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 379754 Fix missing syscall ulock_wait (OS X 10.12)
73 380397 s390x: __GI_strcspn() replacemenet needed
74 381162 possible array overrun in VEX register allocator
75 381272 ppc64 doesn't compile test_isa_2_06_partx.c without VSX support
76 381274 powerpc too chatty even with --sigill-diagnostics=no
77 381289 epoll_pwait can have a NULL sigmask
78 381553 VEX register allocator v3
79 381556 arm64: Handle feature registers access on 4.11 Linux kernel or later
80 381769 Use ucontext_t instead of struct ucontext
81 381805 arm32 needs ld.so index hardwire for new glibc security fixes
82 382256 gz compiler flag test doesn't work for gold
83 382407 vg_perf needs "--terse" command line option
84 382515 "Assertion 'di->have_dinfo' failed." on wine's dlls/mscoree/tests/[..]
85 382563 MIPS MSA ASE support
86 382998 xml-socket doesn't work
87 383275 massif: m_xarray.c:162 (ensureSpaceXA): Assertion '!xa->arr' failed
88 383723 Fix missing kevent_qos syscall (macOS 10.11)
89 == 385604 illegal hardware instruction (OpenCV cv::namedWindow)
90 384096 Mention AddrCheck at Memcheck's command line option [..]
91 384230 vex x86->IR: 0x67 0xE8 0xAB 0x68
92 == 384156 vex x86->IR: 0x67 0xE8 0x6B 0x6A
93 == 386115 vex x86->IR: 0x67 0xE8 0xD3 0x8B any program
94 == 388407 vex x86->IR: 0x67 0xE8 0xAB 0x29
95 == 394903 vex x86->IR: 0x67 0xE8 0x1B 0xDA
96 384337 performance improvements to VEX register allocator v2 and v3
97 384526 reduce number of spill insns generated by VEX register allocator v3
98 384584 Callee saved regs listed first for AMD64, X86, and PPC architectures
99 384631 Sanitise client args as printed with -v
100 384633 Add a simple progress-reporting facility
101 384987 VEX regalloc: allocate caller-save registers for short lived vregs
102 385055 PPC VEX temporary storage exhausted
103 385182 PPC64 is missing support for the DSCR
104 385183 PPC64, Add support for xscmpeqdp, xscmpgtdp, xscmpgedp, xsmincdp
105 385207 PPC64, generate_store_FPRF() generates too many Iops
106 385208 PPC64, xxperm instruction exhausts temporary memory
107 385210 PPC64, vpermr instruction could exhaust temporary memory
108 385279 unhandled syscall: mach:43 (mach_generate_activity_id)
109 == 395136 valgrind: m_syswrap/syswrap-main.c:438 (Bool eq_Syscall[..]
110 == 387045 Valgrind crashing on High Sierra when testing any newly [..]
111 385334 PPC64, fix vpermr, xxperm, xxpermr mask value.
112 385408 s390x: z13 vector "support" instructions not implemented
113 385412 s390x: new non-vector z13 instructions not implemented
114 385868 glibc ld.so _dl_runtime_resolve_avx_slow conditional jump warning.
115 385912 none/tests/rlimit_nofile fails on newer glibc/kernel.
116 385939 Optionally exit on the first error
117 386318 valgrind.org/info/tools.html is missing SGCheck
118 386425 running valgrind + wine on armv7l gives illegal opcode
119 386397 PPC64, valgrind truncates powerpc timebase to 32-bits.
120 387410 MIPSr6 support
121 387664 Memcheck: make expensive-definedness-checks be the default
122 387712 s390x cgijnl reports Conditional jump depends on uninitialised value
123 387766 asm shifts cause false positive "Conditional jump or move depends
124 on uninitialised value"
125 387773 .gnu_debugaltlink paths resolve relative to .debug file, not symlink
126 388786 Support bpf syscall in amd64 Linux
127 388862 Add replacements for wmemchr and wcsnlen on Linux
128 389065 valgrind meets gcc flag -Wlogical-op
129 389373 exp-sgcheck the 'impossible' happened as Ist_LoadG is not instrumented
130 390471 suppression by specification of source-file line number
131 390723 make xtree dump files world wide readable, similar to log files
132 391164 constraint bug in tests/ppc64/test_isa_2_07_part1.c for mtfprwa
133 391861 Massif Assertion 'n_ips >= 1 && n_ips <= VG_(clo_backtrace_size)'
134 392118 unhandled amd64-linux syscall: 332 (statx)
135 392449 callgrind not clearing the number of calls properly
136 393017 Add missing support for xsmaxcdp instruction, bug fixes for xsmincdp,
137 lxssp, stxssp and stxvl instructions.
138 393023 callgrind_control risks using the wrong vgdb
139 393062 build-id ELF phdrs read causes "debuginfo reader: ensure_valid failed"
140 393099 posix_memalign() invalid write if alignment == 0
141 393146 failing assert "is_DebugInfo_active(di)"
142 395709 PPC64 is missing support for the xvnegsp instruction
143 395682 Accept read-only PT_LOAD segments and .rodata by ld -z separate-code
145 396475 valgrind OS-X build: config.h not found (out-of-tree macOS builds)
146 396887 arch_prctl should return EINVAL on unknown option
147 == 397286 crash before launching binary (Unsupported arch_prctl option)
148 == 397393 valgrind: the 'impossible' happened: (Archlinux)
149 == 397521 valgrind: the 'impossible' happened: Unsupported [..]
150 397012 glibc ld.so uses arch_prctl on i386
151 397354 utimensat should ignore timespec tv_sec if tv_nsec is UTIME_NOW/OMIT
152 397424 glibc 2.27 and gdb_server tests
153 398066 s390x: cgijl dep1, 0 reports false unitialised values warning
155 n-i-bz Fix missing workq_ops operations (macOS)
156 n-i-bz fix bug in strspn replacement
157 n-i-bz Add support for the Linux BLKFLSBUF ioctl
158 n-i-bz Add support for the Linux BLKREPORTZONE and BLKRESETZONE ioctls
159 n-i-bz Fix possible stack trashing by semctl syscall wrapping
160 n-i-bz Add support for the Linux membarrier() system call
162 Release 3.13.0 (15 June 2017)
163 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
165 3.13.0 is a feature release with many improvements and the usual collection of
168 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
169 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
170 MIPS64/Linux, ARM/Android, ARM64/Android, MIPS32/Android, X86/Android,
171 X86/Solaris, AMD64/Solaris and AMD64/MacOSX 10.12.
173 * ==================== CORE CHANGES ===================
175 * The translation cache size has been increased to keep up with the demands of
176 large applications. The maximum number of sectors has increased from 24 to
177 48. The default number of sectors has increased from 16 to 32 on all
178 targets except Android, where the increase is from 6 to 12.
180 * The amount of memory that Valgrind can use has been increased from 64GB to
181 128GB. In particular this means your application can allocate up to about
182 60GB when running on Memcheck.
184 * Valgrind's default load address has been changed from 0x3800'0000 to
185 0x5800'0000, so as to make it possible to load larger executables. This
186 should make it possible to load executables of size at least 1200MB.
188 * A massive spaceleak caused by reading compressed debuginfo files has been
189 fixed. Valgrind should now be entirely usable with gcc-7.0 "-gz" created
192 * The C++ demangler has been updated.
194 * Support for demangling Rust symbols has been added.
196 * A new representation of stack traces, the "XTree", has been added. An XTree
197 is a tree of stacktraces with data associated with the stacktraces. This is
198 used by various tools (Memcheck, Helgrind, Massif) to report on the heap
199 consumption of your program. Reporting is controlled by the new options
200 --xtree-memory=none|allocs|full and --xtree-memory-file=<file>.
202 A report can also be produced on demand using the gdbserver monitor command
203 'xtmemory [<filename>]>'. The XTree can be output in 2 formats: 'callgrind
204 format' and 'massif format. The existing visualisers for these formats (e.g.
205 callgrind_annotate, KCachegrind, ms_print) can be used to visualise and
206 analyse these reports.
208 Memcheck can also produce XTree leak reports using the Callgrind file
209 format. For more details, see the user manual.
211 * ================== PLATFORM CHANGES =================
213 * ppc64: support for ISA 3.0B and various fixes for existing 3.0 support
215 * amd64: fixes for JIT failure problems on long AVX2 code blocks
217 * amd64 and x86: support for CET prefixes has been added
219 * arm32: a few missing ARMv8 instructions have been implemented
221 * arm64, mips64, mips32: an alternative implementation of Load-Linked and
222 Store-Conditional instructions has been added. This is to deal with
223 processor implementations that implement the LL/SC specifications strictly
224 and as a result cause Valgrind to hang in certain situations. The
225 alternative implementation is automatically enabled at startup, as required.
226 You can use the option --sim-hints=fallback-llsc to force-enable it if you
229 * Support for OSX 10.12 has been improved.
231 * On Linux, clone handling has been improved to honour CLONE_VFORK that
232 involves a child stack. Note however that CLONE_VFORK | CLONE_VM is handled
233 like CLONE_VFORK (by removing CLONE_VM), so applications that depend on
234 CLONE_VM exact semantics will (still) not work.
236 * The TileGX/Linux port has been removed because it appears to be both unused
239 * ==================== TOOL CHANGES ====================
243 - Memcheck should give fewer false positives when running optimised
244 Clang/LLVM generated code.
246 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
248 - New command line options --xtree-leak=no|yes and --xtree-leak-file=<file>
249 to produce the end of execution leak report in a xtree callgrind format
252 - New option 'xtleak' in the memcheck leak_check monitor command, to produce
253 the leak report in an xtree file.
257 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
259 - For some workloads (typically, for big applications), Massif memory
260 consumption and CPU consumption has decreased significantly.
264 - Support for --xtree-memory and 'xtmemory [<filename>]>'.
266 - addition of client request VALGRIND_HG_GNAT_DEPENDENT_MASTER_JOIN, useful
267 for Ada gnat compiled applications.
269 * ==================== OTHER CHANGES ====================
271 * For Valgrind developers: in an outer/inner setup, the outer Valgrind will
272 append the inner guest stacktrace to the inner host stacktrace. This helps
273 to investigate the errors reported by the outer, when they are caused by the
274 inner guest program (such as an inner regtest). See README_DEVELOPERS for
277 * To allow fast detection of callgrind files by desktop environments and file
278 managers, the format was extended to have an optional first line that
279 uniquely identifies the format ("# callgrind format"). Callgrind creates
280 this line now, as does the new xtree functionality.
282 * File name template arguments (such as --log-file, --xtree-memory-file, ...)
