strchr/index override like x86
591 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
592 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
594 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
595 359871 Incorrect mask handling in ppoll
596 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
597 360008 Contents of Power vr registers contents is not printed correctly when
598 the --vgdb-shadow-registers=yes option is used
599 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
601 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
602 360425 arm64 unsupported instruction ldpsw
604 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
605 360571 Error about the Android Runtime reading below the stack pointer on ARM
606 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
607 360749 kludge for multiple .rodata sections on Solaris no longer needed
608 360752 raise the number of reserved fds in m_main.c from 10 to 12
609 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
610 361226 s390x: risbgn (EC59) not implemented
611 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
612 361354 ppc64[le]: wire up separate socketcalls system calls
613 361615 Inconsistent termination for multithreaded process terminated by signal
614 361926 Unhandled Solaris syscall: sysfs(84)
615 362009 V dumps core on unimplemented functionality before threads are created
616 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
617 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
618 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
619 362953 Request for an update to the Valgrind Developers page
620 363680 add renameat2() support
621 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
622 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
623 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
624 364058 clarify in manual limitations of array overruns detections
625 364413 pselect sycallwrapper mishandles NULL sigmask
626 364728 Power PC, missing support for several HW registers in
627 get_otrack_shadow_offset_wrk()
628 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
629 365273 Invalid write to stack location reported after signal handler runs
630 365912 ppc64BE segfault during jm-insns test (RELRO)
631 366079 FPXX Support for MIPS32 Valgrind
632 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
633 366344 Multiple unhandled instruction for Aarch64
634 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
635 367995 Integration of memcheck with custom memory allocator
636 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
637 368412 False positive result for altivec capability check
638 368416 Add tc06_two_races_xml.exp output for ppc64
639 368419 Perf Events ioctls not implemented
640 368461 mmapunmap test fails on ppc64
641 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
642 369000 AMD64 fma4 instructions unsupported.
643 369169 ppc64 fails jm_int_isa_2_07 test
644 369175 jm_vec_isa_2_07 test crashes on ppc64
645 369209 valgrind loops and eats up all memory if cwd doesn't exist.
646 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
647 369359 msghdr_foreachfield can crash when handling bad iovec
648 369360 Bad sigprocmask old or new sets can crash valgrind
649 369361 vmsplice syscall wrapper crashes on bad iovec
650 369362 Bad sigaction arguments crash valgrind
651 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
652 369402 Bad set/get_thread_area pointer crashes valgrind
653 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
654 369446 valgrind crashes on unknown fcntl command
655 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
656 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
657 370265 ISA 3.0 HW cap stuff needs updating
658 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
659 372195 Power PC, xxsel instruction is not always recognized
661 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
662 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
663 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
664 n-i-bz Fix ppoll_alarm exclusion on OS X
665 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
666 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
667 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
668 n-i-bz amd64: memcheck false positive with shr %edx
669 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
670 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
671 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
672 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
674 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
675 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
676 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
680 Release 3.11.0 (22 September 2015)
681 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
683 3.11.0 is a feature release with many improvements and the usual
684 collection of bug fixes.
686 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
687 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
688 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
689 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
690 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
691 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
693 * ================== PLATFORM CHANGES =================
695 * Support for Solaris/x86 and Solaris/amd64 has been added.
697 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
699 * Preliminary support for the Tilera TileGX architecture has been added.
701 * s390x: It is now required for the host to have the "long displacement"
702 facility. The oldest supported machine model is z990.
704 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
705 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
707 * The JIT's register allocator is significantly faster, making the JIT
708 as a whole somewhat faster, so JIT-intensive activities, for example
709 program startup, are modestly faster, around 5%.
711 * There have been changes to the default settings of several command
712 line flags, as detailed below.
714 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
715 capable hosts, the simulated CPUID will now indicate AVX2 support.
717 * ==================== TOOL CHANGES ====================
721 - The default value for --leak-check-heuristics has been changed from
722 "none" to "all". This helps to reduce the number of possibly
723 lost blocks, in particular for C++ applications.
725 - The default value for --keep-stacktraces has been changed from
726 "malloc-then-free" to "malloc-and-free". This has a small cost in
727 memory (one word per malloc-ed block) but allows Memcheck to show the
728 3 stacktraces of a dangling reference: where the block was allocated,
729 where it was freed, and where it is acccessed after being freed.
731 - The default value for --partial-loads-ok has been changed from "no" to
732 "yes", so as to avoid false positive errors resulting from some kinds
735 - A new monitor command 'xb <addr> <len>' shows the validity bits of
736 <len> bytes at <addr>. The monitor command 'xb' is easier to use
737 than get_vbits when you need to associate byte data value with
738 their corresponding validity bits.
740 - The 'block_list' monitor command has been enhanced:
741 o it can print a range of loss records
742 o it now accepts an optional argument 'limited <max_blocks>'
743 to control the number of blocks printed.
744 o if a block has been found using a heuristic, then
745 'block_list' now shows the heuristic after the block size.
746 o the loss records/blocks to print can be limited to the blocks
747 found via specified heuristics.
749 - The C helper functions used to instrument loads on
750 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
751 replaced by handwritten assembly sequences. This gives speedups
752 in the region of 0% to 7% for those targets only.
754 - A new command line option, --expensive-definedness-checks=yes|no,
755 has been added. This is useful for avoiding occasional invalid
756 uninitialised-value errors in optimised code. Watch out for
757 runtime degradation, as this can be up to 25%. As always, though,
758 the slowdown is highly application specific. The default setting
763 - A new monitor command 'all_snapshots <filename>' dumps all
764 snapshots taken so far.
768 - Significant memory reduction and moderate speedups for
769 --history-level=full for applications accessing a lot of memory
770 with many different stacktraces.
772 - The default value for --conflict-cache-size=N has been doubled to
773 2000000. Users that were not using the default value should
774 preferably also double the value they give.
776 The default was changed due to the changes in the "full history"
777 implementation. Doubling the value gives on average a slightly more
778 complete history and uses similar memory (or significantly less memory
779 in the worst case) than the previous implementation.
781 - The Helgrind monitor command 'info locks' now accepts an optional
782 argument 'lock_addr', which shows information about the lock at the
785 - When using --history-level=full, the new Helgrind monitor command
786 'accesshistory <addr> [<len>]' will show the recorded accesses for
787 <len> (or 1) bytes at <addr>.
789 * ==================== OTHER CHANGES ====================
791 * The default value for the --smc-check option has been changed from
792 "stack" to "all-non-file" on targets that provide automatic D-I
793 cache coherence (x86, amd64 and s390x). The result is to provide,
794 by default, transparent support for JIT generated and self-modifying
797 * Mac OS X only: the default value for the --dsymutil option has been
798 changed from "no" to "yes", since any serious usage on Mac OS X
799 always required it to be "yes".
801 * The command line options --db-attach and --db-command have been removed.
802 They were deprecated in 3.10.0.
804 * When a process dies due to a signal, Valgrind now shows the signal
805 and the stacktrace at default verbosity (i.e. verbosity 1).
807 * The address description logic used by Memcheck and Helgrind now
808 describes addresses in anonymous segments, file mmap-ed segments,
809 shared memory segments and the brk data segment.
811 * The new option --error-markers=<begin>,<end> can be used to mark the
812 begin/end of errors in textual output mode, to facilitate
813 searching/extracting errors in output files that mix valgrind errors
816 * The new option --max-threads=<number> can be used to change the number
817 of threads valgrind can handle. The default is 500 threads which
818 should be more than enough for most applications.
820 * The new option --valgrind-stacksize=<number> can be used to change the
821 size of the private thread stacks used by Valgrind. This is useful
822 for reducing memory use or increasing the stack size if Valgrind
823 segfaults due to stack overflow.
825 * The new option --avg-transtab-entry-size=<number> can be used to specify
826 the expected instrumented block size, either to reduce memory use or
827 to avoid excessive retranslation.
829 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
831 * New and modified GDB server monitor features:
833 - When a signal is reported in GDB, you can now use the GDB convenience
834 variable $_siginfo to examine detailed signal information.
836 - Valgrind's gdbserver now allows the user to change the signal
837 to deliver to the process. So, use 'signal SIGNAL' to continue execution
838 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
839 continue without passing the signal to the process.
841 - With GDB >= 7.10, the command 'target remote'
842 will automatically load the executable file of the process running
843 under Valgrind. This means you do not need to specify the executable
844 file yourself, GDB will discover it itself. See GDB documentation about
845 'qXfer:exec-file:read' packet for more info.
847 * ==================== FIXED BUGS ====================
849 The following bugs have been fixed or resolved. Note that "n-i-bz"
850 stands for "not in bugzilla" -- that is, a bug that was reported to us
851 but never got a bugzilla entry. We encourage you to file bugs in
852 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
853 than mailing the developers (or mailing lists) directly -- bugs that
854 are not entered into bugzilla tend to get forgotten about or ignored.
856 To see details of a given bug, visit
857 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
858 where XXXXXX is the bug number as listed below.
860 116002 VG_(printf): Problems with justification of strings and integers
861 155125 avoid cutting away file:lineno after long function name
862 197259 Unsupported arch_prtctl PR_SET_GS option
863 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
864 201216 Fix Valgrind does not support pthread_sigmask() on OS X
865 201435 Fix Darwin: -v does not show kernel version
866 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
867 211256 Fixed an outdated comment regarding the default platform.
868 211529 Incomplete call stacks for code compiled by newer versions of MSVC
869 211926 Avoid compilation warnings in valgrind.h with -pedantic
870 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
872 226609 Crediting upstream authors in man page
873 231257 Valgrind omits path when executing script from shebang line
874 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
875 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
876 269360 s390x: Fix addressing mode selection for compare-and-swap
877 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
879 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
880 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
881 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
882 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
883 330147 libmpiwrap PMPI_Get_count returns undefined value
884 333051 mmap of huge pages fails due to incorrect alignment
886 334802 valgrind does not always explain why a given option is bad
887 335618 mov.w rN, pc/sp (ARM32)
888 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
893 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
894 338602 AVX2 bit in CPUID missing
895 338606 Strange message for scripts with invalid interpreter
896 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
897 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
898 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
900 339156 gdbsrv not called for fatal signal
901 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
902 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
903 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
904 339442 Fix testsuite build failure on OS X 10.9
905 339542 Enable compilation with Intel's ICC compiler
906 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
907 339688 Mac-specific ASM does not support .version directive (cpuid,
908 tronical and pushfpopf tests)
909 339745 Valgrind crash when check Marmalade app (partial fix)
910 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
911 339778 Linux/TileGx platform support for Valgrind
912 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
913 339789 Fix none/tests/execve test on Mac OS X 10.9
914 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
915 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
916 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
917 340392 Allow user to select more accurate definedness checking in memcheck
918 to avoid invalid complaints on optimised code
919 340430 Fix some grammatical weirdness in the manual.
920 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
921 341419 Signal handler ucontext_t not filled out correctly on OS X
922 341539 VG_(describe_addr) should not describe address as belonging to client
923 segment if it is past the heap end
924 341613 Enable building of manythreads and thread-exits tests on Mac OS X
925 341615 Fix none/tests/darwin/access_extended test on Mac OS X
926 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
927 341789 aarch64: shmat fails with valgrind on ARMv8
928 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
929 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
930 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
931 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
932 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
933 342221 socket connect false positive uninit memory for unknown af family
934 342353 Allow dumping full massif output while valgrind is still running
935 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
939 342603 Add I2C_SMBUS ioctl support
940 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
941 342683 Mark memory past the initial brk limit as unaddressable
942 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
943 342795 Internal glibc __GI_mempcpy call should be intercepted
944 342841 s390x: Support instructions fiebr(a) and fidbr(a)
945 343012 Unhandled syscall 319 (memfd_create)
946 343069 Patch updating v4l2 API support
947 343173 helgrind crash during stack unwind
948 343219 fix GET_STARTREGS for arm
949 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
950 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
951 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
952 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
953 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
954 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
955 343597 ppc64le: incorrect use of offseof macro
956 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
957 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
958 343732 Unhandled syscall 144 (setgid) on aarch64
959 343733 Unhandled syscall 187 (msgctl and related) on aarch64
960 343802 s390x: False positive "conditional jump or move depends on [..]
961 343902 --vgdb=yes doesn't break when --xml=yes is used
962 343967 Don't warn about setuid/setgid/setcap executable for directories
963 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
964 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
965 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
966 344054 www - update info for Solaris/illumos
967 344416 'make regtest' does not work cleanly on OS X
968 344235 Remove duplicate include of pub_core_aspacemgr.h
969 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
970 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
971 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
972 344314 callgrind_annotate ... warnings about commands containing newlines
973 344318 socketcall should wrap recvmmsg and sendmmsg
974 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
975 344416 Fix 'make regtest' does not work cleanly on OS X
976 344499 Fix compilation for Linux kernel >= 4.0.0
977 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
978 unix:349 (__pthread_fchdir)
979 344559 Garbage collection of unused segment names in address space manager
980 344560 Fix stack traces missing penultimate frame on OS X
981 344621 Fix memcheck/tests/err_disable4 test on OS X
982 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
983 344702 Fix missing libobjc suppressions on OS X 10.10
985 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
986 344939 Fix memcheck/tests/xml1 on OS X 10.10
987 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
988 345079 Fix build problems in VEX/useful/test_main.c
989 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
990 345177 arm64: prfm (reg) not implemented
991 345215 Performance improvements for the register allocator
992 345248 add support for Solaris OS in valgrind
993 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
994 345394 Fix memcheck/tests/strchr on OS X
995 345637 Fix memcheck/tests/sendmsg on OS X
996 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
997 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
998 345887 Fix an assertion in the address space manager
999 345928 amd64: callstack only contains current function for small stacks
1000 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1001 345987 MIPS64: Implement cavium LHX instruction
1002 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1003 346185 Fix typo saving altivec register v24
1004 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1005 and LibVEX_GuestPPC64_get_CR()
1006 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1007 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1008 346307 fuse filesystem syscall deadlocks
1009 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1010 346411 MIPS: SysRes::_valEx handling is incorrect
1011 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1012 346474 PPC64 Power 8, spr TEXASRU register not supported
1013 346487 Compiler generates "note" about a future ABI change for PPC64
1014 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1015 and causing spurious "invalid read of size 8" warnings
1016 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1017 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1018 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1019 347322 Power PC regression test cleanup
1020 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1022 347389 unhandled syscall: 373 (Linux ARM syncfs)
1023 347686 Patch set to cleanup PPC64 regtests
1024 347978 Remove bash dependencies where not needed
1025 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1026 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1028 348102 Patch updating v4l2 API support
1029 348247 amd64 front end: jno jumps wrongly when overflow is not set
1030 348269 Improve mmap MAP_HUGETLB support.