283 have a new %n format letter that is replaced by a sequence number.
285 * "--version -v" now shows the SVN revision numbers from which Valgrind was
288 * ==================== FIXED BUGS ====================
290 The following bugs have been fixed or resolved. Note that "n-i-bz"
291 stands for "not in bugzilla" -- that is, a bug that was reported to us
292 but never got a bugzilla entry. We encourage you to file bugs in
293 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
294 than mailing the developers (or mailing lists) directly -- bugs that
295 are not entered into bugzilla tend to get forgotten about or ignored.
297 To see details of a given bug, visit
298 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
299 where XXXXXX is the bug number as listed below.
301 162848 --log-file output isn't split when a program forks
302 340777 Illegal instruction on mips (ar71xx)
303 341481 MIPS64: Iop_CmpNE32 triggers false warning on MIPS64 platforms
304 342040 Valgrind mishandles clone with CLONE_VFORK | CLONE_VM that clones
305 to a different stack.
306 344139 x86 stack-seg overrides, needed by the Wine people
307 344524 store conditional of guest applications always fail - observed on
309 348616 Wine/valgrind: noted but unhandled ioctl 0x5390 [..] (DVD_READ_STRUCT)
310 352395 Please provide SVN revision info in --version -v
311 352767 Wine/valgrind: noted but unhandled ioctl 0x5307 [..] (CDROMSTOP)
312 356374 Assertion 'DRD_(g_threadinfo)[tid].pt_threadid !=
313 INVALID_POSIX_THREADID' failed
314 358213 helgrind/drd bar_bad testcase hangs or crashes with new glibc pthread
315 barrier implementation
316 358697 valgrind.h: Some code remains even when defining NVALGRIND
317 359202 Add musl libc configure/compile
318 360415 amd64 instructions ADCX and ADOX are not implemented in VEX
319 == 372828 (vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10)
320 360429 unhandled ioctl 0x530d with no size/direction hints (CDROMREADMODE1)
321 362223 assertion failed when .valgrindrc is a directory instead of a file
322 367543 bt/btc/btr/bts x86/x86_64 instructions are poorly-handled wrt flags
323 367942 Segfault vgPlain_do_sys_sigaction (m_signals.c:1138)
324 368507 can't malloc chunks larger than about 34GB
325 368529 Android arm target link error, missing atexit and pthread_atfork
326 368863 WARNING: unhandled arm64-linux syscall: 100 (get_robust_list)
327 368865 WARNING: unhandled arm64-linux syscall: 272 (kcmp)
328 368868 disInstr(arm64): unhandled instruction 0xD53BE000 = cntfrq_el0 (ARMv8)
329 368917 WARNING: unhandled arm64-linux syscall: 218 (request_key)
330 368918 WARNING: unhandled arm64-linux syscall: 127 (sched_rr_get_interval)
331 368922 WARNING: unhandled arm64-linux syscall: 161 (sethostname)
332 368924 WARNING: unhandled arm64-linux syscall: 84 (sync_file_range)
333 368925 WARNING: unhandled arm64-linux syscall: 130 (tkill)
334 368926 WARNING: unhandled arm64-linux syscall: 97 (unshare)
335 369459 valgrind on arm64 violates the ARMv8 spec (ldxr/stxr)
336 370028 Reduce the number of compiler warnings on MIPS platforms
337 370635 arm64 missing syscall getcpu
338 371225 Fix order of timer_{gettime,getoverrun,settime} syscalls on arm64
339 371227 Clean AArch64 syscall table
340 371412 Rename wrap_sys_shmat to sys_shmat like other wrappers
341 371471 Valgrind complains about non legit memory leaks on placement new (C++)
342 371491 handleAddrOverrides() is [incorrect] when ASO prefix is used
343 371503 disInstr(arm64): unhandled instruction 0xF89F0000
344 371869 support '%' in symbol Z-encoding
345 371916 execution tree xtree concept
346 372120 c++ demangler demangles symbols which are not c++
347 372185 Support of valgrind on ARMv8 with 32 bit executable
348 372188 vex amd64->IR: 0x66 0xF 0x3A 0x62 0x4A 0x10 0x10 0x48 (PCMPxSTRx $0x10)
349 372195 Power PC, xxsel instruction is not always recognized.
350 372504 Hanging on exit_group
351 372600 process loops forever when fatal signals are arriving quickly
352 372794 LibVEX (arm32 front end): 'Assertion szBlg2 <= 3' failed
353 373046 Stacks registered by core are never deregistered
354 373069 memcheck/tests/leak_cpp_interior fails with GCC 5.1+
355 373086 Implement additional Xen hypercalls
356 373192 Calling posix_spawn in glibc 2.24 completely broken
357 373488 Support for fanotify API on ARM64 architecture
358 == 368864 WARNING: unhandled arm64-linux syscall: 262 (fanotify_init)
359 373555 Rename BBPTR to GSPTR as it denotes guest state pointer only
360 373938 const IRExpr arguments for matchIRExpr()
361 374719 some spelling fixes
362 374963 increase valgrind's load address to prevent mmap failure
363 375514 valgrind_get_tls_addr() does not work in case of static TLS
364 375772 +1 error in get_elf_symbol_info() when computing value of 'hi' address
365 for ML_(find_rx_mapping)()
366 375806 Test helgrind/tests/tc22_exit_w_lock fails with glibc 2.24
367 375839 Temporary storage exhausted, with long sequence of vfmadd231ps insns
368 == 377159 "vex: the `impossible' happened" still present
369 == 375150 Assertion 'tres.status == VexTransOK' failed
370 == 378068 valgrind crashes on AVX2 function in FFmpeg
371 376142 Segfaults on MIPS Cavium Octeon boards
372 376279 disInstr(arm64): unhandled instruction 0xD50320FF
373 376455 Solaris: unhandled syscall lgrpsys(180)
374 376518 Solaris: unhandled fast trap getlgrp(6)
375 376611 ppc64 and arm64 don't know about prlimit64 syscall
376 376729 PPC64, remove R2 from the clobber list
378 376956 syswrap of SNDDRV and DRM_IOCTL_VERSION causing some addresses
379 to be wrongly marked as addressable
380 377066 Some Valgrind unit tests fail to compile on Ubuntu 16.10 with
381 PIE enabled by default
382 377376 memcheck/tests/linux/getregset fails with glibc2.24
383 377427 PPC64, lxv instruction failing on odd destination register
384 377478 PPC64: ISA 3.0 setup fixes
385 377698 Missing memory check for futex() uaddr arg for FUTEX_WAKE
386 and FUTEX_WAKE_BITSET, check only 4 args for FUTEX_WAKE_BITSET,
387 and 2 args for FUTEX_TRYLOCK_PI
388 377717 Fix massive space leak when reading compressed debuginfo sections
389 377891 Update Xen 4.6 domctl wrappers
390 377930 fcntl syscall wrapper is missing flock structure check
391 378524 libvexmultiarch_test regression on s390x and ppc64
392 378535 Valgrind reports INTERNAL ERROR in execve syscall wrapper
393 378673 Update libiberty demangler
394 378931 Add ISA 3.0B additional isnstructions, add OV32, CA32 setting support
395 379039 syscall wrapper for prctl(PR_SET_NAME) must not check more than 16 bytes
396 379094 Valgrind reports INTERNAL ERROR in rt_sigsuspend syscall wrapper
397 379371 UNKNOWN task message [id 3444, to mach_task_self(), reply 0x603]
398 (task_register_dyld_image_infos)
399 379372 UNKNOWN task message [id 3447, to mach_task_self(), reply 0x603]
400 (task_register_dyld_shared_cache_image_info)
401 379390 unhandled syscall: mach:70 (host_create_mach_voucher_trap)
402 379473 MIPS: add support for rdhwr cycle counter register
403 379504 remove TileGX/Linux port
404 379525 Support more x86 nop opcodes
405 379838 disAMode(x86): not an addr!
406 379703 PC ISA 3.0 fixes: stxvx, stxv, xscmpexpdp instructions
407 379890 arm: unhandled instruction: 0xEBAD 0x1B05 (sub.w fp, sp, r5, lsl #4)
408 379895 clock_gettime does not execute POST syscall wrapper
409 379925 PPC64, mtffs does not set the FPCC and C bits in the FPSCR correctly
410 379966 WARNING: unhandled amd64-linux syscall: 313 (finit_module)
411 380200 xtree generated callgrind files refer to files without directory name
412 380202 Assertion failure for cache line size (cls == 64) on aarch64.
413 380397 s390x: __GI_strcspn() replacement needed
414 n-i-bz Fix pub_tool_basics.h build issue with g++ 4.4.7.
416 (3.13.0.RC1: 2 June 2017, vex r3386, valgrind r16434)
417 (3.13.0.RC2: 9 June 2017, vex r3389, valgrind r16443)
418 (3.13.0: 14 June 2017, vex r3396, valgrind r16446)
422 Release 3.12.0 (20 October 2016)
423 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
425 3.12.0 is a feature release with many improvements and the usual
426 collection of bug fixes.
428 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
429 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
430 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
431 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
432 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
433 X86/MacOSX 10.11/12, AMD64/MacOSX 10.11/12 and TILEGX/Linux.
435 * ================== PLATFORM CHANGES =================
437 * POWER: Support for ISA 3.0 has been added
439 * mips: support for O32 FPXX ABI has been added.
440 * mips: improved recognition of different processors
441 * mips: determination of page size now done at run time
443 * amd64: Partial support for AMD FMA4 instructions.
445 * arm, arm64: Support for v8 crypto and CRC instructions.
447 * Improvements and robustification of the Solaris port.
449 * Preliminary support for MacOS 10.12 (Sierra) has been added.
451 Whilst 3.12.0 continues to support the 32-bit x86 instruction set, we
452 would prefer users to migrate to 64-bit x86 (a.k.a amd64 or x86_64)
453 where possible. Valgrind's support for 32-bit x86 has stagnated in
454 recent years and has fallen far behind that for 64-bit x86
455 instructions. By contrast 64-bit x86 is well supported, up to and
458 * ==================== TOOL CHANGES ====================
462 - Added meta mempool support for describing a custom allocator which:
463 - Auto-frees all chunks assuming that destroying a pool destroys all
465 - Uses itself to allocate other memory blocks
467 - New flag --ignore-range-below-sp to ignore memory accesses below
468 the stack pointer, if you really have to. The related flag
469 --workaround-gcc296-bugs=yes is now deprecated. Use
470 --ignore-range-below-sp=1024-1 as a replacement.
474 - Improved thread startup time significantly on non-Linux platforms.
478 - Added collection of the metric "tot-blocks-allocd"
480 * ==================== OTHER CHANGES ====================
482 * Replacement/wrapping of malloc/new related functions is now done not just
483 for system libraries by default, but for any globally defined malloc/new
484 related function (both in shared libraries and statically linked alternative
485 malloc implementations). The dynamic (runtime) linker is excluded, though.
486 To only intercept malloc/new related functions in
487 system libraries use --soname-synonyms=somalloc=nouserintercepts (where
488 "nouserintercepts" can be any non-existing library name).
489 This new functionality is not implemented for MacOS X.
491 * The maximum number of callers in a suppression entry is now equal to
492 the maximum size for --num-callers (500).
493 Note that --gen-suppressions=yes|all similarly generates suppressions
494 containing up to --num-callers frames.
496 * New and modified GDB server monitor features:
498 - Valgrind's gdbserver now accepts the command 'catch syscall'.
499 Note that you must have GDB >= 7.11 to use 'catch syscall' with
502 * New option --run-cxx-freeres=<yes|no> can be used to change whether
503 __gnu_cxx::__freeres() cleanup function is called or not. Default is
506 * Valgrind is able to read compressed debuginfo sections in two formats:
507 - zlib ELF gABI format with SHF_COMPRESSED flag (gcc option -gz=zlib)
508 - zlib GNU format with .zdebug sections (gcc option -gz=zlib-gnu)
510 * Modest JIT-cost improvements: the cost of instrumenting code blocks
511 for the most common use case (x86_64-linux, Memcheck) has been
514 * Improved performance for programs that do a lot of discarding of
515 instruction address ranges of 8KB or less.
517 * The C++ symbol demangler has been updated.
519 * More robustness against invalid syscall parameters on Linux.
521 * ==================== FIXED BUGS ====================
523 The following bugs have been fixed or resolved. Note that "n-i-bz"
524 stands for "not in bugzilla" -- that is, a bug that was reported to us
525 but never got a bugzilla entry. We encourage you to file bugs in
526 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
527 than mailing the developers (or mailing lists) directly -- bugs that
528 are not entered into bugzilla tend to get forgotten about or ignored.
530 To see details of a given bug, visit
531 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
532 where XXXXXX is the bug number as listed below.
534 191069 Exiting due to signal not reported in XML output
535 199468 Suppressions: stack size limited to 25
536 while --num-callers allows more frames
537 212352 vex amd64 unhandled opc_aux = 0x 2, first_opcode == 0xDC (FCOM)
538 278744 cvtps2pd with redundant RexW
539 303877 valgrind doesn't support compressed debuginfo sections.
540 345307 Warning about "still reachable" memory when using libstdc++ from gcc 5
541 348345 Assertion fails for negative lineno
542 348924 MIPS: Load doubles through memory so the code compiles with the FPXX ABI
543 351282 V 3.10.1 MIPS softfloat build broken with GCC 4.9.3 / binutils 2.25.1
544 351692 Dumps created by valgrind are not readable by gdb (mips32 specific)
545 351804 Crash on generating suppressions for "printf" call on OS X 10.10
546 352197 mips: mmap2() not wrapped correctly for page size > 4096
547 353083 arm64 doesn't implement various xattr system calls
548 353084 arm64 doesn't support sigpending system call
549 353137 www: update info for Supported Platforms
550 353138 www: update "The Valgrind Developers" page
551 353370 don't advertise RDRAND in cpuid for Core-i7-4910-like avx2 machine
554 353384 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x62 (pcmpXstrX $0x62)
555 353398 WARNING: unhandled amd64-solaris syscall: 207
556 353660 XML in auxwhat tag not escaping reserved symbols properly
557 353680 s390x: Crash with certain glibc versions due to non-implemented TBEGIN
558 353727 amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x72 (pcmpXstrX $0x72)
559 353802 ELF debug info reader confused with multiple .rodata sections
560 353891 Assert 'bad_scanned_addr < VG_ROUNDDN(start+len, sizeof(Addr))' failed
561 353917 unhandled amd64-solaris syscall fchdir(120)
562 353920 unhandled amd64-solaris syscall: 170
563 354274 arm: unhandled instruction: 0xEBAD 0x0AC1 (sub.w sl, sp, r1, lsl #3)
564 354392 unhandled amd64-solaris syscall: 171
565 354797 Vbit test does not include Iops for Power 8 instruction support
566 354883 tst->os_state.pthread - magic_delta assertion failure on OSX 10.11
570 354933 Fix documentation of --kernel-variant=android-no-hw-tls option
571 355188 valgrind should intercept all malloc related global functions
572 355454 do not intercept malloc related symbols from the runtime linker
573 355455 stderr.exp of test cases wrapmalloc and wrapmallocstatic overconstrained
574 356044 Dwarf line info reader misinterprets is_stmt register
575 356112 mips: replace addi with addiu
576 356393 valgrind (vex) crashes because isZeroU happened
579 356676 arm64-linux: unhandled syscalls 125, 126 (sched_get_priority_max/min)
580 356678 arm64-linux: unhandled syscall 232 (mincore)
581 356817 valgrind.h triggers compiler errors on MSVC when defining NVALGRIND
582 356823 Unsupported ARM instruction: stlex
583 357059 x86/amd64: SSE cvtpi2ps with memory source does transition to MMX state
584 357338 Unhandled instruction for SHA instructions libcrypto Boring SSL
585 357673 crash if I try to run valgrind with a binary link with libcurl
586 357833 Setting RLIMIT_DATA to zero breaks with linux 4.5+
587 357871 pthread_spin_destroy not properly wrapped
588 357887 Calls to VG_(fclose) do not close the file descriptor
589 357932 amd64->IR: accept redundant REX prefixes for {minsd,maxsd} m128, xmm.
590 358030 support direct socket calls on x86 32bit (new in linux 4.3)
591 358478 drd/tests/std_thread.cpp doesn't build with GCC6
592 359133 Assertion 'eltSzB <= ddpa->poolSzB' failed
593 359181 Buffer Overflow during Demangling
594 359201 futex syscall "skips" argument 5 if op is FUTEX_WAIT_BITSET
595 359289 s390x: popcnt (B9E1) not implemented
596 359472 The Power PC vsubuqm instruction doesn't always give the correct result
597 359503 Add missing syscalls for aarch64 (arm64)
598 359645 "You need libc6-dbg" help message could be more helpful
599 359703 s390: wire up separate socketcalls system calls
600 359724 getsockname might crash - deref_UInt should call safe_to_deref
601 359733 amd64 implement ld.so strchr/index override like x86
602 359767 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 1/5
603 359829 Power PC test suite none/tests/ppc64/test_isa_2_07.c uses
605 359838 arm64: Unhandled instruction 0xD5033F5F (clrex)
606 359871 Incorrect mask handling in ppoll
607 359952 Unrecognised PCMPESTRM variants (0x70, 0x19)
608 360008 Contents of Power vr registers contents is not printed correctly when
609 the --vgdb-shadow-registers=yes option is used
610 360035 POWER PC instruction bcdadd and bcdsubtract generate result with
612 360378 arm64: Unhandled instruction 0x5E280844 (sha1h s4, s2)
613 360425 arm64 unsupported instruction ldpsw
615 360519 none/tests/arm64/memory.vgtest might fail with newer gcc
616 360571 Error about the Android Runtime reading below the stack pointer on ARM
617 360574 Wrong parameter type for an ashmem ioctl() call on Android and ARM64
618 360749 kludge for multiple .rodata sections on Solaris no longer needed
619 360752 raise the number of reserved fds in m_main.c from 10 to 12
620 361207 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 2/5
621 361226 s390x: risbgn (EC59) not implemented
622 361253 [s390x] ex_clone.c:42: undefined reference to `pthread_create'
623 361354 ppc64[le]: wire up separate socketcalls system calls
624 361615 Inconsistent termination for multithreaded process terminated by signal
625 361926 Unhandled Solaris syscall: sysfs(84)
626 362009 V dumps core on unimplemented functionality before threads are created
627 362329 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 3/5
628 362894 missing (broken) support for wbit field on mtfsfi instruction (ppc64)
629 362935 [AsusWRT] Assertion 'sizeof(TTEntryC) <= 88' failed
630 362953 Request for an update to the Valgrind Developers page
631 363680 add renameat2() support
632 363705 arm64 missing syscall name_to_handle_at and open_by_handle_at
633 363714 ppc64 missing syscalls sync, waitid and name_to/open_by_handle_at
634 363858 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 4/5
635 364058 clarify in manual limitations of array overruns detections
636 364413 pselect sycallwrapper mishandles NULL sigmask
637 364728 Power PC, missing support for several HW registers in
638 get_otrack_shadow_offset_wrk()
639 364948 Valgrind does not support the IBM POWER ISA 3.0 instructions, part 5/5
640 365273 Invalid write to stack location reported after signal handler runs
641 365912 ppc64BE segfault during jm-insns test (RELRO)
642 366079 FPXX Support for MIPS32 Valgrind
643 366138 Fix configure errors out when using Xcode 8 (clang 8.0.0)
644 366344 Multiple unhandled instruction for Aarch64
645 (0x0EE0E020, 0x1AC15800, 0x4E284801, 0x5E040023, 0x5E056060)
646 367995 Integration of memcheck with custom memory allocator
647 368120 x86_linux asm _start functions do not keep 16-byte aligned stack pointer
648 368412 False positive result for altivec capability check
649 368416 Add tc06_two_races_xml.exp output for ppc64
650 368419 Perf Events ioctls not implemented
651 368461 mmapunmap test fails on ppc64
652 368823 run_a_thread_NORETURN assembly code typo for VGP_arm64_linux target
653 369000 AMD64 fma4 instructions unsupported.
654 369169 ppc64 fails jm_int_isa_2_07 test
655 369175 jm_vec_isa_2_07 test crashes on ppc64
656 369209 valgrind loops and eats up all memory if cwd doesn't exist.