1031 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1032 348345 Assertion fails for negative lineno
1033 348377 Unsupported ARM instruction: yield
1034 348565 Fix detection of command line option availability for clang
1035 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1036 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1037 348748 Fix redundant condition
1038 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1039 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1040 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1041 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1042 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1043 349626 Implemented additional Xen hypercalls
1044 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1045 349790 Clean up of the hardware capability checking utilities.
1046 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1047 349874 Fix typos in source code
1048 349879 memcheck: add handwritten assembly for helperc_LOADV*
1049 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1050 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1051 350202 Add limited param to 'monitor block_list'
1052 350290 s390x: Support instructions fixbr(a)
1053 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1054 350809 Fix none/tests/async-sigs for Solaris
1055 350811 Remove reference to --db-attach which has been removed.
1056 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1057 350854 hard-to-understand code in VG_(load_ELF)()
1058 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1059 351386 Solaris: Cannot run ld.so.1 under Valgrind
1060 351474 Fix VG_(iseqsigset) as obvious
1061 351531 Typo in /include/vki/vki-xen-physdev.h header guard
1062 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
1063 351858 ldsoexec support on Solaris
1064 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
1065 352130 helgrind reports false races for printfs using mempcpy on FILE* state
1066 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
1067 352320 arm64 crash on none/tests/nestedfs
1068 352765 Vbit test fails on Power 6
1069 352768 The mbar instruction is missing from the Power PC support
1070 352769 Power PC program priority register (PPR) is not supported
1071 n-i-bz Provide implementations of certain compiler builtins to support
1072 compilers that may not provide those
1073 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
1074 n-i-bz Fix compilation on distros with glibc < 2.5
1075 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
1076 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
1077 n-i-bz mremap did not work properly on shared memory
1078 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
1079 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
1081 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
1082 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
1083 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
1087 Release 3.10.1 (25 November 2014)
1088 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1089 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
1090 and backports fixes for all reported missing AArch64 ARMv8 instructions
1091 and syscalls from the trunk. If you package or deliver 3.10.0 for others
1092 to use, you might want to consider upgrading to 3.10.1 instead.
1094 The following bugs have been fixed or resolved. Note that "n-i-bz"
1095 stands for "not in bugzilla" -- that is, a bug that was reported to us
1096 but never got a bugzilla entry. We encourage you to file bugs in
1097 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1098 than mailing the developers (or mailing lists) directly -- bugs that
1099 are not entered into bugzilla tend to get forgotten about or ignored.
1101 To see details of a given bug, visit
1102 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1103 where XXXXXX is the bug number as listed below.
1105 335440 arm64: ld1 (single structure) is not implemented
1106 335713 arm64: unhanded instruction: prfm (immediate)
1107 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
1108 339182 ppc64: AvSplat ought to load destination vector register with [..]
1109 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
1110 339433 ppc64 lxvw4x instruction uses four 32-byte loads
1111 339645 Use correct tag names in sys_getdents/64 wrappers
1112 339706 Fix false positive for ioctl(TIOCSIG) on linux
1113 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
1114 339853 arm64 times syscall unknown
1115 339855 arm64 unhandled getsid/setsid syscalls
1116 339858 arm64 dmb sy not implemented
1117 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
1118 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
1119 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
1121 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
1122 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
1123 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
1124 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
1125 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
1126 340509 arm64: unhandled instruction fcvtas
1127 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
1128 340632 arm64: unhandled instruction fcvtas
1129 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
1130 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
1131 340788 warning: unhandled syscall: 318 (getrandom)
1132 340807 disInstr(arm): unhandled instruction: 0xEE989B20
1133 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
1134 340922 arm64: unhandled getgroups/setgroups syscalls
1135 350251 Fix typo in VEX utility program (test_main.c).
1136 350407 arm64: unhandled instruction ucvtf (vector, integer)
1137 350809 none/tests/async-sigs breaks when run under cron on Solaris
1138 350811 update README.solaris after r15445
1139 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
1140 350854 strange code in VG_(load_ELF)()
1141 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1142 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
1143 n-i-bz Add missing ]] to terminate CDATA.
1144 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
1145 n-i-bz Enable sys_fadvise64_64 on arm32.
1146 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
1147 n-i-bz Add test cases for all known arm64 load/store instructions.
1148 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
1149 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
1150 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
1151 n-i-bz arm64: Implement "fcvtpu w, s".
1152 n-i-bz arm64: implement ADDP and various others
1153 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
1154 n-i-bz arm64: enable FCVT{A,N}S X,S.
1156 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
1160 Release 3.10.0 (10 September 2014)
1161 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1163 3.10.0 is a feature release with many improvements and the usual
1164 collection of bug fixes.
1166 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1167 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1168 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
1169 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
1170 significantly improved relative to the 3.9.0 release.
1172 * ================== PLATFORM CHANGES =================
1174 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
1175 is mostly complete, and is usable, but some SIMD instructions are as
1178 * Support for little-endian variant of the 64-bit POWER architecture.
1180 * Support for Android on MIPS32.
1182 * Support for 64bit FPU on MIPS32 platforms.
1184 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
1186 * Configuration for and running on Android targets has changed.
1187 See README.android in the source tree for details.
1189 * ================== DEPRECATED FEATURES =================
1191 * --db-attach is now deprecated and will be removed in the next
1192 valgrind feature release. The built-in GDB server capabilities are
1193 superior and should be used instead. Learn more here:
1194 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
1196 * ==================== TOOL CHANGES ====================
1200 - Client code can now selectively disable and re-enable reporting of
1201 invalid address errors in specific ranges using the new client
1202 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
1203 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
1205 - Leak checker: there is a new leak check heuristic called
1206 "length64". This is used to detect interior pointers pointing 8
1207 bytes inside a block, on the assumption that the first 8 bytes
1208 holds the value "block size - 8". This is used by
1209 sqlite3MemMalloc, for example.
1211 - Checking of system call parameters: if a syscall parameter
1212 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
1213 several fields not initialised, an error is now reported for each
1214 field. Previously, an error was reported only for the first
1215 uninitialised field.
1217 - Mismatched alloc/free checking: a new flag
1218 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
1219 such checks if necessary.
1223 - Improvements to error messages:
1225 o Race condition error message involving heap allocated blocks also
1226 show the thread number that allocated the raced-on block.
1228 o All locks referenced by an error message are now announced.
1229 Previously, some error messages only showed the lock addresses.
1231 o The message indicating where a lock was first observed now also
1232 describes the address/location of the lock.
1234 - Helgrind now understands the Ada task termination rules and
1235 creates a happens-before relationship between a terminated task
1236 and its master. This avoids some false positives and avoids a big
1237 memory leak when a lot of Ada tasks are created and terminated.
1238 The interceptions are only activated with forthcoming releases of
1239 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
1241 - A new GDB server monitor command "info locks" giving the list of
1242 locks, their location, and their status.
1246 - callgrind_control now supports the --vgdb-prefix argument,
1247 which is needed if valgrind was started with this same argument.
1249 * ==================== OTHER CHANGES ====================
1251 * Unwinding through inlined function calls. Stack unwinding can now
1252 make use of Dwarf3 inlined-unwind information if it is available.
1253 The practical effect is that inlined calls become visible in stack
1254 traces. The suppression matching machinery has been adjusted
1255 accordingly. This is controlled by the new option
1256 --read-inline-info=yes|no. Currently this is enabled by default
1257 only on Linux and Android targets and only for the tools Memcheck,
1260 * Valgrind can now read EXIDX unwind information on 32-bit ARM
1261 targets. If an object contains both CFI and EXIDX unwind
1262 information, Valgrind will prefer the CFI over the EXIDX. This
1263 facilitates unwinding through system libraries on arm-android
1266 * Address description logic has been improved and is now common
1267 between Memcheck and Helgrind, resulting in better address
1268 descriptions for some kinds of error messages.
1270 * Error messages about dubious arguments (eg, to malloc or calloc) are
1271 output like other errors. This means that they can be suppressed
1272 and they have a stack trace.
1274 * The C++ demangler has been updated for better C++11 support.
1276 * New and modified GDB server monitor features:
1278 - Thread local variables/storage (__thread) can now be displayed.
1280 - The GDB server monitor command "v.info location <address>"
1281 displays information about an address. The information produced
1282 depends on the tool and on the options given to valgrind.
1283 Possibly, the following are described: global variables, local
1284 (stack) variables, allocated or freed blocks, ...
1286 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
1287 ask the GDB server to stop at the start of program execution, at
1288 the end of the program execution and on Valgrind internal errors.
1290 - A new monitor command "v.info stats" shows various Valgrind core
1291 and tool statistics.
1293 - A new monitor command "v.set hostvisibility" allows the GDB server
1294 to provide access to Valgrind internal host status/memory.
1296 * A new option "--aspace-minaddr=<address>" can in some situations
1297 allow the use of more memory by decreasing the address above which
1298 Valgrind maps memory. It can also be used to solve address
1299 conflicts with system libraries by increasing the default value.
1300 See user manual for details.
1302 * The amount of memory used by Valgrind to store debug info (unwind
1303 info, line number information and symbol data) has been
1304 significantly reduced, even though Valgrind now reads more
1305 information in order to support unwinding of inlined function calls.
1307 * Dwarf3 handling with --read-var-info=yes has been improved:
1309 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
1311 - Code compiled with
1312 -ffunction-sections -fdata-sections -Wl,--gc-sections
1313 no longer causes assertion failures.
1315 * Improved checking for the --sim-hints= and --kernel-variant=
1316 options. Unknown strings are now detected and reported to the user
1319 * The semantics of stack start/end boundaries in the valgrind.h
1320 VALGRIND_STACK_REGISTER client request has been clarified and
1321 documented. The convention is that start and end are respectively
1322 the lowest and highest addressable bytes of the stack.
1324 * ==================== FIXED BUGS ====================
1326 The following bugs have been fixed or resolved. Note that "n-i-bz"
1327 stands for "not in bugzilla" -- that is, a bug that was reported to us
1328 but never got a bugzilla entry. We encourage you to file bugs in
1329 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1330 than mailing the developers (or mailing lists) directly -- bugs that
1331 are not entered into bugzilla tend to get forgotten about or ignored.
1333 To see details of a given bug, visit
1334 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1335 where XXXXXX is the bug number as listed below.
1337 175819 Support for ipv6 socket reporting with --track-fds
1338 232510 make distcheck fails
1339 249435 Analyzing wine programs with callgrind triggers a crash
1340 278972 support for inlined function calls in stacktraces and suppression
1342 291310 FXSAVE instruction marks memory as undefined on amd64
1343 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
1344 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
1345 315199 vgcore file for threaded app does not show which thread crashed
1346 315952 tun/tap ioctls are not supported
1347 323178 Unhandled instruction: PLDW register (ARM)
1348 323179 Unhandled instruction: PLDW immediate (ARM)
1349 324050 Helgrind: SEGV because of unaligned stack when using movdqa
1350 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
1351 325124 [MIPSEL] Compilation error
1352 325477 Phase 4 support for IBM Power ISA 2.07
1353 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
1354 325628 Phase 5 support for IBM Power ISA 2.07
1355 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
1356 325751 Missing the two privileged Power PC Transactional Memory Instructions
1357 325816 Phase 6 support for IBM Power ISA 2.07
1358 325856 Make SGCheck fail gracefully on unsupported platforms
1359 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
1360 326436 DRD: False positive in libstdc++ std::list::push_back
1361 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
1362 326462 Refactor vgdb to isolate invoker stuff into separate module
1363 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
1364 326623 DRD: false positive conflict report in a field assignment
1365 326724 Valgrind does not compile on OSX 1.9 Mavericks
1366 326816 Intercept for __strncpy_sse2_unaligned missing?
1367 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
1368 326983 Clear direction flag after tests on amd64.
1369 327212 Do not prepend the current directory to absolute path names.
1370 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
1371 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
1372 327284 s390x: Fix translation of the risbg instruction
1373 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
1374 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
1375 327916 DW_TAG_typedef may have no name
1376 327943 s390x: add a redirection for the 'index' function
1377 328100 XABORT not implemented
1378 328205 Implement additional Xen hypercalls
1379 328454 add support Backtraces with ARM unwind tables (EXIDX)
1380 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
1381 328711 valgrind.1 manpage "memcheck options" section is badly generated
1382 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
1383 329612 Incorrect handling of AT_BASE for image execution
1384 329694 clang warns about using uninitialized variable
1385 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
1386 330228 mmap must align to VKI_SHMLBA on mips32
1387 330257 LLVM does not support `-mno-dynamic-no-pic` option
1388 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
1389 330459 --track-fds=yes doesn't track eventfds
1390 330469 Add clock_adjtime syscall support
1391 330594 Missing sysalls on PowerPC / uClibc
1392 330622 Add test to regression suite for POWER instruction: dcbzl
1393 330939 Support for AMD's syscall instruction on x86
1395 330941 Typo in PRE(poll) syscall wrapper
1396 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
1397 331254 Fix expected output for memcheck/tests/dw4
1398 331255 Fix race condition in test none/tests/coolo_sigaction
1399 331257 Fix type of jump buffer in test none/tests/faultstatus
1400 331305 configure uses bash specific syntax
1401 331337 s390x WARNING: unhandled syscall: 326 (dup3)
1402 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
1403 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
1404 331829 Unexpected ioctl opcode sign extension
1405 331830 ppc64: WARNING: unhandled syscall: 96/97
1406 331839 drd/tests/sem_open specifies invalid semaphore name
1407 331847 outcome of drd/tests/thread_name is nondeterministic
1408 332037 Valgrind cannot handle Thumb "add pc, reg"
1409 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
1410 consistency checks enabled
1411 332263 intercepts for pthread_rwlock_timedrdlock and
1412 pthread_rwlock_timedwrlock are incorrect
1413 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
1415 332276 Implement additional Xen hypercalls
1416 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
1417 332765 Fix ms_print to create temporary files in a proper directory
1418 333072 drd: Add semaphore annotations
1419 333145 Tests for missaligned PC+#imm access for arm
1420 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
1421 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
1422 333248 WARNING: unhandled syscall: unix:443
1423 333428 ldr.w pc [rD, #imm] instruction leads to assertion
1424 333501 cachegrind: assertion: Cache set count is not a power of two.
1427 333666 Recognize MPX instructions and bnd prefix.
1428 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
1429 333817 Valgrind reports the memory areas written to by the SG_IO
1431 334049 lzcnt fails silently (x86_32)
1432 334384 Valgrind does not have support Little Endian support for
1434 334585 recvmmsg unhandled (+patch) (arm)
1435 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
1436 334727 Build fails with -Werror=format-security
1437 334788 clarify doc about --log-file initial program directory
1438 334834 PPC64 Little Endian support, patch 2
1439 334836 PPC64 Little Endian support, patch 3 testcase fixes
1440 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
1441 335034 Unhandled ioctl: HCIGETDEVLIST
1442 335155 vgdb, fix error print statement.
1443 335262 arm64: movi 8bit version is not supported
1444 335263 arm64: dmb instruction is not implemented
1445 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
1446 335496 arm64: sbc/abc instructions are not implemented
1447 335554 arm64: unhandled instruction: abs
1448 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
1449 335735 arm64: unhandled instruction: cnt
1450 335736 arm64: unhandled instruction: uaddlv
1451 335848 arm64: unhandled instruction: {s,u}cvtf
1452 335902 arm64: unhandled instruction: sli
1453 335903 arm64: unhandled instruction: umull (vector)
1454 336055 arm64: unhandled instruction: mov (element)
1455 336062 arm64: unhandled instruction: shrn{,2}
1456 336139 mip64: [...] valgrind hangs and spins on a single core [...]
1457 336189 arm64: unhandled Instruction: mvn
1458 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
1459 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
1460 336772 Make moans about unknown ioctls more informative
1461 336957 Add a section about the Solaris/illumos port on the webpage
1462 337094 ifunc wrapper is broken on ppc64
1463 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
1464 337528 leak check heuristic for block prefixed by length as 64bit number
1465 337740 Implement additional Xen hypercalls
1466 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
1467 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
1468 337871 deprecate --db-attach
1469 338023 Add support for all V4L2/media ioctls
1470 338024 inlined functions are not shown if DW_AT_ranges is used
1471 338106 Add support for 'kcmp' syscall
1472 338115 DRD: computed conflict set differs from actual after fork
1473 338160 implement display of thread local storage in gdbsrv
1474 338205 configure.ac and check for -Wno-tautological-compare
1475 338300 coredumps are missing one byte of every segment
1476 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
1477 338499 --sim-hints parsing broken due to wrong order in tokens
1478 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
1479 338681 Unable to unwind through clone thread created on i386-linux
1480 338698 race condition between gdbsrv and vgdb on startup
1481 338703 helgrind on arm-linux gets false positives in dynamic loader
1482 338791 alt dwz files can be relative of debug/main file
1483 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
1484 338932 build V-trunk with gcc-trunk
1485 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
1486 345079 Fix build problems in VEX/useful/test_main.c
1487 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
1488 n-i-bz s390x: Fix memory corruption for multithreaded applications
1489 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
1490 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
1491 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
1492 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
1493 n-i-bz Add command-line option --default-suppressions
1494 n-i-bz Add support for BLKDISCARDZEROES ioctl
1495 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
1496 n-i-bz Add support for sys_pivot_root and sys_unshare
1498 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
1499 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
1500 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
1504 Release 3.9.0 (31 October 2013)
1505 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1506 3.9.0 is a feature release with many improvements and the usual
1507 collection of bug fixes.
1509 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
1510 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
1511 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
1512 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
1514 * ================== PLATFORM CHANGES =================
1516 * Support for MIPS64 LE and BE running Linux. Valgrind has been
1517 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
1519 * Support for MIPS DSP ASE on MIPS32 platforms.
1521 * Support for s390x Decimal Floating Point instructions on hosts that
1522 have the DFP facility installed.
1524 * Support for POWER8 (Power ISA 2.07) instructions
1526 * Support for Intel AVX2 instructions. This is available only on 64
1529 * Initial support for Intel Transactional Synchronization Extensions,
1532 * Initial support for Hardware Transactional Memory on POWER.
1534 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
1535 run large GUI apps tolerably well.
1537 * ==================== TOOL CHANGES ====================
1541 - Improvements in handling of vectorised code, leading to
1542 significantly fewer false error reports. You need to use the flag
1543 --partial-loads-ok=yes to get the benefits of these changes.
1545 - Better control over the leak checker. It is now possible to
1546 specify which leak kinds (definite/indirect/possible/reachable)
1547 should be displayed, which should be regarded as errors, and which
1548 should be suppressed by a given leak suppression. This is done
1549 using the options --show-leak-kinds=kind1,kind2,..,
1550 --errors-for-leak-kinds=kind1,kind2,.. and an optional
1551 "match-leak-kinds:" line in suppression entries, respectively.
1553 Note that generated leak suppressions contain this new line and
1554 are therefore more specific than in previous releases. To get the
1555 same behaviour as previous releases, remove the "match-leak-kinds:"
1556 line from generated suppressions before using them.
1558 - Reduced "possible leak" reports from the leak checker by the use
1559 of better heuristics. The available heuristics provide detection
1560 of valid interior pointers to std::stdstring, to new[] allocated
1561 arrays with elements having destructors and to interior pointers
1562 pointing to an inner part of a C++ object using multiple
1563 inheritance. They can be selected individually using the
1564 option --leak-check-heuristics=heur1,heur2,...
1566 - Better control of stacktrace acquisition for heap-allocated
1567 blocks. Using the --keep-stacktraces option, it is possible to
1568 control independently whether a stack trace is acquired for each
1569 allocation and deallocation. This can be used to create better
1570 "use after free" errors or to decrease Valgrind's resource
1571 consumption by recording less information.
1573 - Better reporting of leak suppression usage. The list of used
1574 suppressions (shown when the -v option is given) now shows, for
1575 each leak suppressions, how many blocks and bytes it suppressed
1576 during the last leak search.
1580 - False errors resulting from the use of statically initialised
1581 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
1584 - False errors resulting from the use of pthread_cond_waits that
1585 timeout, have been removed.
1587 * ==================== OTHER CHANGES ====================
1589 * Some attempt to tune Valgrind's space requirements to the expected
1590 capabilities of the target:
1592 - The default size of the translation cache has been reduced from 8
1593 sectors to 6 on Android platforms, since each sector occupies
1594 about 40MB when using Memcheck.
1596 - The default size of the translation cache has been increased to 16
1597 sectors on all other platforms, reflecting the fact that large
1598 applications require instrumentation and storage of huge amounts
1599 of code. For similar reasons, the number of memory mapped
1600 segments that can be tracked has been increased by a factor of 6.
1602 - In all cases, the maximum number of sectors in the translation
1603 cache can be controlled by the new flag --num-transtab-sectors.
1605 * Changes in how debug info (line numbers, etc) is read:
1607 - Valgrind no longer temporarily mmaps the entire object to read
1608 from it. Instead, reading is done through a small fixed sized
1609 buffer. This avoids virtual memory usage spikes when Valgrind
1610 reads debuginfo from large shared objects.
1612 - A new experimental remote debug info server. Valgrind can read
1613 debug info from a different machine (typically, a build host)
1614 where debuginfo objects are stored. This can save a lot of time
1615 and hassle when running Valgrind on resource-constrained targets
1616 (phones, tablets) when the full debuginfo objects are stored
1617 somewhere else. This is enabled by the --debuginfo-server=
1620 - Consistency checking between main and debug objects can be
1621 disabled using the --allow-mismatched-debuginfo option.
1623 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
1624 scanning can recover stack traces in some cases when the normal
1625 unwind mechanisms fail. Stack scanning is best described as "a
1626 nasty, dangerous and misleading hack" and so is disabled by default.
1627 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
1630 * Detection and merging of recursive stack frame cycles. When your
1631 program has recursive algorithms, this limits the memory used by
1632 Valgrind for recorded stack traces and avoids recording
1633 uninteresting repeated calls. This is controlled by the command
1634 line option --merge-recursive-frame and by the monitor command
1635 "v.set merge-recursive-frames".
1637 * File name and line numbers for used suppressions. The list of used
1638 suppressions (shown when the -v option is given) now shows, for each
1639 used suppression, the file name and line number where the suppression
1642 * New and modified GDB server monitor features:
1644 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
1645 that can be used to execute gdbserver monitor commands from the
1648 - A new monitor command, "v.info open_fds", that gives the list of
1649 open file descriptors and additional details.
1651 - An optional message in the "v.info n_errs_found" monitor command,
1652 for example "v.info n_errs_found test 1234 finished", allowing a
1653 comment string to be added to the process output, perhaps for the
1654 purpose of separating errors of different tests or test phases.
1656 - A new monitor command "v.info execontext" that shows information
1657 about the stack traces recorded by Valgrind.
1659 - A new monitor command "v.do expensive_sanity_check_general" to run
1660 some internal consistency checks.
1662 * New flag --sigill-diagnostics to control whether a diagnostic
1663 message is printed when the JIT encounters an instruction it can't
1664 translate. The actual behavior -- delivery of SIGILL to the
1665 application -- is unchanged.
1667 * The maximum amount of memory that Valgrind can use on 64 bit targets
1668 has been increased from 32GB to 64GB. This should make it possible
1669 to run applications on Memcheck that natively require up to about 35GB.
1671 * ==================== FIXED BUGS ====================
1673 The following bugs have been fixed or resolved. Note that "n-i-bz"
1674 stands for "not in bugzilla" -- that is, a bug that was reported to us
1675 but never got a bugzilla entry. We encourage you to file bugs in
1676 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1677 than mailing the developers (or mailing lists) directly -- bugs that
1678 are not entered into bugzilla tend to get forgotten about or ignored.
1680 To see details of a given bug, visit
1681 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1682 where XXXXXX is the bug number as listed below.
1684 123837 system call: 4th argument is optional, depending on cmd
1685 135425 memcheck should tell you where Freed blocks were Mallocd
1686 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
1687 207815 Adds some of the drm ioctls to syswrap-linux.c
1688 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
1689 252955 Impossible to compile with ccache
1690 253519 Memcheck reports auxv pointer accesses as invalid reads.
1691 263034 Crash when loading some PPC64 binaries
1692 269599 Increase deepest backtrace
1693 274695 s390x: Support "compare to/from logical" instructions (z196)
1694 275800 s390x: Autodetect cache info (part 2)
1695 280271 Valgrind reports possible memory leaks on still-reachable std::string
1696 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
1697 289578 Backtraces with ARM unwind tables (stack scan flags)
1698 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
1699 304832 ppc32: build failure
1700 305431 Use find_buildid shdr fallback for separate .debug files
1701 305728 Add support for AVX2 instructions
1702 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
1703 306035 s390x: Fix IR generation for LAAG and friends
1704 306054 s390x: Condition code computation for convert-to-int/logical
1705 306098 s390x: alternate opcode form for convert to/from fixed
1706 306587 Fix cache line detection from auxiliary vector for PPC.
1707 306783 Mips unhandled syscall : 4025 / 4079 / 4182
1708 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
1709 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
1710 307101 sys_capget second argument can be NULL
1711 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
1712 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
1713 307113 s390x: DFP support
1714 307141 valgrind does't work in mips-linux system
1715 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
1716 307285 x86_amd64 feature test for avx in test suite is wrong
1717 307290 memcheck overlap testcase needs memcpy version filter
1718 307463 Please add "&limit=0" to the "all open bugs" link
1719 307465 --show-possibly-lost=no should reduce the error count / exit code
1720 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
1721 307729 pkgconfig support broken valgrind.pc
1722 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
1723 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
1724 308089 Unhandled syscall on ppc64: prctl
1725 308135 PPC32 MPC8xx has 16 bytes cache size
1726 308321 testsuite memcheck filter interferes with gdb_filter
1728 308341 vgdb should report process exit (or fatal signal)
1729 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
1730 308495 Remove build dependency on installed Xen headers
1731 308573 Internal error on 64-bit instruction executed in 32-bit mode
1733 308627 pmovmskb validity bit propagation is imprecise
1734 308644 vgdb command for having the info for the track-fds option
1735 308711 give more info about aspacemgr and arenas in out_of_memory
1736 308717 ARM: implement fixed-point VCVT.F64.[SU]32
1737 308718 ARM implement SMLALBB family of instructions
1738 308886 Missing support for PTRACE_SET/GETREGSET
1739 308930 syscall name_to_handle_at (303 on amd64) not handled
1740 309229 V-bit tester does not report number of tests generated
1741 309323 print unrecognized instuction on MIPS
1742 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
1743 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
1744 309430 Self hosting ppc64 encounters a vassert error on operand type
1745 309600 valgrind is a bit confused about 0-sized sections
1746 309823 Generate errors for still reachable blocks
1747 309921 PCMPISTRI validity bit propagation is imprecise
1748 309922 none/tests/ppc64/test_dfp5 sometimes fails
1749 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
1750 310424 --read-var-info does not properly describe static variables
1751 310792 search additional path for debug symbols
1752 310931 s390x: Message-security assist (MSA) instruction extension [..]
1753 311100 PPC DFP implementation of the integer operands is inconsistent [..]
1754 311318 ARM: "128-bit constant is not implemented" error message
1755 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
1756 311690 V crashes because it redirects branches inside of a redirected function
1757 311880 x86_64: make regtest hangs at shell_valid1
1758 311922 WARNING: unhandled syscall: 170
1760 312171 ppc: insn selection for DFP
1761 312571 Rounding mode call wrong for the DFP Iops [..]
1762 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
1763 312913 Dangling pointers error should also report the alloc stack trace
1764 312980 Building on Mountain Lion generates some compiler warnings
1765 313267 Adding MIPS64/Linux port to Valgrind
1768 313811 Buffer overflow in assert_fail
1769 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
1770 314269 ppc: dead code in insn selection
1771 314718 ARM: implement integer divide instruction (sdiv and udiv)
1772 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
1773 315441 sendmsg syscall should ignore unset msghdr msg_flags
1774 315534 msgrcv inside a thread causes valgrind to hang (block)
1775 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
1776 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
1777 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
1778 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
1779 316144 valgrind.1 manpage contains unknown ??? strings [..]