657 369356 pre_mem_read_sockaddr syscall wrapper can crash with bad sockaddr
658 369359 msghdr_foreachfield can crash when handling bad iovec
659 369360 Bad sigprocmask old or new sets can crash valgrind
660 369361 vmsplice syscall wrapper crashes on bad iovec
661 369362 Bad sigaction arguments crash valgrind
662 369383 x86 sys_modify_ldt wrapper crashes on bad ptr
663 369402 Bad set/get_thread_area pointer crashes valgrind
664 369441 bad lvec argument crashes process_vm_readv/writev syscall wrappers
665 369446 valgrind crashes on unknown fcntl command
666 369439 S390x: Unhandled insns RISBLG/RISBHG and LDE/LDER
667 369468 Remove quadratic metapool algorithm using VG_(HT_remove_at_Iter)
668 370265 ISA 3.0 HW cap stuff needs updating
669 371128 BCD add and subtract instructions on Power BE in 32-bit mode do not work
670 372195 Power PC, xxsel instruction is not always recognized
672 n-i-bz Fix incorrect (or infinite loop) unwind on RHEL7 x86 and amd64
673 n-i-bz massif --pages-as-heap=yes does not report peak caused by mmap+munmap
674 n-i-bz false positive leaks due to aspacemgr merging heap & non heap segments
675 n-i-bz Fix ppoll_alarm exclusion on OS X
676 n-i-bz Document brk segment limitation, reference manual in limit reached msg.
677 n-i-bz Fix clobber list in none/tests/amd64/xacq_xrel.c [valgrind r15737]
678 n-i-bz Bump allowed shift value for "add.w reg, sp, reg, lsl #N" [vex r3206]
679 n-i-bz amd64: memcheck false positive with shr %edx
680 n-i-bz arm3: Allow early writeback of SP base register in "strd rD, [sp, #-16]"
681 n-i-bz ppc: Fix two cases of PPCAvFpOp vs PPCFpOp enum confusion
682 n-i-bz arm: Fix incorrect register-number constraint check for LDAEX{,B,H,D}
683 n-i-bz DHAT: added collection of the metric "tot-blocks-allocd"
685 (3.12.0.RC1: 20 October 2016, vex r3282, valgrind r16094)
686 (3.12.0.RC2: 20 October 2016, vex r3282, valgrind r16096)
687 (3.12.0: 21 October 2016, vex r3282, valgrind r16098)
691 Release 3.11.0 (22 September 2015)
692 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
694 3.11.0 is a feature release with many improvements and the usual
695 collection of bug fixes.
697 This release supports X86/Linux, AMD64/Linux, ARM32/Linux,
698 ARM64/Linux, PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux,
699 MIPS32/Linux, MIPS64/Linux, ARM/Android, ARM64/Android,
700 MIPS32/Android, X86/Android, X86/Solaris, AMD64/Solaris, X86/MacOSX
701 10.10 and AMD64/MacOSX 10.10. There is also preliminary support for
702 X86/MacOSX 10.11, AMD64/MacOSX 10.11 and TILEGX/Linux.
704 * ================== PLATFORM CHANGES =================
706 * Support for Solaris/x86 and Solaris/amd64 has been added.
708 * Preliminary support for Mac OS X 10.11 (El Capitan) has been added.
710 * Preliminary support for the Tilera TileGX architecture has been added.
712 * s390x: It is now required for the host to have the "long displacement"
713 facility. The oldest supported machine model is z990.
715 * x86: on an SSE2 only host, Valgrind in 32 bit mode now claims to be a
716 Pentium 4. 3.10.1 wrongly claimed to be a Core 2, which is SSSE3.
718 * The JIT's register allocator is significantly faster, making the JIT
719 as a whole somewhat faster, so JIT-intensive activities, for example
720 program startup, are modestly faster, around 5%.
722 * There have been changes to the default settings of several command
723 line flags, as detailed below.
725 * Intel AVX2 support is more complete (64 bit targets only). On AVX2
726 capable hosts, the simulated CPUID will now indicate AVX2 support.
728 * ==================== TOOL CHANGES ====================
732 - The default value for --leak-check-heuristics has been changed from
733 "none" to "all". This helps to reduce the number of possibly
734 lost blocks, in particular for C++ applications.
736 - The default value for --keep-stacktraces has been changed from
737 "malloc-then-free" to "malloc-and-free". This has a small cost in
738 memory (one word per malloc-ed block) but allows Memcheck to show the
739 3 stacktraces of a dangling reference: where the block was allocated,
740 where it was freed, and where it is acccessed after being freed.
742 - The default value for --partial-loads-ok has been changed from "no" to
743 "yes", so as to avoid false positive errors resulting from some kinds
746 - A new monitor command 'xb <addr> <len>' shows the validity bits of
747 <len> bytes at <addr>. The monitor command 'xb' is easier to use
748 than get_vbits when you need to associate byte data value with
749 their corresponding validity bits.
751 - The 'block_list' monitor command has been enhanced:
752 o it can print a range of loss records
753 o it now accepts an optional argument 'limited <max_blocks>'
754 to control the number of blocks printed.
755 o if a block has been found using a heuristic, then
756 'block_list' now shows the heuristic after the block size.
757 o the loss records/blocks to print can be limited to the blocks
758 found via specified heuristics.
760 - The C helper functions used to instrument loads on
761 x86-{linux,solaris} and arm-linux (both 32-bit only) have been
762 replaced by handwritten assembly sequences. This gives speedups
763 in the region of 0% to 7% for those targets only.
765 - A new command line option, --expensive-definedness-checks=yes|no,
766 has been added. This is useful for avoiding occasional invalid
767 uninitialised-value errors in optimised code. Watch out for
768 runtime degradation, as this can be up to 25%. As always, though,
769 the slowdown is highly application specific. The default setting
774 - A new monitor command 'all_snapshots <filename>' dumps all
775 snapshots taken so far.
779 - Significant memory reduction and moderate speedups for
780 --history-level=full for applications accessing a lot of memory
781 with many different stacktraces.
783 - The default value for --conflict-cache-size=N has been doubled to
784 2000000. Users that were not using the default value should
785 preferably also double the value they give.
787 The default was changed due to the changes in the "full history"
788 implementation. Doubling the value gives on average a slightly more
789 complete history and uses similar memory (or significantly less memory
790 in the worst case) than the previous implementation.
792 - The Helgrind monitor command 'info locks' now accepts an optional
793 argument 'lock_addr', which shows information about the lock at the
796 - When using --history-level=full, the new Helgrind monitor command
797 'accesshistory <addr> [<len>]' will show the recorded accesses for
798 <len> (or 1) bytes at <addr>.
800 * ==================== OTHER CHANGES ====================
802 * The default value for the --smc-check option has been changed from
803 "stack" to "all-non-file" on targets that provide automatic D-I
804 cache coherence (x86, amd64 and s390x). The result is to provide,
805 by default, transparent support for JIT generated and self-modifying
808 * Mac OS X only: the default value for the --dsymutil option has been
809 changed from "no" to "yes", since any serious usage on Mac OS X
810 always required it to be "yes".
812 * The command line options --db-attach and --db-command have been removed.
813 They were deprecated in 3.10.0.
815 * When a process dies due to a signal, Valgrind now shows the signal
816 and the stacktrace at default verbosity (i.e. verbosity 1).
818 * The address description logic used by Memcheck and Helgrind now
819 describes addresses in anonymous segments, file mmap-ed segments,
820 shared memory segments and the brk data segment.
822 * The new option --error-markers=<begin>,<end> can be used to mark the
823 begin/end of errors in textual output mode, to facilitate
824 searching/extracting errors in output files that mix valgrind errors
827 * The new option --max-threads=<number> can be used to change the number
828 of threads valgrind can handle. The default is 500 threads which
829 should be more than enough for most applications.
831 * The new option --valgrind-stacksize=<number> can be used to change the
832 size of the private thread stacks used by Valgrind. This is useful
833 for reducing memory use or increasing the stack size if Valgrind
834 segfaults due to stack overflow.
836 * The new option --avg-transtab-entry-size=<number> can be used to specify
837 the expected instrumented block size, either to reduce memory use or
838 to avoid excessive retranslation.
840 * Valgrind can be built with Intel's ICC compiler, version 14.0 or later.
842 * New and modified GDB server monitor features:
844 - When a signal is reported in GDB, you can now use the GDB convenience
845 variable $_siginfo to examine detailed signal information.
847 - Valgrind's gdbserver now allows the user to change the signal
848 to deliver to the process. So, use 'signal SIGNAL' to continue execution
849 with SIGNAL instead of the signal reported to GDB. Use 'signal 0' to
850 continue without passing the signal to the process.
852 - With GDB >= 7.10, the command 'target remote'
853 will automatically load the executable file of the process running
854 under Valgrind. This means you do not need to specify the executable
855 file yourself, GDB will discover it itself. See GDB documentation about
856 'qXfer:exec-file:read' packet for more info.
858 * ==================== FIXED BUGS ====================
860 The following bugs have been fixed or resolved. Note that "n-i-bz"
861 stands for "not in bugzilla" -- that is, a bug that was reported to us
862 but never got a bugzilla entry. We encourage you to file bugs in
863 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
864 than mailing the developers (or mailing lists) directly -- bugs that
865 are not entered into bugzilla tend to get forgotten about or ignored.
867 To see details of a given bug, visit
868 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
869 where XXXXXX is the bug number as listed below.
871 116002 VG_(printf): Problems with justification of strings and integers
872 155125 avoid cutting away file:lineno after long function name
873 197259 Unsupported arch_prtctl PR_SET_GS option
874 201152 ppc64: Assertion in ppc32g_dirtyhelper_MFSPR_268_269
875 201216 Fix Valgrind does not support pthread_sigmask() on OS X
876 201435 Fix Darwin: -v does not show kernel version
877 208217 "Warning: noted but unhandled ioctl 0x2000747b" on Mac OS X
878 211256 Fixed an outdated comment regarding the default platform.
879 211529 Incomplete call stacks for code compiled by newer versions of MSVC
880 211926 Avoid compilation warnings in valgrind.h with -pedantic
881 212291 Fix unhandled syscall: unix:132 (mkfifo) on OS X
883 226609 Crediting upstream authors in man page
884 231257 Valgrind omits path when executing script from shebang line
885 254164 OS X task_info: UNKNOWN task message [id 3405, to mach_task_self() [..]
886 294065 Improve the pdb file reader by avoiding hardwired absolute pathnames
887 269360 s390x: Fix addressing mode selection for compare-and-swap
888 302630 Memcheck: Assertion failed: 'sizeof(UWord) == sizeof(UInt)'
890 312989 ioctl handling needs to do POST handling on generic ioctls and [..]
891 319274 Fix unhandled syscall: unix:410 (sigsuspend_nocancel) on OS X
892 324181 mmap does not handle MAP_32BIT (handle it now, rather than fail it)
893 327745 Fix valgrind 3.9.0 build fails on Mac OS X 10.6.8
894 330147 libmpiwrap PMPI_Get_count returns undefined value
895 333051 mmap of huge pages fails due to incorrect alignment
897 334802 valgrind does not always explain why a given option is bad
898 335618 mov.w rN, pc/sp (ARM32)
899 335785 amd64->IR 0xC4 0xE2 0x75 0x2F (vmaskmovpd)
904 335907 segfault when running wine's ddrawex/tests/surface.c under valgrind
905 338602 AVX2 bit in CPUID missing
906 338606 Strange message for scripts with invalid interpreter
907 338731 ppc: Fix testuite build for toolchains not supporting -maltivec
908 338995 shmat with hugepages (SHM_HUGETLB) fails with EINVAL
909 339045 Getting valgrind to compile and run on OS X Yosemite (10.10)
911 339156 gdbsrv not called for fatal signal
912 339215 Valgrind 3.10.0 contain 2013 in copyrights notice
913 339288 support Cavium Octeon MIPS specific BBIT*32 instructions
914 339636 Use fxsave64 and fxrstor64 mnemonics instead of old-school rex64 prefix
915 339442 Fix testsuite build failure on OS X 10.9
916 339542 Enable compilation with Intel's ICC compiler
917 339563 The DVB demux DMX_STOP ioctl doesn't have a wrapper
918 339688 Mac-specific ASM does not support .version directive (cpuid,
919 tronical and pushfpopf tests)
920 339745 Valgrind crash when check Marmalade app (partial fix)
921 339755 Fix known deliberate memory leak in setenv() on Mac OS X 10.9
922 339778 Linux/TileGx platform support for Valgrind
923 339780 Fix known uninitialised read in pthread_rwlock_init() on Mac OS X 10.9
924 339789 Fix none/tests/execve test on Mac OS X 10.9
925 339808 Fix none/tests/rlimit64_nofile test on Mac OS X 10.9
926 339820 vex amd64->IR: 0x66 0xF 0x3A 0x63 0xA 0x42 0x74 0x9 (pcmpistri $0x42)
927 340115 Fix none/tests/cmdline[1|2] tests on systems which define TMPDIR
928 340392 Allow user to select more accurate definedness checking in memcheck
929 to avoid invalid complaints on optimised code
930 340430 Fix some grammatical weirdness in the manual.
931 341238 Recognize GCC5/DWARFv5 DW_LANG constants (Go, C11, C++11, C++14)
932 341419 Signal handler ucontext_t not filled out correctly on OS X
933 341539 VG_(describe_addr) should not describe address as belonging to client
934 segment if it is past the heap end
935 341613 Enable building of manythreads and thread-exits tests on Mac OS X
936 341615 Fix none/tests/darwin/access_extended test on Mac OS X
937 341698 Valgrind's AESKEYGENASSIST gives wrong result in words 0 and 2 [..]
938 341789 aarch64: shmat fails with valgrind on ARMv8
939 341997 MIPS64: Cavium OCTEON insns - immediate operand handled incorrectly
940 342008 valgrind.h needs type cast [..] for clang/llvm in 64-bit mode
941 342038 Unhandled syscalls on aarch64 (mbind/get/set_mempolicy)
942 342063 wrong format specifier for test mcblocklistsearch in gdbserver_tests
943 342117 Hang when loading PDB file for MSVC compiled Firefox under Wine
944 342221 socket connect false positive uninit memory for unknown af family
945 342353 Allow dumping full massif output while valgrind is still running
946 342571 Valgrind chokes on AVX compare intrinsic with _CMP_GE_QS
950 342603 Add I2C_SMBUS ioctl support
951 342635 OS X 10.10 (Yosemite) - missing system calls and fcntl code
952 342683 Mark memory past the initial brk limit as unaddressable
953 342783 arm: unhandled instruction 0xEEFE1ACA = "vcvt.s32.f32 s3, s3, #12"
954 342795 Internal glibc __GI_mempcpy call should be intercepted
955 342841 s390x: Support instructions fiebr(a) and fidbr(a)
956 343012 Unhandled syscall 319 (memfd_create)
957 343069 Patch updating v4l2 API support
958 343173 helgrind crash during stack unwind
959 343219 fix GET_STARTREGS for arm
960 343303 Fix known deliberate memory leak in setenv() on Mac OS X 10.10
961 343306 OS X 10.10: UNKNOWN mach_msg unhandled MACH_SEND_TRAILER option
962 343332 Unhandled instruction 0x9E310021 (fcvtmu) on aarch64
963 343335 unhandled instruction 0x1E638400 (fccmp) aarch64
964 343523 OS X mach_ports_register: UNKNOWN task message [id 3403, to [..]
965 343525 OS X host_get_special_port: UNKNOWN host message [id 412, to [..]
966 343597 ppc64le: incorrect use of offseof macro
967 343649 OS X host_create_mach_voucher: UNKNOWN host message [id 222, to [..]
968 343663 OS X 10.10 Memchecj always reports a leak regardless of [..]
969 343732 Unhandled syscall 144 (setgid) on aarch64
970 343733 Unhandled syscall 187 (msgctl and related) on aarch64
971 343802 s390x: False positive "conditional jump or move depends on [..]
972 343902 --vgdb=yes doesn't break when --xml=yes is used
973 343967 Don't warn about setuid/setgid/setcap executable for directories
974 343978 Recognize DWARF5/GCC5 DW_LANG_Fortran 2003 and 2008 constants
975 344007 accept4 syscall unhandled on arm64 (242) and ppc64 (344)
976 344033 Helgrind on ARM32 loses track of mutex state in pthread_cond_wait
977 344054 www - update info for Solaris/illumos
978 344416 'make regtest' does not work cleanly on OS X
979 344235 Remove duplicate include of pub_core_aspacemgr.h
980 344279 syscall sendmmsg on arm64 (269) and ppc32/64 (349) unhandled
981 344295 syscall recvmmsg on arm64 (243) and ppc32/64 (343) unhandled
982 344307 2 unhandled syscalls on aarch64/arm64: umount2(39), mount (40)
983 344314 callgrind_annotate ... warnings about commands containing newlines
984 344318 socketcall should wrap recvmmsg and sendmmsg
985 344337 Fix unhandled syscall: mach:41 (_kernelrpc_mach_port_guard_trap)
986 344416 Fix 'make regtest' does not work cleanly on OS X
987 344499 Fix compilation for Linux kernel >= 4.0.0
988 344512 OS X: unhandled syscall: unix:348 (__pthread_chdir),
989 unix:349 (__pthread_fchdir)
990 344559 Garbage collection of unused segment names in address space manager
991 344560 Fix stack traces missing penultimate frame on OS X
992 344621 Fix memcheck/tests/err_disable4 test on OS X
993 344686 Fix suppression for pthread_rwlock_init on OS X 10.10
994 344702 Fix missing libobjc suppressions on OS X 10.10
996 344936 Fix unhandled syscall: unix:473 (readlinkat) on OS X 10.10
997 344939 Fix memcheck/tests/xml1 on OS X 10.10
998 345016 helgrind/tests/locked_vs_unlocked2 is failing sometimes
999 345079 Fix build problems in VEX/useful/test_main.c
1000 345126 Incorrect handling of VIDIOC_G_AUDIO and G_AUDOUT
1001 345177 arm64: prfm (reg) not implemented
1002 345215 Performance improvements for the register allocator
1003 345248 add support for Solaris OS in valgrind
1004 345338 TIOCGSERIAL and TIOCSSERIAL ioctl support on Linux
1005 345394 Fix memcheck/tests/strchr on OS X
1006 345637 Fix memcheck/tests/sendmsg on OS X
1007 345695 Add POWERPC support for AT_DCACHESIZE and HWCAP2
1008 345824 Fix aspacem segment mismatch: seen with none/tests/bigcode
1009 345887 Fix an assertion in the address space manager
1010 345928 amd64: callstack only contains current function for small stacks
1011 345984 disInstr(arm): unhandled instruction: 0xEE193F1E
1012 345987 MIPS64: Implement cavium LHX instruction
1013 346031 MIPS: Implement support for the CvmCount register (rhwr %0, 31)
1014 346185 Fix typo saving altivec register v24
1015 346267 Compiler warnings for PPC64 code on call to LibVEX_GuestPPC64_get_XER()
1016 and LibVEX_GuestPPC64_get_CR()
1017 346270 Regression tests none/tests/jm_vec/isa_2_07 and
1018 none/tests/test_isa_2_07_part2 have failures on PPC64 little endian
1019 346307 fuse filesystem syscall deadlocks
1020 346324 PPC64 missing support for lbarx, lharx, stbcx and sthcx instructions
1021 346411 MIPS: SysRes::_valEx handling is incorrect
1022 346416 Add support for LL_IOC_PATH2FID and LL_IOC_GETPARENT Lustre ioctls
1023 346474 PPC64 Power 8, spr TEXASRU register not supported
1024 346487 Compiler generates "note" about a future ABI change for PPC64
1025 346562 MIPS64: lwl/lwr instructions are performing 64bit loads
1026 and causing spurious "invalid read of size 8" warnings
1027 346801 Fix link error on OS X: _vgModuleLocal_sf_maybe_extend_stack
1028 347151 Fix suppression for pthread_rwlock_init on OS X 10.8
1029 347233 Fix memcheck/tests/strchr on OS X 10.10 (Haswell)
1030 347322 Power PC regression test cleanup
1031 347379 valgrind --leak-check=full leak errors from system libs on OS X 10.8
1033 347389 unhandled syscall: 373 (Linux ARM syncfs)
1034 347686 Patch set to cleanup PPC64 regtests
1035 347978 Remove bash dependencies where not needed
1036 347982 OS X: undefined symbols for architecture x86_64: "_global" [..]
1037 347988 Memcheck: the 'impossible' happened: unexpected size for Addr (OSX/wine)
1039 348102 Patch updating v4l2 API support
1040 348247 amd64 front end: jno jumps wrongly when overflow is not set
1041 348269 Improve mmap MAP_HUGETLB support.
1042 348334 (ppc) valgrind does not simulate dcbfl - then my program terminates
1043 348345 Assertion fails for negative lineno
1044 348377 Unsupported ARM instruction: yield
1045 348565 Fix detection of command line option availability for clang
1046 348574 vex amd64->IR pcmpistri SSE4.2 unsupported (pcmpistri $0x18)
1047 348728 Fix broken check for VIDIOC_G_ENC_INDEX
1048 348748 Fix redundant condition
1049 348890 Fix clang warning about unsupported --param inline-unit-growth=900
1050 348949 Bogus "ERROR: --ignore-ranges: suspiciously large range"
1051 349034 Add Lustre ioctls LL_IOC_GROUP_LOCK and LL_IOC_GROUP_UNLOCK
1052 349086 Fix UNKNOWN task message [id 3406, to mach_task_self(), [..]