1780 316145 callgrind command line options in manpage reference (unknown) [..]
1781 316145 callgrind command line options in manpage reference [..]
1782 316181 drd: Fixed a 4x slowdown for certain applications
1783 316503 Valgrind does not support SSE4 "movntdqa" instruction
1784 316535 Use of |signed int| instead of |size_t| in valgrind messages
1785 316696 fluidanimate program of parsec 2.1 stuck
1786 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
1787 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
1788 317186 "Impossible happens" when occurs VCVT instruction on ARM
1789 317318 Support for Threading Building Blocks "scalable_malloc"
1790 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
1791 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
1792 317463 bmi testcase IR SANITY CHECK FAILURE
1793 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
1794 318050 libmpiwrap fails to compile with out-of-source build
1795 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
1796 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
1797 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
1798 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
1799 318932 Add missing PPC64 and PPC32 system call support
1800 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
1801 319395 Crash with unhandled instruction on STRT (Thumb) instructions
1802 319494 VEX Makefile-gcc standalone build update after r2702
1803 319505 [MIPSEL] Crash: unhandled UNRAY operator.
1804 319858 disInstr(thumb): unhandled instruction on instruction STRBT
1805 319932 disInstr(thumb): unhandled instruction on instruction STRHT
1806 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
1807 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
1808 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
1809 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
1810 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
1811 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
1812 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
1813 320895 add fanotify support (patch included)
1814 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
1815 321065 Valgrind updates for Xen 4.3
1816 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
1817 321363 Unhandled instruction: SSAX (ARM + Thumb)
1818 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
1819 321466 Unhandled instruction: SHASX (ARM + Thumb)
1820 321467 Unhandled instruction: SHSAX (ARM + Thumb)
1821 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
1822 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
1823 321620 Unhandled instruction: UASX (ARM + Thumb)
1824 321621 Unhandled instruction: USAX (ARM + Thumb)
1825 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
1826 321693 Unhandled instruction: LDRSBT (Thumb)
1827 321694 Unhandled instruction: UQASX (ARM + Thumb)
1828 321696 Unhandled instruction: UQSAX (Thumb + ARM)
1829 321697 Unhandled instruction: UHASX (ARM + Thumb)
1830 321703 Unhandled instruction: UHSAX (ARM + Thumb)
1831 321704 Unhandled instruction: REVSH (ARM + Thumb)
1832 321730 Add cg_diff and cg_merge man pages
1833 321738 Add vgdb and valgrind-listener man pages
1835 321891 Unhandled instruction: LDRHT (Thumb)
1836 321960 pthread_create() then alloca() causing invalid stack write errors
1837 321969 ppc32 and ppc64 don't support [lf]setxattr
1838 322254 Show threadname together with tid if set by application
1839 322294 Add initial support for IBM Power ISA 2.07
1840 322368 Assertion failure in wqthread_hijack under OS X 10.8
1841 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
1842 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
1843 322851 0bXXX binary literal syntax is not standard
1844 323035 Unhandled instruction: LDRSHT(Thumb)
1845 323036 Unhandled instruction: SMMLS (ARM and Thumb)
1846 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
1847 323175 Unhandled instruction: SMLALD (ARM + Thumb)
1848 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
1849 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
1850 323437 Phase 2 support for IBM Power ISA 2.07
1851 323713 Support mmxext (integer sse) subset on i386 (athlon)
1852 323803 Transactional memory instructions are not supported for Power
1853 323893 SSE3 not available on amd cpus in valgrind
1854 323905 Probable false positive from Valgrind/drd on close()
1855 323912 valgrind.h header isn't compatible for mingw64
1856 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
1857 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
1858 324181 mmap does not handle MAP_32BIT
1859 324227 memcheck false positive leak when a thread calls exit+block [..]
1860 324421 Support for fanotify API on ARM architecture
1861 324514 gdbserver monitor cmd output behaviour consistency [..]
1862 324518 ppc64: Emulation of dcbt instructions does not handle [..]
1863 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
1864 324582 When access is made to freed memory, report both allocation [..]
1865 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
1866 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
1867 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
1868 324834 Unhandled instructions in Microsoft C run-time for x86_64
1869 324894 Phase 3 support for IBM Power ISA 2.07
1870 326091 drd: Avoid false race reports from optimized strlen() impls
1871 326113 valgrind libvex hwcaps error on AMD64
1872 n-i-bz Some wrong command line options could be ignored
1873 n-i-bz patch to allow fair-sched on android
1874 n-i-bz report error for vgdb snapshot requested before execution
1875 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
1877 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
1881 Release 3.8.1 (19 September 2012)
1882 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1883 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
1884 that occur moderately frequently in real use cases, adds support for
1885 some missing instructions on ARM, and fixes a deadlock condition on
1886 MacOSX. If you package or deliver 3.8.0 for others to use, you might
1887 want to consider upgrading to 3.8.1 instead.
1889 The following bugs have been fixed or resolved. Note that "n-i-bz"
1890 stands for "not in bugzilla" -- that is, a bug that was reported to us
1891 but never got a bugzilla entry. We encourage you to file bugs in
1892 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1893 than mailing the developers (or mailing lists) directly -- bugs that
1894 are not entered into bugzilla tend to get forgotten about or ignored.
1896 To see details of a given bug, visit
1897 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1898 where XXXXXX is the bug number as listed below.
1901 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
1902 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
1903 298281 wcslen causes false(?) uninitialised value warnings
1904 301281 valgrind hangs on OS X when the process calls system()
1905 304035 disInstr(arm): unhandled instruction 0xE1023053
1906 304867 implement MOVBE instruction in x86 mode
1907 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
1908 305042 amd64: implement 0F 7F encoding of movq between two registers
1909 305199 ARM: implement QDADD and QDSUB
1910 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
1911 305513 killed by fatal signal: SIGSEGV
1912 305690 DRD reporting invalid semaphore when sem_trywait fails
1913 305926 Invalid alignment checks for some AVX instructions
1914 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
1915 306310 3.8.0 release tarball missing some files
1916 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
1917 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
1918 n-i-bz shmat of a segment > 4Gb does not work
1919 n-i-bz simulate_control_c script wrong USR1 signal number on mips
1920 n-i-bz vgdb ptrace calls wrong on mips [...]
1921 n-i-bz Fixes for more MPI false positives
1922 n-i-bz exp-sgcheck's memcpy causes programs to segfault
1923 n-i-bz OSX build w/ clang: asserts at startup
1924 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
1925 n-i-bz fix a couple of union tag-vs-field mixups
1926 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
1928 The following bugs were fixed in 3.8.0 but not listed in this NEWS
1931 254088 Valgrind should know about UD2 instruction
1934 304754 NEWS blows TeX's little mind
1936 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
1940 Release 3.8.0 (10 August 2012)
1941 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1942 3.8.0 is a feature release with many improvements and the usual
1943 collection of bug fixes.
1945 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
1946 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
1947 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
1948 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
1949 There is initial support for MacOSX 10.8, but it is not usable for
1950 serious work at present.
1952 * ================== PLATFORM CHANGES =================
1954 * Support for MIPS32 platforms running Linux. Valgrind has been
1955 tested on MIPS32 and MIPS32r2 platforms running different Debian
1956 Squeeze and MeeGo distributions. Both little-endian and big-endian
1957 cores are supported. The tools Memcheck, Massif and Lackey have
1958 been tested and are known to work. See README.mips for more details.
1960 * Preliminary support for Android running on x86.
1962 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
1964 * Support for Intel AVX instructions and for AES instructions. This
1965 support is available only for 64 bit code.
1967 * Support for POWER Decimal Floating Point instructions.
1969 * ==================== TOOL CHANGES ====================
1971 * Non-libc malloc implementations are now supported. This is useful
1972 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
1973 Using the new option --soname-synonyms, such tools can be informed
1974 that the malloc implementation is either linked statically into the
1975 executable, or is present in some other shared library different
1976 from libc.so. This makes it possible to process statically linked
1977 programs, and programs using other malloc libraries, for example
1978 TCMalloc or JEMalloc.
1980 * For tools that provide their own replacement for malloc et al, the
1981 option --redzone-size=<number> allows users to specify the size of
1982 the padding blocks (redzones) added before and after each client
1983 allocated block. Smaller redzones decrease the memory needed by
1984 Valgrind. Bigger redzones increase the chance to detect blocks
1985 overrun or underrun. Prior to this change, the redzone size was
1986 hardwired to 16 bytes in Memcheck.
1990 - The leak_check GDB server monitor command now can
1991 control the maximum nr of loss records to output.
1993 - Reduction of memory use for applications allocating
1994 many blocks and/or having many partially defined bytes.
1996 - Addition of GDB server monitor command 'block_list' that lists
1997 the addresses/sizes of the blocks of a leak search loss record.
1999 - Addition of GDB server monitor command 'who_points_at' that lists
2000 the locations pointing at a block.
2002 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2003 detect an invalid access of these redzones, by marking them
2004 noaccess. Similarly, if a redzone size is given for a memory
2005 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2006 This still allows to find some bugs if the user has forgotten to
2007 mark the pool superblock noaccess.
2009 - Performance of memory leak check has been improved, especially in
2010 cases where there are many leaked blocks and/or many suppression
2011 rules used to suppress leak reports.
2013 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2014 more precise analysis, which is important for LLVM/Clang
2015 generated code. This is at the cost of somewhat reduced
2016 performance. Note there is no change to analysis precision or
2017 costs on Linux targets.
2021 - Added even more facilities that can help finding the cause of a data
2022 race, namely the command-line option --ptrace-addr and the macro
2023 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2025 - Fixed a subtle bug that could cause false positive data race reports.
2027 * ==================== OTHER CHANGES ====================
2029 * The C++ demangler has been updated so as to work well with C++
2030 compiled by up to at least g++ 4.6.
2032 * Tool developers can make replacement/wrapping more flexible thanks
2033 to the new option --soname-synonyms. This was reported above, but
2034 in fact is very general and applies to all function
2035 replacement/wrapping, not just to malloc-family functions.
2037 * Round-robin scheduling of threads can be selected, using the new
2038 option --fair-sched= yes. Prior to this change, the pipe-based
2039 thread serialisation mechanism (which is still the default) could
2040 give very unfair scheduling. --fair-sched=yes improves
2041 responsiveness of interactive multithreaded applications, and
2042 improves repeatability of results from the thread checkers Helgrind
2045 * For tool developers: support to run Valgrind on Valgrind has been
2046 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2048 * gdbserver now shows the float shadow registers as integer
2049 rather than float values, as the shadow values are mostly
2050 used as bit patterns.
2052 * Increased limit for the --num-callers command line flag to 500.
2054 * Performance improvements for error matching when there are many
2055 suppression records in use.
2057 * Improved support for DWARF4 debugging information (bug 284184).
2059 * Initial support for DWZ compressed Dwarf debug info.
2061 * Improved control over the IR optimiser's handling of the tradeoff
2062 between performance and precision of exceptions. Specifically,
2063 --vex-iropt-precise-memory-exns has been removed and replaced by
2064 --vex-iropt-register-updates, with extended functionality. This
2065 allows the Valgrind gdbserver to always show up to date register
2068 * Modest performance gains through the use of translation chaining for
2071 * ==================== FIXED BUGS ====================
2073 The following bugs have been fixed or resolved. Note that "n-i-bz"
2074 stands for "not in bugzilla" -- that is, a bug that was reported to us
2075 but never got a bugzilla entry. We encourage you to file bugs in
2076 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2077 than mailing the developers (or mailing lists) directly -- bugs that
2078 are not entered into bugzilla tend to get forgotten about or ignored.
2080 To see details of a given bug, visit
2081 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2082 where XXXXXX is the bug number as listed below.
2084 197914 Building valgrind from svn now requires automake-1.10
2085 203877 increase to 16Mb maximum allowed alignment for memalign et al
2086 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
2087 247386 make perf does not run all performance tests
2088 270006 Valgrind scheduler unfair
2089 270777 Adding MIPS/Linux port to Valgrind
2090 270796 s390x: Removed broken support for the TS insn
2091 271438 Fix configure for proper SSE4.2 detection
2092 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
2093 273475 Add support for AVX instructions
2094 274078 improved configure logic for mpicc
2095 276993 fix mremap 'no thrash checks'
2096 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
2097 281482 memcheck incorrect byte allocation count in realloc() for silly argument
2098 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
2099 283413 Fix wrong sanity check
2100 283671 Robustize alignment computation in LibVEX_Alloc
2101 283961 Adding support for some HCI IOCTLs
2102 284124 parse_type_DIE: confused by: DWARF 4
2103 284864 == 273475 (Add support for AVX instructions)
2104 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
2105 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
2106 285725 == 273475 (Add support for AVX instructions)
2107 286261 add wrapper for linux I2C_RDWR ioctl
2108 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
2109 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
2110 286384 configure fails "checking for a supported version of gcc"
2111 286497 == 273475 (Add support for AVX instructions)
2112 286596 == 273475 (Add support for AVX instructions)
2113 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
2114 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
2115 287260 Incorrect conditional jump or move depends on uninitialised value(s)
2116 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
2117 287307 == 273475 (Add support for AVX instructions)
2118 287858 VG_(strerror): unknown error
2119 288298 (MacOSX) unhandled syscall shm_unlink
2120 288995 == 273475 (Add support for AVX instructions)
2121 289470 Loading of large Mach-O thin binaries fails.
2122 289656 == 273475 (Add support for AVX instructions)
2123 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
2124 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
2125 289839 s390x: Provide support for unicode conversion instructions
2126 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
2127 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
2128 290655 Add support for AESKEYGENASSIST instruction
2129 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
2130 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
2131 291253 ES register not initialised in valgrind simulation
2132 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
2133 291865 s390x: Support the "Compare Double and Swap" family of instructions
2134 292300 == 273475 (Add support for AVX instructions)
2135 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
2136 292493 == 273475 (Add support for AVX instructions)
2137 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
2138 292627 Missing support for some SCSI ioctls
2139 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
2140 292841 == 273475 (Add support for AVX instructions)
2141 292993 implement the getcpu syscall on amd64-linux
2142 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
2143 293088 Add some VEX sanity checks for ppc64 unhandled instructions
2144 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
2145 293754 PCMPxSTRx not implemented for 16-bit characters
2146 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
2147 293808 CLFLUSH not supported by latest VEX for amd64
2148 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
2149 294048 MPSADBW instruction not implemented
2150 294055 regtest none/tests/shell fails when locale is not set to C
2151 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
2152 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
2153 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
2154 294260 disInstr_AMD64: disInstr miscalculated next %rip
2155 294523 --partial-loads-ok=yes causes false negatives
2156 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
2157 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
2158 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
2159 295089 can not annotate source for both helgrind and drd
2160 295221 POWER Processor decimal floating point instruction support missing
2161 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
2162 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
2163 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
2164 295617 ARM - Add some missing syscalls
2165 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
2166 296229 Linux user input device ioctls missing wrappers
2167 296318 ELF Debug info improvements (more than one rx/rw mapping)
2168 296422 Add translation chaining support
2169 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
2170 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
2171 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
2172 297078 gdbserver signal handling problems [..]
2173 297147 drd false positives on newly allocated memory
2174 297329 disallow decoding of IBM Power DFP insns on some machines
2175 297497 POWER Processor decimal floating point instruction support missing
2176 297701 Another alias for strncasecmp_l in libc-2.13.so
2177 297911 'invalid write' not reported when using APIs for custom mem allocators.
2178 297976 s390x: revisit EX implementation
2179 297991 Valgrind interferes with mmap()+ftell()
2180 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
2181 297993 Fix compilation of valgrind with gcc -g3.
2182 298080 POWER Processor DFP support missing, part 3
2183 298227 == 273475 (Add support for AVX instructions)
2184 298335 == 273475 (Add support for AVX instructions)
2185 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
2186 298394 s390x: Don't bail out on an unknown machine model. [..]
2187 298421 accept4() syscall (366) support is missing for ARM
2188 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
2189 298732 valgrind installation problem in ubuntu with kernel version 3.x
2190 298862 POWER Processor DFP instruction support missing, part 4
2191 298864 DWARF reader mis-parses DW_FORM_ref_addr
2192 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
2193 299053 Support DWARF4 DW_AT_high_pc constant form
2194 299104 == 273475 (Add support for AVX instructions)
2195 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
2196 299629 dup3() syscall (358) support is missing for ARM
2197 299694 POWER Processor DFP instruction support missing, part 5
2198 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
2199 299803 == 273475 (Add support for AVX instructions)
2200 299804 == 273475 (Add support for AVX instructions)
2201 299805 == 273475 (Add support for AVX instructions)
2202 300140 ARM - Missing (T1) SMMUL
2203 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
2204 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
2205 300414 FCOM and FCOMP unimplemented for amd64 guest
2206 301204 infinite loop in canonicaliseSymtab with ifunc symbol
2207 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
2208 301265 add x86 support to Android build
2209 301984 configure script doesn't detect certain versions of clang
2210 302205 Fix compiler warnings for POWER VEX code and POWER test cases
2211 302287 Unhandled movbe instruction on Atom processors
2212 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
2213 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
2214 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
2215 302656 == 273475 (Add support for AVX instructions)
2216 302709 valgrind for ARM needs extra tls support for android emulator [..]
2217 302827 add wrapper for CDROM_GET_CAPABILITY
2218 302901 Valgrind crashes with dwz optimized debuginfo
2219 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
2220 303116 Add support for the POWER instruction popcntb
2221 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
2222 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
2223 303466 == 273475 (Add support for AVX instructions)
2224 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
2225 303963 strstr() function produces wrong results under valgrind callgrind
2226 304054 CALL_FN_xx macros need to enforce stack alignment
2227 304561 tee system call not supported
2228 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
2229 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
2230 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
2231 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
2232 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
2233 n-i-bz s390x: Shadow registers can now be examined using vgdb
2235 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
2236 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
2240 Release 3.7.0 (5 November 2011)
2241 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2242 3.7.0 is a feature release with many significant improvements and the
2243 usual collection of bug fixes.
2245 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2246 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
2247 Support for recent distros and toolchain components (glibc 2.14, gcc
2248 4.6, MacOSX 10.7) has been added.
2250 * ================== PLATFORM CHANGES =================
2252 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
2253 analyse 64-bit programs running on z/Architecture. Most user space
2254 instructions up to and including z10 are supported. Valgrind has
2255 been tested extensively on z9, z10, and z196 machines running SLES
2256 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
2257 known to work well. Callgrind, Helgrind, and DRD work reasonably
2258 well on z9 and later models. See README.s390 for more details.
2260 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
2261 64-bit processes are supported. Some complex threaded applications
2262 (Firefox) are observed to hang when run as 32 bit applications,
2263 whereas 64-bit versions run OK. The cause is unknown. Memcheck
2264 will likely report some false errors. In general, expect some rough
2265 spots. This release also supports MacOSX 10.6, but drops support
2268 * Preliminary support for Android (on ARM). Valgrind can now run
2269 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
2270 README.android for more details, plus instructions on how to get
2273 * Support for the IBM Power ISA 2.06 (Power7 instructions)
2275 * General correctness and performance improvements for ARM/Linux, and,
2276 by extension, ARM/Android.
2278 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
2279 instruction set support is under development but is not available in
2282 * Support for AIX5 has been removed.
2284 * ==================== TOOL CHANGES ====================
2286 * Memcheck: some incremental changes:
2288 - reduction of memory use in some circumstances
2290 - improved handling of freed memory, which in some circumstances
2291 can cause detection of use-after-free that would previously have
2294 - fix of a longstanding bug that could cause false negatives (missed
2295 errors) in programs doing vector saturated narrowing instructions.
2297 * Helgrind: performance improvements and major memory use reductions,
2298 particularly for large, long running applications which perform many
2299 synchronisation (lock, unlock, etc) events. Plus many smaller
2302 - display of locksets for both threads involved in a race
2304 - general improvements in formatting/clarity of error messages
2306 - addition of facilities and documentation regarding annotation
2307 of thread safe reference counted C++ classes
2309 - new flag --check-stack-refs=no|yes [yes], to disable race checking
2310 on thread stacks (a performance hack)
2312 - new flag --free-is-write=no|yes [no], to enable detection of races
2313 where one thread accesses heap memory but another one frees it,
2314 without any coordinating synchronisation event
2316 * DRD: enabled XML output; added support for delayed thread deletion
2317 in order to detect races that occur close to the end of a thread
2318 (--join-list-vol); fixed a memory leak triggered by repeated client
2319 memory allocatation and deallocation; improved Darwin support.
2321 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
2323 * exp-sgcheck: this tool has been reduced in scope so as to improve
2324 performance and remove checking that Memcheck does better.
2325 Specifically, the ability to check for overruns for stack and global
2326 arrays is unchanged, but the ability to check for overruns of heap
2327 blocks has been removed. The tool has accordingly been renamed to
2328 exp-sgcheck ("Stack and Global Array Checking").
2330 * ==================== OTHER CHANGES ====================
2332 * GDB server: Valgrind now has an embedded GDB server. That means it
2333 is possible to control a Valgrind run from GDB, doing all the usual
2334 things that GDB can do (single stepping, breakpoints, examining
2335 data, etc). Tool-specific functionality is also available. For
2336 example, it is possible to query the definedness state of variables
2337 or memory from within GDB when running Memcheck; arbitrarily large
2338 memory watchpoints are supported, etc. To use the GDB server, start
2339 Valgrind with the flag --vgdb-error=0 and follow the on-screen
2342 * Improved support for unfriendly self-modifying code: a new option
2343 --smc-check=all-non-file is available. This adds the relevant
2344 consistency checks only to code that originates in non-file-backed
2345 mappings. In effect this confines the consistency checking only to
2346 code that is or might be JIT generated, and avoids checks on code
2347 that must have been compiled ahead of time. This significantly
2348 improves performance on applications that generate code at run time.
2350 * It is now possible to build a working Valgrind using Clang-2.9 on
2353 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
2354 These enable and disable error reporting on a per-thread, and
2355 nestable, basis. This is useful for hiding errors in particularly
2356 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
2357 now uses this facility.
2359 * Added the --mod-funcname option to cg_diff.
2361 * ==================== FIXED BUGS ====================
2363 The following bugs have been fixed or resolved. Note that "n-i-bz"
2364 stands for "not in bugzilla" -- that is, a bug that was reported to us
2365 but never got a bugzilla entry. We encourage you to file bugs in
2366 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2367 mailing the developers (or mailing lists) directly -- bugs that are
2368 not entered into bugzilla tend to get forgotten about or ignored.
2370 To see details of a given bug, visit
2371 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2372 where XXXXXX is the bug number as listed below.
2374 79311 malloc silly arg warning does not give stack trace
2375 210935 port valgrind.h (not valgrind) to win32 to support client requests
2376 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
2377 243404 Port to zSeries
2378 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
2379 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
2380 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
2381 253206 Some fixes for the faultstatus testcase
2382 255223 capget testcase fails when running as root
2383 256703 xlc_dbl_u32.c testcase broken
2384 256726 Helgrind tests have broken inline asm
2385 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
2386 264800 testcase compile failure on zseries
2387 265762 make public VEX headers compilable by G++ 3.x
2388 265771 assertion in jumps.c (r11523) fails with glibc-2.3
2389 266753 configure script does not give the user the option to not use QtCore
2390 266931 gen_insn_test.pl is broken
2391 266961 ld-linux.so.2 i?86-linux strlen issues
2392 266990 setns instruction causes false positive
2393 267020 Make directory for temporary files configurable at run-time.
2394 267342 == 267997 (segmentation fault on Mac OS 10.6)
2395 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
2396 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
2397 267488 regtest: darwin support for 64-bit build
2398 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
2399 267630 Add support for IBM Power ISA 2.06 -- stage 1
2400 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
2401 267819 Add client request for informing the core about reallocation
2402 267925 laog data structure quadratic for a single sequence of lock
2403 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
2404 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
2405 268513 missed optimizations in fold_Expr
2406 268619 s390x: fpr - gpr transfer facility
2407 268620 s390x: reconsider "long displacement" requirement
2408 268621 s390x: improve IR generation for XC
2409 268715 s390x: FLOGR is not universally available
2410 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
2411 268930 s390x: MHY is not universally available
2412 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
2413 269079 Support ptrace system call on ARM
2414 269144 missing "Bad option" error message
2415 269209 conditional load and store facility (z196)
2416 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
2417 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
2418 269736 s390x: minor code generation tweaks
2419 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
2420 269863 s390x: remove unused function parameters
2421 269864 s390x: tweak s390_emit_load_cc
2422 269884 == 250101 (overhead for huge blocks exhausts space too soon)
2423 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
2424 270115 s390x: rewrite some testcases
2425 270309 == 267997 (valgrind crash on startup)
2426 270320 add support for Linux FIOQSIZE ioctl() call
2427 270326 segfault while trying to sanitize the environment passed to execle
2428 270794 IBM POWER7 support patch causes regression in none/tests
2429 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
2430 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
2431 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
2432 270959 s390x: invalid use of R0 as base register
2433 271042 VSX configure check fails when it should not
2434 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
2435 271259 s390x: fix code confusion
2436 271337 == 267997 (Valgrind segfaults on MacOS X)
2437 271385 s390x: Implement Ist_MBE
2438 271501 s390x: misc cleanups
2439 271504 s390x: promote likely and unlikely
2440 271579 ppc: using wrong enum type
2441 271615 unhandled instruction "popcnt" (arch=amd10h)
2442 271730 Fix bug when checking ioctls: duplicate check
2443 271776 s390x: provide STFLE instruction support
2444 271779 s390x: provide clock instructions like STCK
2445 271799 Darwin: ioctls without an arg report a memory error
2446 271820 arm: fix type confusion
2447 271917 pthread_cond_timedwait failure leads to not-locked false positive
2448 272067 s390x: fix DISP20 macro
2449 272615 A typo in debug output in mc_leakcheck.c
2450 272661 callgrind_annotate chokes when run from paths containing regex chars
2451 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
2452 272955 Unhandled syscall error for pwrite64 on ppc64 arch
2453 272967 make documentation build-system more robust
2454 272986 Fix gcc-4.6 warnings with valgrind.h
2455 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
2456 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
2457 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
2458 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
2459 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
2460 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
2461 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
2462 273778 exp-ptrcheck: unhandled sysno == 259
2463 274089 exp-ptrcheck: unhandled sysno == 208
2464 274378 s390x: Various dispatcher tweaks
2465 274447 WARNING: unhandled syscall: 340
2466 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
2467 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
2468 274926 valgrind does not build against linux-3
2469 275148 configure FAIL with glibc-2.14
2470 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
2471 275168 Make Valgrind work for MacOSX 10.7 Lion
2472 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
2473 275278 valgrind does not build on Linux kernel 3.0.* due to silly
2474 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
2475 275308 Fix implementation for ppc64 fres instruc
2476 275339 s390x: fix testcase compile warnings
2477 275517 s390x: Provide support for CKSM instruction
2478 275710 s390x: get rid of redundant address mode calculation
2479 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
2480 275852 == 250101 (valgrind uses all swap space and is killed)
2481 276784 Add support for IBM Power ISA 2.06 -- stage 3
2482 276987 gdbsrv: fix tests following recent commits
2483 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
2484 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
2485 277471 Unhandled syscall: 340
2486 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
2487 277653 ARM: support Thumb2 PLD instruction
2488 277663 ARM: NEON float VMUL by scalar incorrect
2489 277689 ARM: tests for VSTn with register post-index are broken
2490 277694 ARM: BLX LR instruction broken in ARM mode
2491 277780 ARM: VMOV.F32 (immediate) instruction is broken
2492 278057 fuse filesystem syscall deadlocks
2493 278078 Unimplemented syscall 280 on ppc32
2494 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
2495 278454 VALGRIND_STACK_DEREGISTER has wrong output type
2496 278502 == 275284 (Valgrind confuses memcpy() and memmove())
2497 278892 gdbsrv: factorize gdb version handling, fix doc and typos
2498 279027 Support for MVCL and CLCL instruction
2499 279027 s390x: Provide support for CLCL and MVCL instructions
2500 279062 Remove a redundant check in the insn selector for ppc.
2501 279071 JDK creates PTEST with redundant REX.W prefix
2502 279212 gdbsrv: add monitor cmd v.info scheduler.