1053 349087 Fix UNKNOWN task message [id 3410, to mach_task_self(), [..]
1054 349626 Implemented additional Xen hypercalls
1055 349769 Clang/osx: ld: warning: -read_only_relocs cannot be used with x86_64
1056 349790 Clean up of the hardware capability checking utilities.
1057 349828 memcpy intercepts memmove causing src/dst overlap error (ppc64 ld.so)
1058 349874 Fix typos in source code
1059 349879 memcheck: add handwritten assembly for helperc_LOADV*
1060 349941 di_notify_mmap might create wrong start/size DebugInfoMapping
1061 350062 vex x86->IR: 0x66 0xF 0x3A 0xB (ROUNDSD) on OS X
1062 350202 Add limited param to 'monitor block_list'
1063 350290 s390x: Support instructions fixbr(a)
1064 350359 memcheck/tests/x86/fxsave hangs indefinetely on OS X
1065 350809 Fix none/tests/async-sigs for Solaris
1066 350811 Remove reference to --db-attach which has been removed.
1067 350813 Memcheck/x86: enable handwritten assembly helpers for x86/Solaris too
1068 350854 hard-to-understand code in VG_(load_ELF)()
1069 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1070 351386 Solaris: Cannot run ld.so.1 under Valgrind
1071 351474 Fix VG_(iseqsigset) as obvious
1072 351531 Typo in /include/vki/vki-xen-physdev.h header guard
1073 351756 Intercept platform_memchr$VARIANT$Haswell on OS X
1074 351858 ldsoexec support on Solaris
1075 351873 Newer gcc doesn't allow __builtin_tabortdc[i] in ppc32 mode
1076 352130 helgrind reports false races for printfs using mempcpy on FILE* state
1077 352284 s390: Conditional jump depends on uninitialised value(s) in vfprintf
1078 352320 arm64 crash on none/tests/nestedfs
1079 352765 Vbit test fails on Power 6
1080 352768 The mbar instruction is missing from the Power PC support
1081 352769 Power PC program priority register (PPR) is not supported
1082 n-i-bz Provide implementations of certain compiler builtins to support
1083 compilers that may not provide those
1084 n-i-bz Old STABS code is still being compiled, but never used. Remove it.
1085 n-i-bz Fix compilation on distros with glibc < 2.5
1086 n-i-bz (vex 3098) Avoid generation of Neon insns on non-Neon hosts
1087 n-i-bz Enable rt_sigpending syscall on ppc64 linux.
1088 n-i-bz mremap did not work properly on shared memory
1089 n-i-bz Fix incorrect sizeof expression in syswrap-xen.c reported by Coverity
1090 n-i-bz In VALGRIND_PRINTF write out thread name, if any, to xml
1092 (3.11.0.TEST1: 8 September 2015, vex r3187, valgrind r15646)
1093 (3.11.0.TEST2: 21 September 2015, vex r3193, valgrind r15667)
1094 (3.11.0: 22 September 2015, vex r3195, valgrind r15674)
1098 Release 3.10.1 (25 November 2014)
1099 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1100 3.10.1 is a bug fix release. It fixes various bugs reported in 3.10.0
1101 and backports fixes for all reported missing AArch64 ARMv8 instructions
1102 and syscalls from the trunk. If you package or deliver 3.10.0 for others
1103 to use, you might want to consider upgrading to 3.10.1 instead.
1105 The following bugs have been fixed or resolved. Note that "n-i-bz"
1106 stands for "not in bugzilla" -- that is, a bug that was reported to us
1107 but never got a bugzilla entry. We encourage you to file bugs in
1108 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1109 than mailing the developers (or mailing lists) directly -- bugs that
1110 are not entered into bugzilla tend to get forgotten about or ignored.
1112 To see details of a given bug, visit
1113 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1114 where XXXXXX is the bug number as listed below.
1116 335440 arm64: ld1 (single structure) is not implemented
1117 335713 arm64: unhanded instruction: prfm (immediate)
1118 339020 ppc64: memcheck/tests/ppc64/power_ISA2_05 failing in nightly build
1119 339182 ppc64: AvSplat ought to load destination vector register with [..]
1120 339336 PPC64 store quad instruction (stq) is not supposed to change [..]
1121 339433 ppc64 lxvw4x instruction uses four 32-byte loads
1122 339645 Use correct tag names in sys_getdents/64 wrappers
1123 339706 Fix false positive for ioctl(TIOCSIG) on linux
1124 339721 assertion 'check_sibling == sibling' failed in readdwarf3.c ...
1125 339853 arm64 times syscall unknown
1126 339855 arm64 unhandled getsid/setsid syscalls
1127 339858 arm64 dmb sy not implemented
1128 339926 Unhandled instruction 0x1E674001 (frintx) on aarm64
1129 339927 Unhandled instruction 0x9E7100C6 (fcvtmu) on aarch64
1130 339938 disInstr(arm64): unhandled instruction 0x4F8010A4 (fmla)
1132 339940 arm64: unhandled syscall: 83 (sys_fdatasync) + patch
1133 340033 arm64: unhandled insn dmb ishld and some other isb-dmb-dsb variants
1134 340028 unhandled syscalls for arm64 (msync, pread64, setreuid and setregid)
1135 340036 arm64: Unhandled instruction ld4 (multiple structures, no offset)
1136 340236 arm64: unhandled syscalls: mknodat, fchdir, chroot, fchownat
1137 340509 arm64: unhandled instruction fcvtas
1138 340630 arm64: fchmod (52) and fchown (55) syscalls not recognized
1139 340632 arm64: unhandled instruction fcvtas
1140 340722 Resolve "UNKNOWN attrlist flags 0:0x10000000"
1141 340725 AVX2: Incorrect decoding of vpbroadcast{b,w} reg,reg forms
1142 340788 warning: unhandled syscall: 318 (getrandom)
1143 340807 disInstr(arm): unhandled instruction: 0xEE989B20
1144 340856 disInstr(arm64): unhandled instruction 0x1E634C45 (fcsel)
1145 340922 arm64: unhandled getgroups/setgroups syscalls
1146 350251 Fix typo in VEX utility program (test_main.c).
1147 350407 arm64: unhandled instruction ucvtf (vector, integer)
1148 350809 none/tests/async-sigs breaks when run under cron on Solaris
1149 350811 update README.solaris after r15445
1150 350813 Use handwritten memcheck assembly helpers on x86/Solaris [..]
1151 350854 strange code in VG_(load_ELF)()
1152 351140 arm64 syscalls setuid (146) and setresgid (149) not implemented
1153 n-i-bz DRD and Helgrind: Handle Imbe_CancelReservation (clrex on ARM)
1154 n-i-bz Add missing ]] to terminate CDATA.
1155 n-i-bz Glibc versions prior to 2.5 do not define PTRACE_GETSIGINFO
1156 n-i-bz Enable sys_fadvise64_64 on arm32.
1157 n-i-bz Add test cases for all remaining AArch64 SIMD, FP and memory insns.
1158 n-i-bz Add test cases for all known arm64 load/store instructions.
1159 n-i-bz PRE(sys_openat): when checking whether ARG1 == VKI_AT_FDCWD [..]
1160 n-i-bz Add detection of old ppc32 magic instructions from bug 278808.
1161 n-i-bz exp-dhat: Implement missing function "dh_malloc_usable_size".
1162 n-i-bz arm64: Implement "fcvtpu w, s".
1163 n-i-bz arm64: implement ADDP and various others
1164 n-i-bz arm64: Implement {S,U}CVTF (scalar, fixedpt).
1165 n-i-bz arm64: enable FCVT{A,N}S X,S.
1167 (3.10.1: 25 November 2014, vex r3026, valgrind r14785)
1171 Release 3.10.0 (10 September 2014)
1172 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1174 3.10.0 is a feature release with many improvements and the usual
1175 collection of bug fixes.
1177 This release supports X86/Linux, AMD64/Linux, ARM32/Linux, ARM64/Linux,
1178 PPC32/Linux, PPC64BE/Linux, PPC64LE/Linux, S390X/Linux, MIPS32/Linux,
1179 MIPS64/Linux, ARM/Android, MIPS32/Android, X86/Android, X86/MacOSX 10.9
1180 and AMD64/MacOSX 10.9. Support for MacOSX 10.8 and 10.9 is
1181 significantly improved relative to the 3.9.0 release.
1183 * ================== PLATFORM CHANGES =================
1185 * Support for the 64-bit ARM Architecture (AArch64 ARMv8). This port
1186 is mostly complete, and is usable, but some SIMD instructions are as
1189 * Support for little-endian variant of the 64-bit POWER architecture.
1191 * Support for Android on MIPS32.
1193 * Support for 64bit FPU on MIPS32 platforms.
1195 * Both 32- and 64-bit executables are supported on MacOSX 10.8 and 10.9.
1197 * Configuration for and running on Android targets has changed.
1198 See README.android in the source tree for details.
1200 * ================== DEPRECATED FEATURES =================
1202 * --db-attach is now deprecated and will be removed in the next
1203 valgrind feature release. The built-in GDB server capabilities are
1204 superior and should be used instead. Learn more here:
1205 http://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.gdbserver
1207 * ==================== TOOL CHANGES ====================
1211 - Client code can now selectively disable and re-enable reporting of
1212 invalid address errors in specific ranges using the new client
1213 requests VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE and
1214 VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE.
1216 - Leak checker: there is a new leak check heuristic called
1217 "length64". This is used to detect interior pointers pointing 8
1218 bytes inside a block, on the assumption that the first 8 bytes
1219 holds the value "block size - 8". This is used by
1220 sqlite3MemMalloc, for example.
1222 - Checking of system call parameters: if a syscall parameter
1223 (e.g. bind struct sockaddr, sendmsg struct msghdr, ...) has
1224 several fields not initialised, an error is now reported for each
1225 field. Previously, an error was reported only for the first
1226 uninitialised field.
1228 - Mismatched alloc/free checking: a new flag
1229 --show-mismatched-frees=no|yes [yes] makes it possible to turn off
1230 such checks if necessary.
1234 - Improvements to error messages:
1236 o Race condition error message involving heap allocated blocks also
1237 show the thread number that allocated the raced-on block.
1239 o All locks referenced by an error message are now announced.
1240 Previously, some error messages only showed the lock addresses.
1242 o The message indicating where a lock was first observed now also
1243 describes the address/location of the lock.
1245 - Helgrind now understands the Ada task termination rules and
1246 creates a happens-before relationship between a terminated task
1247 and its master. This avoids some false positives and avoids a big
1248 memory leak when a lot of Ada tasks are created and terminated.
1249 The interceptions are only activated with forthcoming releases of
1250 gnatpro >= 7.3.0w-20140611 and gcc >= 5.0.
1252 - A new GDB server monitor command "info locks" giving the list of
1253 locks, their location, and their status.
1257 - callgrind_control now supports the --vgdb-prefix argument,
1258 which is needed if valgrind was started with this same argument.
1260 * ==================== OTHER CHANGES ====================
1262 * Unwinding through inlined function calls. Stack unwinding can now
1263 make use of Dwarf3 inlined-unwind information if it is available.
1264 The practical effect is that inlined calls become visible in stack
1265 traces. The suppression matching machinery has been adjusted
1266 accordingly. This is controlled by the new option
1267 --read-inline-info=yes|no. Currently this is enabled by default
1268 only on Linux and Android targets and only for the tools Memcheck,
1271 * Valgrind can now read EXIDX unwind information on 32-bit ARM
1272 targets. If an object contains both CFI and EXIDX unwind
1273 information, Valgrind will prefer the CFI over the EXIDX. This
1274 facilitates unwinding through system libraries on arm-android
1277 * Address description logic has been improved and is now common
1278 between Memcheck and Helgrind, resulting in better address
1279 descriptions for some kinds of error messages.
1281 * Error messages about dubious arguments (eg, to malloc or calloc) are
1282 output like other errors. This means that they can be suppressed
1283 and they have a stack trace.
1285 * The C++ demangler has been updated for better C++11 support.
1287 * New and modified GDB server monitor features:
1289 - Thread local variables/storage (__thread) can now be displayed.
1291 - The GDB server monitor command "v.info location <address>"
1292 displays information about an address. The information produced
1293 depends on the tool and on the options given to valgrind.
1294 Possibly, the following are described: global variables, local
1295 (stack) variables, allocated or freed blocks, ...
1297 - The option "--vgdb-stop-at=event1,event2,..." allows the user to
1298 ask the GDB server to stop at the start of program execution, at
1299 the end of the program execution and on Valgrind internal errors.
1301 - A new monitor command "v.info stats" shows various Valgrind core
1302 and tool statistics.
1304 - A new monitor command "v.set hostvisibility" allows the GDB server
1305 to provide access to Valgrind internal host status/memory.
1307 * A new option "--aspace-minaddr=<address>" can in some situations
1308 allow the use of more memory by decreasing the address above which
1309 Valgrind maps memory. It can also be used to solve address
1310 conflicts with system libraries by increasing the default value.
1311 See user manual for details.
1313 * The amount of memory used by Valgrind to store debug info (unwind
1314 info, line number information and symbol data) has been
1315 significantly reduced, even though Valgrind now reads more
1316 information in order to support unwinding of inlined function calls.
1318 * Dwarf3 handling with --read-var-info=yes has been improved:
1320 - Ada and C struct containing VLAs no longer cause a "bad DIE" error
1322 - Code compiled with
1323 -ffunction-sections -fdata-sections -Wl,--gc-sections
1324 no longer causes assertion failures.
1326 * Improved checking for the --sim-hints= and --kernel-variant=
1327 options. Unknown strings are now detected and reported to the user
1330 * The semantics of stack start/end boundaries in the valgrind.h
1331 VALGRIND_STACK_REGISTER client request has been clarified and
1332 documented. The convention is that start and end are respectively
1333 the lowest and highest addressable bytes of the stack.
1335 * ==================== FIXED BUGS ====================
1337 The following bugs have been fixed or resolved. Note that "n-i-bz"
1338 stands for "not in bugzilla" -- that is, a bug that was reported to us
1339 but never got a bugzilla entry. We encourage you to file bugs in
1340 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1341 than mailing the developers (or mailing lists) directly -- bugs that
1342 are not entered into bugzilla tend to get forgotten about or ignored.
1344 To see details of a given bug, visit
1345 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1346 where XXXXXX is the bug number as listed below.
1348 175819 Support for ipv6 socket reporting with --track-fds
1349 232510 make distcheck fails
1350 249435 Analyzing wine programs with callgrind triggers a crash
1351 278972 support for inlined function calls in stacktraces and suppression
1353 291310 FXSAVE instruction marks memory as undefined on amd64
1354 303536 ioctl for SIOCETHTOOL (ethtool(8)) isn't wrapped
1355 308729 vex x86->IR: unhandled instruction bytes 0xf 0x5 (syscall)
1356 315199 vgcore file for threaded app does not show which thread crashed
1357 315952 tun/tap ioctls are not supported
1358 323178 Unhandled instruction: PLDW register (ARM)
1359 323179 Unhandled instruction: PLDW immediate (ARM)
1360 324050 Helgrind: SEGV because of unaligned stack when using movdqa
1361 325110 Add test-cases for Power ISA 2.06 insns: divdo/divdo. and divduo/divduo.
1362 325124 [MIPSEL] Compilation error
1363 325477 Phase 4 support for IBM Power ISA 2.07
1364 325538 cavium octeon mips64, valgrind reported "dumping core" [...]
1365 325628 Phase 5 support for IBM Power ISA 2.07
1366 325714 Empty vgcore but RLIMIT_CORE is big enough (too big)
1367 325751 Missing the two privileged Power PC Transactional Memory Instructions
1368 325816 Phase 6 support for IBM Power ISA 2.07
1369 325856 Make SGCheck fail gracefully on unsupported platforms
1370 326026 Iop names for count leading zeros/sign bits incorrectly imply [..]
1371 326436 DRD: False positive in libstdc++ std::list::push_back
1372 326444 Cavium MIPS Octeon Specific Load Indexed Instructions
1373 326462 Refactor vgdb to isolate invoker stuff into separate module
1374 326469 amd64->IR: 0x66 0xF 0x3A 0x63 0xC1 0xE (pcmpistri 0x0E)
1375 326623 DRD: false positive conflict report in a field assignment
1376 326724 Valgrind does not compile on OSX 1.9 Mavericks
1377 326816 Intercept for __strncpy_sse2_unaligned missing?
1378 326921 coregrind fails to compile m_trampoline.S with MIPS/Linux port of V
1379 326983 Clear direction flag after tests on amd64.
1380 327212 Do not prepend the current directory to absolute path names.
1381 327223 Support for Cavium MIPS Octeon Atomic and Count Instructions
1382 327238 Callgrind Assertion 'passed <= last_bb->cjmp_count' failed
1383 327284 s390x: Fix translation of the risbg instruction
1384 327639 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x34
1385 327837 dwz compressed alternate .debug_info and .debug_str not read correctly
1386 327916 DW_TAG_typedef may have no name
1387 327943 s390x: add a redirection for the 'index' function
1388 328100 XABORT not implemented
1389 328205 Implement additional Xen hypercalls
1390 328454 add support Backtraces with ARM unwind tables (EXIDX)
1391 328455 s390x: SIGILL after emitting wrong register pair for ldxbr
1392 328711 valgrind.1 manpage "memcheck options" section is badly generated
1393 328878 vex amd64->IR pcmpestri SSE4.2 instruction is unsupported 0x14
1394 329612 Incorrect handling of AT_BASE for image execution
1395 329694 clang warns about using uninitialized variable
1396 329956 valgrind crashes when lmw/stmw instructions are used on ppc64
1397 330228 mmap must align to VKI_SHMLBA on mips32
1398 330257 LLVM does not support `-mno-dynamic-no-pic` option
1399 330319 amd64->IR: unhandled instruction bytes: 0xF 0x1 0xD5 (xend)
1400 330459 --track-fds=yes doesn't track eventfds
1401 330469 Add clock_adjtime syscall support
1402 330594 Missing sysalls on PowerPC / uClibc
1403 330622 Add test to regression suite for POWER instruction: dcbzl
1404 330939 Support for AMD's syscall instruction on x86
1406 330941 Typo in PRE(poll) syscall wrapper
1407 331057 unhandled instruction: 0xEEE01B20 (vfma.f64) (has patch)
1408 331254 Fix expected output for memcheck/tests/dw4
1409 331255 Fix race condition in test none/tests/coolo_sigaction
1410 331257 Fix type of jump buffer in test none/tests/faultstatus
1411 331305 configure uses bash specific syntax
1412 331337 s390x WARNING: unhandled syscall: 326 (dup3)
1413 331380 Syscall param timer_create(evp) points to uninitialised byte(s)
1414 331476 Patch to handle ioctl 0x5422 on Linux (x86 and amd64)
1415 331829 Unexpected ioctl opcode sign extension
1416 331830 ppc64: WARNING: unhandled syscall: 96/97
1417 331839 drd/tests/sem_open specifies invalid semaphore name
1418 331847 outcome of drd/tests/thread_name is nondeterministic
1419 332037 Valgrind cannot handle Thumb "add pc, reg"
1420 332055 drd asserts on platforms with VG_STACK_REDZONE_SZB == 0 and
1421 consistency checks enabled
1422 332263 intercepts for pthread_rwlock_timedrdlock and
1423 pthread_rwlock_timedwrlock are incorrect
1424 332265 drd could do with post-rwlock_init and pre-rwlock_destroy
1426 332276 Implement additional Xen hypercalls
1427 332658 ldrd.w r1, r2, [PC, #imm] does not adjust for 32bit alignment
1428 332765 Fix ms_print to create temporary files in a proper directory
1429 333072 drd: Add semaphore annotations
1430 333145 Tests for missaligned PC+#imm access for arm
1431 333228 AAarch64 Missing instruction encoding: mrs %[reg], ctr_el0
1432 333230 AAarch64 missing instruction encodings: dc, ic, dsb.
1433 333248 WARNING: unhandled syscall: unix:443
1434 333428 ldr.w pc [rD, #imm] instruction leads to assertion
1435 333501 cachegrind: assertion: Cache set count is not a power of two.
1438 333666 Recognize MPX instructions and bnd prefix.
1439 333788 Valgrind does not support the CDROM_DISC_STATUS ioctl (has patch)
1440 333817 Valgrind reports the memory areas written to by the SG_IO
1442 334049 lzcnt fails silently (x86_32)
1443 334384 Valgrind does not have support Little Endian support for
1445 334585 recvmmsg unhandled (+patch) (arm)
1446 334705 sendmsg and recvmsg should guard against bogus msghdr fields.
1447 334727 Build fails with -Werror=format-security
1448 334788 clarify doc about --log-file initial program directory
1449 334834 PPC64 Little Endian support, patch 2
1450 334836 PPC64 Little Endian support, patch 3 testcase fixes
1451 334936 patch to fix false positives on alsa SNDRV_CTL_* ioctls
1452 335034 Unhandled ioctl: HCIGETDEVLIST
1453 335155 vgdb, fix error print statement.