2503 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
2504 279698 memcheck discards valid-bits for packuswb
2505 279795 memcheck reports uninitialised values for mincore on amd64
2506 279994 Add support for IBM Power ISA 2.06 -- stage 3
2507 280083 mempolicy syscall check errors
2508 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
2509 280710 s390x: config files for nightly builds
2510 280757 /tmp dir still used by valgrind even if TMPDIR is specified
2511 280965 Valgrind breaks fcntl locks when program does mmap
2512 281138 WARNING: unhandled syscall: 340
2513 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
2514 281304 == 275168 (Darwin: dyld "cannot load inserted library")
2515 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
2516 281468 s390x: handle do_clone and gcc clones in call traces
2517 281488 ARM: VFP register corruption
2518 281828 == 275284 (false memmove warning: "Source and destination overlap")
2519 281883 s390x: Fix system call wrapper for "clone".
2520 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
2521 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
2522 282238 SLES10: make check fails
2523 282979 strcasestr needs replacement with recent(>=2.12) glibc
2524 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
2525 283243 Regression in ppc64 memcheck tests
2526 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
2527 283427 re-connect epoll_pwait syscall on ARM linux
2528 283600 gdbsrv: android: port vgdb.c
2529 283709 none/tests/faultstatus needs to account for page size
2530 284305 filter_gdb needs enhancement to work on ppc64
2531 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
2532 284472 Thumb2 ROR.W encoding T2 not implemented
2533 284621 XML-escape process command line in XML output
2534 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
2535 that have non-power-of-2 sizes (also AMDs)
2536 n-i-bz don't be spooked by libraries mashed by elfhack
2537 n-i-bz don't be spooked by libxul.so linked with gold
2538 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
2540 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
2541 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
2542 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
2546 Release 3.6.1 (16 February 2011)
2547 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2548 3.6.1 is a bug fix release. It adds support for some SSE4
2549 instructions that were omitted in 3.6.0 due to lack of time. Initial
2550 support for glibc-2.13 has been added. A number of bugs causing
2551 crashing or assertion failures have been fixed.
2553 The following bugs have been fixed or resolved. Note that "n-i-bz"
2554 stands for "not in bugzilla" -- that is, a bug that was reported to us
2555 but never got a bugzilla entry. We encourage you to file bugs in
2556 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2557 mailing the developers (or mailing lists) directly -- bugs that are
2558 not entered into bugzilla tend to get forgotten about or ignored.
2560 To see details of a given bug, visit
2561 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2562 where XXXXXX is the bug number as listed below.
2564 188572 Valgrind on Mac should suppress setenv() mem leak
2565 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
2566 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
2567 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
2568 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
2569 254420 memory pool tracking broken
2570 254957 Test code failing to compile due to changes in memcheck.h
2571 255009 helgrind/drd: crash on chmod with invalid parameter
2572 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
2573 255355 helgrind/drd: crash on threaded programs doing fork
2575 255418 (SSE4.x) rint call compiled with ICC
2576 255822 --gen-suppressions can create invalid files: "too many callers [...]"
2577 255888 closing valgrindoutput tag outputted to log-stream on error
2578 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
2579 255966 Slowness when using mempool annotations
2580 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
2581 256600 super-optimized strcasecmp() false positive
2582 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
2583 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
2584 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
2585 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
2586 257276 Missing case in memcheck --track-origins=yes
2587 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
2588 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
2589 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
2590 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
2591 263099 callgrind_annotate counts Ir improperly [...]
2592 263877 undefined coprocessor instruction on ARMv7
2593 265964 configure FAIL with glibc-2.13
2594 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
2595 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
2596 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
2597 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
2598 n-i-bz DRD: disable --free-is-write due to implementation difficulties
2600 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
2604 Release 3.6.0 (21 October 2010)
2605 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2606 3.6.0 is a feature release with many significant improvements and the
2607 usual collection of bug fixes.
2609 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2610 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
2611 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
2613 -------------------------
2615 Here are some highlights. Details are shown further down:
2617 * Support for ARM/Linux.
2619 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
2621 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
2623 * Support for the SSE4.2 instruction set.
2625 * Enhancements to the Callgrind profiler, including the ability to
2626 handle CPUs with three levels of cache.
2628 * A new experimental heap profiler, DHAT.
2630 * A huge number of bug fixes and small enhancements.
2632 -------------------------
2634 Here are details of the above changes, together with descriptions of
2635 many other changes, and a list of fixed bugs.
2637 * ================== PLATFORM CHANGES =================
2639 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
2640 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
2641 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
2643 This requires a CPU capable of running the ARMv7-A instruction set
2644 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
2645 of the user space instruction set, including ARM and Thumb integer
2646 code, VFPv3, NEON and V6 media instructions. The Memcheck,
2647 Cachegrind and Massif tools work properly; other tools work to
2650 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
2651 with support for recent releases of the underlying toolchain
2652 components, notably gcc-4.5 and glibc-2.12.
2654 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
2655 support also works much better on OS X 10.5, and is as solid as
2658 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
2659 64-bit mode. In 32-bit mode, support is only available up to and
2660 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
2661 supported in 64-bit mode, and 32-bit mode does in fact support the
2662 bare minimum SSE4 instructions to needed to run programs on Mac OS X
2663 10.6 on 32-bit targets.
2665 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
2666 and including version 2.05 is supported.
2668 * ==================== TOOL CHANGES ====================
2670 * Cachegrind has a new processing script, cg_diff, which finds the
2671 difference between two profiles. It's very useful for evaluating
2672 the performance effects of a change in a program.
2674 Related to this change, the meaning of cg_annotate's (rarely-used)
2675 --threshold option has changed; this is unlikely to affect many
2676 people, if you do use it please see the user manual for details.
2678 * Callgrind now can do branch prediction simulation, similar to
2679 Cachegrind. In addition, it optionally can count the number of
2680 executed global bus events. Both can be used for a better
2681 approximation of a "Cycle Estimation" as derived event (you need to
2682 update the event formula in KCachegrind yourself).
2684 * Cachegrind and Callgrind now refer to the LL (last-level) cache
2685 rather than the L2 cache. This is to accommodate machines with
2686 three levels of caches -- if Cachegrind/Callgrind auto-detects the
2687 cache configuration of such a machine it will run the simulation as
2688 if the L2 cache isn't present. This means the results are less
2689 likely to match the true result for the machine, but
2690 Cachegrind/Callgrind's results are already only approximate, and
2691 should not be considered authoritative. The results are still
2692 useful for giving a general idea about a program's locality.
2694 * Massif has a new option, --pages-as-heap, which is disabled by
2695 default. When enabled, instead of tracking allocations at the level
2696 of heap blocks (as allocated with malloc/new/new[]), it instead
2697 tracks memory allocations at the level of memory pages (as mapped by
2698 mmap, brk, etc). Each mapped page is treated as its own block.
2699 Interpreting the page-level output is harder than the heap-level
2700 output, but this option is useful if you want to account for every
2701 byte of memory used by a program.
2703 * DRD has two new command-line options: --free-is-write and
2704 --trace-alloc. The former allows to detect reading from already freed
2705 memory, and the latter allows tracing of all memory allocations and
2708 * DRD has several new annotations. Custom barrier implementations can
2709 now be annotated, as well as benign races on static variables.
2711 * DRD's happens before / happens after annotations have been made more
2712 powerful, so that they can now also be used to annotate e.g. a smart
2713 pointer implementation.
2715 * Helgrind's annotation set has also been drastically improved, so as
2716 to provide to users a general set of annotations to describe locks,
2717 semaphores, barriers and condition variables. Annotations to
2718 describe thread-safe reference counted heap objects have also been
2721 * Memcheck has a new command-line option, --show-possibly-lost, which
2722 is enabled by default. When disabled, the leak detector will not
2723 show possibly-lost blocks.
2725 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
2726 has been added. DHAT keeps track of allocated heap blocks, and also
2727 inspects every memory reference to see which block (if any) is being
2728 accessed. This gives a lot of insight into block lifetimes,
2729 utilisation, turnover, liveness, and the location of hot and cold
2730 fields. You can use DHAT to do hot-field profiling.
2732 * ==================== OTHER CHANGES ====================
2734 * Improved support for unfriendly self-modifying code: the extra
2735 overhead incurred by --smc-check=all has been reduced by
2736 approximately a factor of 5 as compared with 3.5.0.
2738 * Ability to show directory names for source files in error messages.
2739 This is combined with a flexible mechanism for specifying which
2740 parts of the paths should be shown. This is enabled by the new flag
2743 * A new flag, --require-text-symbol, which will stop the run if a
2744 specified symbol is not found it a given shared object when it is
2745 loaded into the process. This makes advanced working with function
2746 intercepting and wrapping safer and more reliable.
2748 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
2749 and control of Valgrind is now available for the tools Memcheck and
2750 Helgrind. XML output from Valgrind is available for Memcheck,
2751 Helgrind and exp-Ptrcheck.
2753 * More reliable stack unwinding on amd64-linux, particularly in the
2754 presence of function wrappers, and with gcc-4.5 compiled code.
2756 * Modest scalability (performance improvements) for massive
2757 long-running applications, particularly for those with huge amounts
2760 * Support for analyzing programs running under Wine with has been
2761 improved. The header files <valgrind/valgrind.h>,
2762 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
2763 Windows-programs compiled with MinGW or one of the Microsoft Visual
2766 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
2767 The 32-bit simulator was not affected. This did not occur often,
2768 but when it did would usually crash the program under test.
2771 * A large number of bugs were fixed. These are shown below.
2773 * A number of bugs were investigated, and were candidates for fixing,
2774 but are not fixed in 3.6.0, due to lack of developer time. They may
2775 get fixed in later releases. They are:
2777 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
2778 212419 false positive "lock order violated" (A+B vs A)
2779 213685 Undefined value propagates past dependency breaking instruction
2780 216837 Incorrect instrumentation of NSOperationQueue on Darwin
2781 237920 valgrind segfault on fork failure
2782 242137 support for code compiled by LLVM-2.8
2783 242423 Another unknown Intel cache config value
2784 243232 Inconsistent Lock Orderings report with trylock
2785 243483 ppc: callgrind triggers VEX assertion failure
2786 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
2787 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
2789 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
2790 249435 Analyzing wine programs with callgrind triggers a crash
2791 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
2792 250065 Handling large allocations
2793 250101 huge "free" memory usage due to m_mallocfree.c
2794 "superblocks fragmentation"
2795 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
2796 252091 Callgrind on ARM does not detect function returns correctly
2797 252600 [PATCH] Allow lhs to be a pointer for shl/shr
2798 254420 memory pool tracking broken
2799 n-i-bz support for adding symbols for JIT generated code
2802 The following bugs have been fixed or resolved. Note that "n-i-bz"
2803 stands for "not in bugzilla" -- that is, a bug that was reported to us
2804 but never got a bugzilla entry. We encourage you to file bugs in
2805 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2806 mailing the developers (or mailing lists) directly -- bugs that are
2807 not entered into bugzilla tend to get forgotten about or ignored.
2809 To see details of a given bug, visit
2810 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2811 where XXXXXX is the bug number as listed below.
2813 135264 dcbzl instruction missing
2815 153699 Valgrind should report unaligned reads with movdqa
2817 190429 Valgrind reports lost of errors in ld.so
2818 with x86_64 2.9.90 glibc
2819 197266 valgrind appears to choke on the xmms instruction
2821 197988 Crash when demangling very large symbol names
2822 202315 unhandled syscall: 332 (inotify_init1)
2823 203256 Add page-level profiling to Massif
2824 205093 dsymutil=yes needs quotes, locking (partial fix)
2825 205241 Snow Leopard 10.6 support (partial fix)
2826 206600 Leak checker fails to upgrade indirect blocks when their
2827 parent becomes reachable
2828 210935 port valgrind.h (not valgrind) to win32 so apps run under
2829 wine can make client requests
2830 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
2831 within Linux ip-stack checksum functions
2832 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
2834 213685 Undefined value propagates past dependency breaking instruction
2836 215914 Valgrind inserts bogus empty environment variable
2838 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
2839 222545 shmat fails under valgind on some arm targets
2840 222560 ARM NEON support
2843 232509 Docs build fails with formatting inside <title></title> elements
2845 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
2846 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
2847 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
2848 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
2849 237485 symlink (syscall 57) is not supported on Mac OS
2850 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
2852 238208 is_just_below_ESP doesn't take into account red-zone
2853 238345 valgrind passes wrong $0 when executing a shell script
2854 238679 mq_timedreceive syscall doesn't flag the reception buffer
2856 238696 fcntl command F_DUPFD_CLOEXEC not supported
2857 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
2858 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
2859 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
2861 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
2867 242606 unhandled syscall: setegid (in Ptrcheck)
2868 242814 Helgrind "Impossible has happened" during
2869 QApplication::initInstance();
2870 243064 Valgrind attempting to read debug information from iso
2871 243270 Make stack unwinding in Valgrind wrappers more reliable
2872 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
2873 sysno = 277 (mq_open)
2874 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
2875 244493 ARM VFP d16-d31 registers support
2876 244670 add support for audit_session_self syscall on Mac OS 10.6
2877 244921 The xml report of helgrind tool is not well format
2878 244923 In the xml report file, the <preamble> not escape the
2879 xml char, eg '<','&','>'
2880 245535 print full path names in plain text reports
2881 245925 x86-64 red zone handling problem
2882 246258 Valgrind not catching integer underruns + new [] s
2883 246311 reg/reg cmpxchg doesn't work on amd64
2884 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
2885 246888 Improve Makefile.vex.am
2886 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
2887 to [f]chmod_extended
2888 247526 IBM POWER6 (ISA 2.05) support is incomplete
2889 247561 Some leak testcases fails due to reachable addresses in
2891 247875 sizeofIRType to handle Ity_I128
2892 247894 [PATCH] unhandled syscall sys_readahead
2893 247980 Doesn't honor CFLAGS passed to configure
2894 248373 darwin10.supp is empty in the trunk
2895 248822 Linux FIBMAP ioctl has int parameter instead of long
2896 248893 [PATCH] make readdwarf.c big endianess safe to enable
2897 unwinding on big endian systems
2898 249224 Syscall 336 not supported (SYS_proc_info)
2900 249775 Incorrect scheme for detecting NEON capabilities of host CPU
2901 249943 jni JVM init fails when using valgrind
2902 249991 Valgrind incorrectly declares AESKEYGENASSIST support
2904 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
2905 250799 frexp$fenv_access_off function generates SIGILL
2906 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
2907 251251 support pclmulqdq insn
2908 251362 valgrind: ARM: attach to debugger either fails or provokes
2910 251674 Unhandled syscall 294
2913 254257 Add support for debugfiles found by build-id
2914 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
2915 254646 Wrapped functions cause stack misalignment on OS X
2916 (and possibly Linux)
2917 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
2919 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
2923 Release 3.5.0 (19 August 2009)
2924 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2925 3.5.0 is a feature release with many significant improvements and the
2926 usual collection of bug fixes. The main improvement is that Valgrind
2927 now works on Mac OS X.
2929 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
2930 and X86/Darwin. Support for recent distros and toolchain components
2931 (glibc 2.10, gcc 4.5) has been added.
2933 -------------------------
2935 Here is a short summary of the changes. Details are shown further
2938 * Support for Mac OS X (10.5.x).
2940 * Improvements and simplifications to Memcheck's leak checker.
2942 * Clarification and simplifications in various aspects of Valgrind's
2945 * XML output for Helgrind and Ptrcheck.
2947 * Performance and stability improvements for Helgrind and DRD.
2949 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
2951 * A new experimental tool, BBV, useful for computer architecture
2954 * Improved Wine support, including ability to read Windows PDB
2957 -------------------------
2959 Here are details of the above changes, followed by descriptions of
2960 many other minor changes, and a list of fixed bugs.
2963 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
2964 called "Darwin" because that is the name of the OS core, which is the
2965 level that Valgrind works at.)
2969 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
2970 because it would require work and 10.4 is only becoming less common.
2972 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
2973 fairly well. For 10.5.x, 32-bit programs are the default even on
2974 64-bit machines, so it handles most current programs.
2976 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
2977 officially supported, but simple programs at least will probably work.
2978 However, start-up is slow.
2980 - PowerPC machines are not supported.
2982 Things that don't work:
2984 - The Ptrcheck tool.
2986 - Objective-C garbage collection.
2990 - If you have Rogue Amoeba's "Instant Hijack" program installed,
2991 Valgrind will fail with a SIGTRAP at start-up. See
2992 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
2997 - You will likely find --dsymutil=yes a useful option, as error
2998 messages may be imprecise without it.
3000 - Mac OS X support is new and therefore will be less robust than the
3001 Linux support. Please report any bugs you find.
3003 - Threaded programs may run more slowly than on Linux.
3005 Many thanks to Greg Parker for developing this port over several years.
3008 * Memcheck's leak checker has been improved.
3010 - The results for --leak-check=summary now match the summary results
3011 for --leak-check=full. Previously they could differ because
3012 --leak-check=summary counted "indirectly lost" blocks and
3013 "suppressed" blocks as "definitely lost".
3015 - Blocks that are only reachable via at least one interior-pointer,
3016 but are directly pointed to by a start-pointer, were previously
3017 marked as "still reachable". They are now correctly marked as
3020 - The default value for the --leak-resolution option has been
3021 changed from "low" to "high". In general, this means that more
3022 leak reports will be produced, but each leak report will describe
3023 fewer leaked blocks.
3025 - With --leak-check=full, "definitely lost" and "possibly lost"
3026 leaks are now considered as proper errors, ie. they are counted
3027 for the "ERROR SUMMARY" and affect the behaviour of
3028 --error-exitcode. These leaks are not counted as errors if
3029 --leak-check=summary is specified, however.
3031 - Documentation for the leak checker has been improved.
3034 * Various aspects of Valgrind's text output have changed.
3036 - Valgrind's start-up message has changed. It is shorter but also
3037 includes the command being run, which makes it easier to use
3038 --trace-children=yes. An example:
3040 - Valgrind's shut-down messages have also changed. This is most
3041 noticeable with Memcheck, where the leak summary now occurs before
3042 the error summary. This change was necessary to allow leaks to be
3043 counted as proper errors (see the description of the leak checker
3044 changes above for more details). This was also necessary to fix a
3045 longstanding bug in which uses of suppressions against leaks were
3046 not "counted", leading to difficulties in maintaining suppression
3047 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3049 - Behavior of -v has changed. In previous versions, -v printed out
3050 a mixture of marginally-user-useful information, and tool/core
3051 statistics. The statistics printing has now been moved to its own
3052 flag, --stats=yes. This means -v is less verbose and more likely
3053 to convey useful end-user information.
3055 - The format of some (non-XML) stack trace entries has changed a
3056 little. Previously there were six possible forms:
3058 0x80483BF: really (a.c:20)
3059 0x80483BF: really (in /foo/a.out)
3061 0x80483BF: (within /foo/a.out)
3062 0x80483BF: ??? (a.c:20)
3065 The third and fourth of these forms have been made more consistent
3066 with the others. The six possible forms are now:
3068 0x80483BF: really (a.c:20)
3069 0x80483BF: really (in /foo/a.out)
3070 0x80483BF: really (in ???)
3071 0x80483BF: ??? (in /foo/a.out)
3072 0x80483BF: ??? (a.c:20)
3075 Stack traces produced when --xml=yes is specified are different
3079 * Helgrind and Ptrcheck now support XML output, so they can be used
3080 from GUI tools. Also, the XML output mechanism has been
3083 - The XML format has been overhauled and generalised, so it is more
3084 suitable for error reporting tools in general. The Memcheck
3085 specific aspects of it have been removed. The new format, which
3086 is an evolution of the old format, is described in
3087 docs/internals/xml-output-protocol4.txt.
3089 - Memcheck has been updated to use the new format.
3091 - Helgrind and Ptrcheck are now able to emit output in this format.
3093 - The XML output mechanism has been overhauled. XML is now output
3094 to its own file descriptor, which means that:
3096 * Valgrind can output text and XML independently.
3098 * The longstanding problem of XML output being corrupted by
3099 unexpected un-tagged text messages is solved.
3101 As before, the destination for text output is specified using
3102 --log-file=, --log-fd= or --log-socket=.
3104 As before, XML output for a tool is enabled using --xml=yes.
3106 Because there's a new XML output channel, the XML output
3107 destination is now specified by --xml-file=, --xml-fd= or
3110 Initial feedback has shown this causes some confusion. To
3111 clarify, the two envisaged usage scenarios are:
3113 (1) Normal text output. In this case, do not specify --xml=yes
3114 nor any of --xml-file=, --xml-fd= or --xml-socket=.
3116 (2) XML output. In this case, specify --xml=yes, and one of
3117 --xml-file=, --xml-fd= or --xml-socket= to select the XML
3118 destination, one of --log-file=, --log-fd= or --log-socket=
3119 to select the destination for any remaining text messages,
3120 and, importantly, -q.
3122 -q makes Valgrind completely silent on the text channel,
3123 except in the case of critical failures, such as Valgrind
3124 itself segfaulting, or failing to read debugging information.
3125 Hence, in this scenario, it suffices to check whether or not
3126 any output appeared on the text channel. If yes, then it is
3127 likely to be a critical error which should be brought to the
3128 attention of the user. If no (the text channel produced no
3129 output) then it can be assumed that the run was successful.
3131 This allows GUIs to make the critical distinction they need to
3132 make (did the run fail or not?) without having to search or
3133 filter the text output channel in any way.
3135 It is also recommended to use --child-silent-after-fork=yes in
3139 * Improvements and changes in Helgrind:
3141 - XML output, as described above
3143 - Checks for consistent association between pthread condition
3144 variables and their associated mutexes are now performed.
3146 - pthread_spinlock functions are supported.
3148 - Modest performance improvements.
3150 - Initial (skeletal) support for describing the behaviour of
3151 non-POSIX synchronisation objects through ThreadSanitizer
3152 compatible ANNOTATE_* macros.
3154 - More controllable tradeoffs between performance and the level of
3155 detail of "previous" accesses in a race. There are now three
3158 * --history-level=full. This is the default, and was also the
3159 default in 3.4.x. It shows both stacks involved in a race, but
3160 requires a lot of memory and can be very slow in programs that
3161 do many inter-thread synchronisation events.
3163 * --history-level=none. This only shows the later stack involved
3164 in a race. This can be much faster than --history-level=full,
3165 but makes it much more difficult to find the other access
3166 involved in the race.
3168 The new intermediate setting is
3170 * --history-level=approx
3172 For the earlier (other) access, two stacks are presented. The
3173 earlier access is guaranteed to be somewhere in between the two
3174 program points denoted by those stacks. This is not as useful
3175 as showing the exact stack for the previous access (as per
3176 --history-level=full), but it is better than nothing, and it's
3177 almost as fast as --history-level=none.
3180 * New features and improvements in DRD:
3182 - The error messages printed by DRD are now easier to interpret.
3183 Instead of using two different numbers to identify each thread
3184 (Valgrind thread ID and DRD thread ID), DRD does now identify
3185 threads via a single number (the DRD thread ID). Furthermore
3186 "first observed at" information is now printed for all error
3187 messages related to synchronization objects.
3189 - Added support for named semaphores (sem_open() and sem_close()).
3191 - Race conditions between pthread_barrier_wait() and
3192 pthread_barrier_destroy() calls are now reported.
3194 - Added support for custom allocators through the macros
3195 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
3196 in <valgrind/valgrind.h>). An alternative for these two macros is
3197 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
3200 - Added support for annotating non-POSIX synchronization objects
3201 through several new ANNOTATE_*() macros.
3203 - OpenMP: added support for the OpenMP runtime (libgomp) included
3204 with gcc versions 4.3.0 and 4.4.0.
3208 - Added two new command-line options (--first-race-only and
3209 --segment-merging-interval).
3212 * Genuinely atomic support for x86/amd64/ppc atomic instructions
3214 Valgrind will now preserve (memory-access) atomicity of LOCK-
3215 prefixed x86/amd64 instructions, and any others implying a global
3216 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
3218 This means that Valgrinded processes will "play nicely" in
3219 situations where communication with other processes, or the kernel,
3220 is done through shared memory and coordinated with such atomic
3221 instructions. Prior to this change, such arrangements usually
3222 resulted in hangs, races or other synchronisation failures, because
3223 Valgrind did not honour atomicity of such instructions.
3226 * A new experimental tool, BBV, has been added. BBV generates basic
3227 block vectors for use with the SimPoint analysis tool, which allows
3228 a program's overall behaviour to be approximated by running only a
3229 fraction of it. This is useful for computer architecture
3230 researchers. You can run BBV by specifying --tool=exp-bbv (the
3231 "exp-" prefix is short for "experimental"). BBV was written by
3235 * Modestly improved support for running Windows applications under
3236 Wine. In particular, initial support for reading Windows .PDB debug
3237 information has been added.
3240 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
3241 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
3245 * The Valgrind client requests VALGRIND_PRINTF and
3246 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
3247 the string was always printed immediately on its own line. Now, the
3248 string will be added to a buffer but not printed until a newline is
3249 encountered, or other Valgrind output is printed (note that for
3250 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
3251 "other Valgrind output"). This allows you to use multiple
3252 VALGRIND_PRINTF calls to build up a single output line, and also to
3253 print multiple output lines with a single request (by embedding
3254 multiple newlines in the string).
3257 * The graphs drawn by Massif's ms_print program have changed slightly:
3259 - The half-height chars '.' and ',' are no longer drawn, because
3260 they are confusing. The --y option can be used if the default
3261 y-resolution is not high enough.
3263 - Horizontal lines are now drawn after the top of a snapshot if
3264 there is a gap until the next snapshot. This makes it clear that
3265 the memory usage has not dropped to zero between snapshots.
3268 * Something that happened in 3.4.0, but wasn't clearly announced: the
3269 option --read-var-info=yes can be used by some tools (Memcheck,
3270 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
3271 variable type and location information. This makes those tools
3272 start up more slowly and increases memory consumption, but
3273 descriptions of data addresses in error messages become more
3277 * exp-Omega, an experimental instantaneous leak-detecting tool, was
3278 disabled in 3.4.0 due to a lack of interest and maintenance,
3279 although the source code was still in the distribution. The source
3280 code has now been removed from the distribution. For anyone
3281 interested, the removal occurred in SVN revision r10247.
3284 * Some changes have been made to the build system.
3286 - VEX/ is now integrated properly into the build system. This means
3287 that dependency tracking within VEX/ now works properly, "make
3288 install" will work without requiring "make" before it, and
3289 parallel builds (ie. 'make -j') now work (previously a
3290 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
3291 was effectively ignored).
3293 - The --with-vex configure option has been removed. It was of
3294 little use and removing it simplified the build system.
3296 - The location of some install files has changed. This should not
3297 affect most users. Those who might be affected:
3299 * For people who use Valgrind with MPI programs, the installed
3300 libmpiwrap.so library has moved from
3301 $(INSTALL)/<platform>/libmpiwrap.so to
3302 $(INSTALL)/libmpiwrap-<platform>.so.
3304 * For people who distribute standalone Valgrind tools, the
3305 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
3306 have moved to $(INSTALL)/libcoregrind-<platform>.a.
3308 These changes simplify the build system.
3310 - Previously, all the distributed suppression (*.supp) files were
3311 installed. Now, only default.supp is installed. This should not
3312 affect users as the other installed suppression files were not
3313 read; the fact that they were installed was a mistake.
3316 * KNOWN LIMITATIONS:
3318 - Memcheck is unusable with the Intel compiler suite version 11.1,
3319 when it generates code for SSE2-and-above capable targets. This
3320 is because of icc's use of highly optimised inlined strlen
3321 implementations. It causes Memcheck to report huge numbers of
3322 false errors even in simple programs. Helgrind and DRD may also
3325 Versions 11.0 and earlier may be OK, but this has not been
3329 The following bugs have been fixed or resolved. Note that "n-i-bz"
3330 stands for "not in bugzilla" -- that is, a bug that was reported to us
3331 but never got a bugzilla entry. We encourage you to file bugs in
3332 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3333 mailing the developers (or mailing lists) directly -- bugs that are
3334 not entered into bugzilla tend to get forgotten about or ignored.
3336 To see details of a given bug, visit
3337 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3338 where XXXXXX is the bug number as listed below.
3340 84303 How about a LockCheck tool?
3341 91633 dereference of null ptr in vgPlain_st_basetype
3342 97452 Valgrind doesn't report any pthreads problems
3343 100628 leak-check gets assertion failure when using
3344 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
3345 108528 NPTL pthread cleanup handlers not called
3346 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
3347 110128 mallinfo is not implemented...
3348 110770 VEX: Generated files not always updated when making valgrind
3349 111102 Memcheck: problems with large (memory footprint) applications
3350 115673 Vex's decoder should never assert
3351 117564 False positive: Syscall param clone(child_tidptr) contains
3352 uninitialised byte(s)
3353 119404 executing ssh from inside valgrind fails
3354 133679 Callgrind does not write path names to sources with dwarf debug
3356 135847 configure.in problem with non gnu compilers (and possible fix)
3357 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
3358 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
3359 136230 memcheck reports "possibly lost", should be "still reachable"
3360 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
3361 137904 Valgrind reports a memory leak when using POSIX threads,
3363 139076 valgrind VT_GETSTATE error
3364 142228 complaint of elf_dynamic_do_rela in trivial usage
3365 145347 spurious warning with USBDEVFS_REAPURB
3366 148441 (wine) can't find memory leak in Wine, win32 binary
3368 148742 Leak-check fails assert on exit
3369 149878 add (proper) check for calloc integer overflow
3370 150606 Call graph is broken when using callgrind control
3371 152393 leak errors produce an exit code of 0. I need some way to
3372 cause leak errors to result in a nonzero exit code.
3373 157154 documentation (leak-resolution doc speaks about num-callers
3374 def=4) + what is a loss record
3375 159501 incorrect handling of ALSA ioctls
3376 162020 Valgrinding an empty/zero-byte file crashes valgrind
3377 162482 ppc: Valgrind crashes while reading stabs information
3378 162718 x86: avoid segment selector 0 in sys_set_thread_area()
3379 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
3380 163560 VEX/test_main.c is missing from valgrind-3.3.1
3381 164353 malloc_usable_size() doesn't return a usable size
3382 165468 Inconsistent formatting in memcheck manual -- please fix
3383 169505 main.c:286 (endOfInstr):
3384 Assertion 'ii->cost_offset == *cost_offset' failed
3385 177206 Generate default.supp during compile instead of configure
3386 177209 Configure valt_load_address based on arch+os
3387 177305 eventfd / syscall 323 patch lost
3388 179731 Tests fail to build because of inlining of non-local asm labels
3389 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
3390 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
3391 181594 Bogus warning for empty text segment
3392 181707 dwarf doesn't require enumerations to have name
3393 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3394 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3395 Assertion '!already_present' failed.
3396 185359 exp-ptrcheck: unhandled syscall getresuid()
3397 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
3398 185816 Valgrind is unable to handle debug info for files with split
3399 debug info that are prelinked afterwards
3400 185980 [darwin] unhandled syscall: sem_open
3401 186238 bbToIR_AMD64: disInstr miscalculated next %rip
3402 186507 exp-ptrcheck unhandled syscalls prctl, etc.
3403 186790 Suppression pattern used for leaks are not reported
3404 186796 Symbols with length>200 in suppression files are ignored
3405 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
3406 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
3407 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
3408 188046 bashisms in the configure script
3409 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
3410 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
3411 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
3412 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
3413 assertion fail "!lock->heldBy"
3414 188427 Add support for epoll_create1 (with patch)
3415 188530 Support for SIOCGSTAMPNS
3416 188560 Include valgrind.spec in the tarball
3417 188572 Valgrind on Mac should suppress setenv() mem leak
3418 189054 Valgrind fails to build because of duplicate non-local asm labels
3419 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
3420 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
3421 189763 drd assertion failure: s_threadinfo[tid].is_recording
3422 190219 unhandled syscall: 328 (x86-linux)
3423 190391 dup of 181394; see above
3424 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
3425 190820 No debug information on powerpc-linux
3426 191095 PATCH: Improve usbdevfs ioctl handling
3427 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
3429 191189 --xml=yes should obey --gen-suppressions=all
3430 191192 syslog() needs a suppression on macosx
3431 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
3432 191761 getrlimit on MacOSX
3433 191992 multiple --fn-skip only works sometimes; dependent on order
3434 192634 V. reports "aspacem sync_check_mapping_callback:
3435 segment mismatch" on Darwin
3436 192954 __extension__ missing on 2 client requests
3437 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
3438 194474 "INSTALL" file has different build instructions than "README"
3439 194671 Unhandled syscall (sem_wait?) from mac valgrind
3440 195069 memcheck: reports leak (memory still reachable) for
3442 195169 drd: (vgDrd_barrier_post_wait):
3443 Assertion 'r->sg[p->post_iteration]' failed.
3444 195268 valgrind --log-file doesn't accept ~/...
3445 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
3446 195860 WARNING: unhandled syscall: unix:223
3447 196528 need a error suppression for pthread_rwlock_init under os x?
3448 197227 Support aio_* syscalls on Darwin
3449 197456 valgrind should reject --suppressions=(directory)
3450 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
3451 197591 unhandled syscall 27 (mincore)
3452 197793 Merge DCAS branch to the trunk == 85756, 142103
3453 197794 Avoid duplicate filenames in Vex
3454 197898 make check fails on current SVN
3455 197901 make check fails also under exp-ptrcheck in current SVN
3456 197929 Make --leak-resolution=high the default
3457 197930 Reduce spacing between leak reports
3458 197933 Print command line of client at start-up, and shorten preamble
3459 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
3460 198395 add BBV to the distribution as an experimental tool
3461 198624 Missing syscalls on Darwin: 82, 167, 281, 347
3462 198649 callgrind_annotate doesn't cumulate counters
3463 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
3464 199977 Valgrind complains about an unrecognized instruction in the
3465 atomic_incs test program
3466 200029 valgrind isn't able to read Fedora 12 debuginfo
3467 200760 darwin unhandled syscall: unix:284
3468 200827 DRD doesn't work on Mac OS X
3469 200990 VG_(read_millisecond_timer)() does not work correctly
3470 201016 Valgrind does not support pthread_kill() on Mac OS
3471 201169 Document --read-var-info
3472 201323 Pre-3.5.0 performance sanity checking
3473 201384 Review user manual for the 3.5.0 release
3474 201585 mfpvr not implemented on ppc
3475 201708 tests failing because x86 direction flag is left set
3476 201757 Valgrind doesn't handle any recent sys_futex additions
3477 204377 64-bit valgrind can not start a shell script
3478 (with #!/path/to/shell) if the shell is a 32-bit executable
3479 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
3480 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
3481 about memory allocation events with memory access tracing enabled
3482 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
3484 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
3488 Release 3.4.1 (28 February 2009)
3489 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3490 3.4.1 is a bug-fix release that fixes some regressions and assertion
3491 failures in debug info reading in 3.4.0, most notably incorrect stack
3492 traces on amd64-linux on older (glibc-2.3 based) systems. Various
3493 other debug info problems are also fixed. A number of bugs in the
3494 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
3496 In view of the fact that 3.4.0 contains user-visible regressions
3497 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
3498 encouraged to ship 3.4.1 in preference to 3.4.0.
3500 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
3501 bugzilla" -- that is, a bug that was reported to us but never got a
3502 bugzilla entry. We encourage you to file bugs in bugzilla
3503 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
3504 developers (or mailing lists) directly -- bugs that are not entered
3505 into bugzilla tend to get forgotten about or ignored.
3507 n-i-bz Fix various bugs reading icc-11 generated debug info
3508 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
3509 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
3510 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
3511 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
3512 179618 exp-ptrcheck crashed / exit prematurely
3513 179624 helgrind: false positive races with pthread_create and
3514 recv/open/close/read
3515 134207 pkg-config output contains @VG_PLATFORM@
3516 176926 floating point exception at valgrind startup with PPC 440EPX
3517 181594 Bogus warning for empty text segment
3518 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
3519 181707 Dwarf3 doesn't require enumerations to have name
3520 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3521 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3522 Assertion '!already_present' failed.
3523 185359 exp-ptrcheck unhandled syscall getresuid()
3525 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
3526 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
3530 Release 3.4.0 (2 January 2009)
3531 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3532 3.4.0 is a feature release with many significant improvements and the
3533 usual collection of bug fixes. This release supports X86/Linux,
3534 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
3535 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
3537 3.4.0 brings some significant tool improvements. Memcheck can now
3538 report the origin of uninitialised values, the thread checkers
3539 Helgrind and DRD are much improved, and we have a new experimental
3540 tool, exp-Ptrcheck, which is able to detect overruns of stack and
3541 global arrays. In detail:
3543 * Memcheck is now able to track the origin of uninitialised values.
3544 When it reports an uninitialised value error, it will try to show
3545 the origin of the value, as either a heap or stack allocation.
3546 Origin tracking is expensive and so is not enabled by default. To
3547 use it, specify --track-origins=yes. Memcheck's speed will be
3548 essentially halved, and memory usage will be significantly
3549 increased. Nevertheless it can drastically reduce the effort
3550 required to identify the root cause of uninitialised value errors,
3551 and so is often a programmer productivity win, despite running more
3554 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
3555 3.4.0, will be released shortly.
3557 * Helgrind's race detection algorithm has been completely redesigned
3558 and reimplemented, to address usability and scalability concerns:
3560 - The new algorithm has a lower false-error rate: it is much less
3561 likely to report races that do not really exist.
3563 - Helgrind will display full call stacks for both accesses involved
3564 in a race. This makes it easier to identify the root causes of
3567 - Limitations on the size of program that can run have been removed.
3569 - Performance has been modestly improved, although that is very
3572 - Direct support for Qt4 threading has been added.
3574 - pthread_barriers are now directly supported.
3576 - Helgrind works well on all supported Linux targets.
3578 * The DRD thread debugging tool has seen major improvements:
3580 - Greatly improved performance and significantly reduced memory
3583 - Support for several major threading libraries (Boost.Thread, Qt4,
3584 glib, OpenMP) has been added.
3586 - Support for atomic instructions, POSIX semaphores, barriers and
3587 reader-writer locks has been added.
3589 - Works now on PowerPC CPUs too.
3591 - Added support for printing thread stack usage at thread exit time.
3593 - Added support for debugging lock contention.
3595 - Added a manual for Drd.
3597 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
3598 checks for misuses of pointers. In that sense it is a bit like
3599 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
3600 detect overruns of stack and global arrays, it can detect
3601 arbitrarily far out-of-bounds accesses to heap blocks, and it can
3602 detect accesses heap blocks that have been freed a very long time
3603 ago (millions of blocks in the past).
3605 Ptrcheck currently works only on x86-linux and amd64-linux. To use
3606 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
3607 of the main Valgrind documentation. As this is an experimental
3608 tool, we would be particularly interested in hearing about your
3609 experiences with it.
3611 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
3612 longer built by default, although the code remains in the repository
3613 and the tarball. This is due to three factors: a perceived lack of
3614 users, a lack of maintenance, and concerns that it may not be
3615 possible to achieve reliable operation using the existing design.
3617 * As usual, support for the latest Linux distros and toolchain
3618 components has been added. It should work well on Fedora Core 10,
3619 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
3620 state) is supported, as is glibc-2.9. The C++ demangler has been
3621 updated so as to work well with C++ compiled by even the most recent
3624 * You can now use frame-level wildcards in suppressions. This was a
3625 frequently-requested enhancement. A line "..." in a suppression now
3626 matches zero or more frames. This makes it easier to write
3627 suppressions which are precise yet insensitive to changes in
3630 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
3632 * Very basic support for IBM Power6 has been added (64-bit processes only).
3634 * Valgrind is now cross-compilable. For example, it is possible to
3635 cross compile Valgrind on an x86/amd64-linux host, so that it runs
3636 on a ppc32/64-linux target.
3638 * You can set the main thread's stack size at startup using the
3639 new --main-stacksize= flag (subject of course to ulimit settings).
3640 This is useful for running apps that need a lot of stack space.
3642 * The limitation that you can't use --trace-children=yes together
3643 with --db-attach=yes has been removed.
3645 * The following bugs have been fixed. Note that "n-i-bz" stands for
3646 "not in bugzilla" -- that is, a bug that was reported to us but
3647 never got a bugzilla entry. We encourage you to file bugs in
3648 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3649 mailing the developers (or mailing lists) directly.
3651 n-i-bz Make return types for some client requests 64-bit clean
3652 n-i-bz glibc 2.9 support
3653 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
3654 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
3655 n-i-bz Building in an env without gdb gives bogus gdb attach
3656 92456 Tracing the origin of uninitialised memory
3657 106497 Valgrind does not demangle some C++ template symbols
3659 151612 Suppression with "..." (frame-level wildcards in .supp files)
3660 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
3661 159285 unhandled syscall:25 (stime, on x86-linux)
3662 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
3663 160954 ppc build of valgrind crashes with illegal instruction (isel)
3664 160956 mallinfo implementation, w/ patch
3665 162092 Valgrind fails to start gnome-system-monitor
3666 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
3667 163794 assertion failure with "--track-origins=yes"
3668 163933 sigcontext.err and .trapno must be set together
3669 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
3670 164476 Missing kernel module loading system calls
3671 164669 SVN regression: mmap() drops posix file locks
3672 166581 Callgrind output corruption when program forks
3673 167288 Patch file for missing system calls on Cell BE
3674 168943 unsupported scas instruction pentium
3675 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
3676 172417 x86->IR: 0x82 ...
3677 172563 amd64->IR: 0xD9 0xF5 - fprem1
3678 173099 .lds linker script generation error
3679 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
3680 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
3682 174908 --log-file value not expanded correctly for core file
3683 175044 Add lookup_dcookie for amd64
3684 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
3686 Developer-visible changes:
3688 * Valgrind's debug-info reading machinery has been majorly overhauled.
3689 It can now correctly establish the addresses for ELF data symbols,
3690 which is something that has never worked properly before now.
3692 Also, Valgrind can now read DWARF3 type and location information for
3693 stack and global variables. This makes it possible to use the
3694 framework to build tools that rely on knowing the type and locations
3695 of stack and global variables, for example exp-Ptrcheck.
3697 Reading of such information is disabled by default, because most
3698 tools don't need it, and because it is expensive in space and time.
3699 However, you can force Valgrind to read it, using the
3700 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
3701 make use of such information, if present, to provide source-level
3702 descriptions of data addresses in the error messages they create.
3704 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
3705 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).