1454 335262 arm64: movi 8bit version is not supported
1455 335263 arm64: dmb instruction is not implemented
1456 335441 unhandled ioctl 0x8905 (SIOCATMARK) when running wine under valgrind
1457 335496 arm64: sbc/abc instructions are not implemented
1458 335554 arm64: unhandled instruction: abs
1459 335564 arm64: unhandled instruction: fcvtpu Xn, Sn
1460 335735 arm64: unhandled instruction: cnt
1461 335736 arm64: unhandled instruction: uaddlv
1462 335848 arm64: unhandled instruction: {s,u}cvtf
1463 335902 arm64: unhandled instruction: sli
1464 335903 arm64: unhandled instruction: umull (vector)
1465 336055 arm64: unhandled instruction: mov (element)
1466 336062 arm64: unhandled instruction: shrn{,2}
1467 336139 mip64: [...] valgrind hangs and spins on a single core [...]
1468 336189 arm64: unhandled Instruction: mvn
1469 336435 Valgrind hangs in pthread_spin_lock consuming 100% CPU
1470 336619 valgrind --read-var-info=yes doesn't handle DW_TAG_restrict_type
1471 336772 Make moans about unknown ioctls more informative
1472 336957 Add a section about the Solaris/illumos port on the webpage
1473 337094 ifunc wrapper is broken on ppc64
1474 337285 fcntl commands F_OFD_SETLK, F_OFD_SETLKW, and F_OFD_GETLK not supported
1475 337528 leak check heuristic for block prefixed by length as 64bit number
1476 337740 Implement additional Xen hypercalls
1477 337762 guest_arm64_toIR.c:4166 (dis_ARM64_load_store): Assertion `0' failed.
1478 337766 arm64-linux: unhandled syscalls mlock (228) and mlockall (230)
1479 337871 deprecate --db-attach
1480 338023 Add support for all V4L2/media ioctls
1481 338024 inlined functions are not shown if DW_AT_ranges is used
1482 338106 Add support for 'kcmp' syscall
1483 338115 DRD: computed conflict set differs from actual after fork
1484 338160 implement display of thread local storage in gdbsrv
1485 338205 configure.ac and check for -Wno-tautological-compare
1486 338300 coredumps are missing one byte of every segment
1487 338445 amd64 vbit-test fails with unknown opcodes used by arm64 VEX
1488 338499 --sim-hints parsing broken due to wrong order in tokens
1489 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7
1490 338681 Unable to unwind through clone thread created on i386-linux
1491 338698 race condition between gdbsrv and vgdb on startup
1492 338703 helgrind on arm-linux gets false positives in dynamic loader
1493 338791 alt dwz files can be relative of debug/main file
1494 338878 on MacOS: assertion 'VG_IS_PAGE_ALIGNED(clstack_end+1)' failed
1495 338932 build V-trunk with gcc-trunk
1496 338974 glibc 2.20 changed size of struct sigaction sa_flags field on s390
1497 345079 Fix build problems in VEX/useful/test_main.c
1498 n-i-bz Fix KVM_CREATE_IRQCHIP ioctl handling
1499 n-i-bz s390x: Fix memory corruption for multithreaded applications
1500 n-i-bz vex arm->IR: allow PC as basereg in some LDRD cases
1501 n-i-bz internal error in Valgrind if vgdb transmit signals when ptrace invoked
1502 n-i-bz Fix mingw64 support in valgrind.h (dev@, 9 May 2014)
1503 n-i-bz drd manual: Document how to C++11 programs that use class "std::thread"
1504 n-i-bz Add command-line option --default-suppressions
1505 n-i-bz Add support for BLKDISCARDZEROES ioctl
1506 n-i-bz ppc32/64: fix a regression with the mtfsb0/mtfsb1 instructions
1507 n-i-bz Add support for sys_pivot_root and sys_unshare
1509 (3.10.0.BETA1: 2 September 2014, vex r2940, valgrind r14428)
1510 (3.10.0.BETA2: 8 September 2014, vex r2950, valgrind r14503)
1511 (3.10.0: 10 September 2014, vex r2950, valgrind r14514)
1515 Release 3.9.0 (31 October 2013)
1516 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1517 3.9.0 is a feature release with many improvements and the usual
1518 collection of bug fixes.
1520 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
1521 PPC64/Linux, S390X/Linux, MIPS32/Linux, MIPS64/Linux, ARM/Android,
1522 X86/Android, X86/MacOSX 10.7 and AMD64/MacOSX 10.7. Support for
1523 MacOSX 10.8 is significantly improved relative to the 3.8.0 release.
1525 * ================== PLATFORM CHANGES =================
1527 * Support for MIPS64 LE and BE running Linux. Valgrind has been
1528 tested on MIPS64 Debian Squeeze and Debian Wheezy distributions.
1530 * Support for MIPS DSP ASE on MIPS32 platforms.
1532 * Support for s390x Decimal Floating Point instructions on hosts that
1533 have the DFP facility installed.
1535 * Support for POWER8 (Power ISA 2.07) instructions
1537 * Support for Intel AVX2 instructions. This is available only on 64
1540 * Initial support for Intel Transactional Synchronization Extensions,
1543 * Initial support for Hardware Transactional Memory on POWER.
1545 * Improved support for MacOSX 10.8 (64-bit only). Memcheck can now
1546 run large GUI apps tolerably well.
1548 * ==================== TOOL CHANGES ====================
1552 - Improvements in handling of vectorised code, leading to
1553 significantly fewer false error reports. You need to use the flag
1554 --partial-loads-ok=yes to get the benefits of these changes.
1556 - Better control over the leak checker. It is now possible to
1557 specify which leak kinds (definite/indirect/possible/reachable)
1558 should be displayed, which should be regarded as errors, and which
1559 should be suppressed by a given leak suppression. This is done
1560 using the options --show-leak-kinds=kind1,kind2,..,
1561 --errors-for-leak-kinds=kind1,kind2,.. and an optional
1562 "match-leak-kinds:" line in suppression entries, respectively.
1564 Note that generated leak suppressions contain this new line and
1565 are therefore more specific than in previous releases. To get the
1566 same behaviour as previous releases, remove the "match-leak-kinds:"
1567 line from generated suppressions before using them.
1569 - Reduced "possible leak" reports from the leak checker by the use
1570 of better heuristics. The available heuristics provide detection
1571 of valid interior pointers to std::stdstring, to new[] allocated
1572 arrays with elements having destructors and to interior pointers
1573 pointing to an inner part of a C++ object using multiple
1574 inheritance. They can be selected individually using the
1575 option --leak-check-heuristics=heur1,heur2,...
1577 - Better control of stacktrace acquisition for heap-allocated
1578 blocks. Using the --keep-stacktraces option, it is possible to
1579 control independently whether a stack trace is acquired for each
1580 allocation and deallocation. This can be used to create better
1581 "use after free" errors or to decrease Valgrind's resource
1582 consumption by recording less information.
1584 - Better reporting of leak suppression usage. The list of used
1585 suppressions (shown when the -v option is given) now shows, for
1586 each leak suppressions, how many blocks and bytes it suppressed
1587 during the last leak search.
1591 - False errors resulting from the use of statically initialised
1592 mutexes and condition variables (PTHREAD_MUTEX_INITIALISER, etc)
1595 - False errors resulting from the use of pthread_cond_waits that
1596 timeout, have been removed.
1598 * ==================== OTHER CHANGES ====================
1600 * Some attempt to tune Valgrind's space requirements to the expected
1601 capabilities of the target:
1603 - The default size of the translation cache has been reduced from 8
1604 sectors to 6 on Android platforms, since each sector occupies
1605 about 40MB when using Memcheck.
1607 - The default size of the translation cache has been increased to 16
1608 sectors on all other platforms, reflecting the fact that large
1609 applications require instrumentation and storage of huge amounts
1610 of code. For similar reasons, the number of memory mapped
1611 segments that can be tracked has been increased by a factor of 6.
1613 - In all cases, the maximum number of sectors in the translation
1614 cache can be controlled by the new flag --num-transtab-sectors.
1616 * Changes in how debug info (line numbers, etc) is read:
1618 - Valgrind no longer temporarily mmaps the entire object to read
1619 from it. Instead, reading is done through a small fixed sized
1620 buffer. This avoids virtual memory usage spikes when Valgrind
1621 reads debuginfo from large shared objects.
1623 - A new experimental remote debug info server. Valgrind can read
1624 debug info from a different machine (typically, a build host)
1625 where debuginfo objects are stored. This can save a lot of time
1626 and hassle when running Valgrind on resource-constrained targets
1627 (phones, tablets) when the full debuginfo objects are stored
1628 somewhere else. This is enabled by the --debuginfo-server=
1631 - Consistency checking between main and debug objects can be
1632 disabled using the --allow-mismatched-debuginfo option.
1634 * Stack unwinding by stack scanning, on ARM. Unwinding by stack
1635 scanning can recover stack traces in some cases when the normal
1636 unwind mechanisms fail. Stack scanning is best described as "a
1637 nasty, dangerous and misleading hack" and so is disabled by default.
1638 Use --unw-stack-scan-thresh and --unw-stack-scan-frames to enable
1641 * Detection and merging of recursive stack frame cycles. When your
1642 program has recursive algorithms, this limits the memory used by
1643 Valgrind for recorded stack traces and avoids recording
1644 uninteresting repeated calls. This is controlled by the command
1645 line option --merge-recursive-frame and by the monitor command
1646 "v.set merge-recursive-frames".
1648 * File name and line numbers for used suppressions. The list of used
1649 suppressions (shown when the -v option is given) now shows, for each
1650 used suppression, the file name and line number where the suppression
1653 * New and modified GDB server monitor features:
1655 - valgrind.h has a new client request, VALGRIND_MONITOR_COMMAND,
1656 that can be used to execute gdbserver monitor commands from the
1659 - A new monitor command, "v.info open_fds", that gives the list of
1660 open file descriptors and additional details.
1662 - An optional message in the "v.info n_errs_found" monitor command,
1663 for example "v.info n_errs_found test 1234 finished", allowing a
1664 comment string to be added to the process output, perhaps for the
1665 purpose of separating errors of different tests or test phases.
1667 - A new monitor command "v.info execontext" that shows information
1668 about the stack traces recorded by Valgrind.
1670 - A new monitor command "v.do expensive_sanity_check_general" to run
1671 some internal consistency checks.
1673 * New flag --sigill-diagnostics to control whether a diagnostic
1674 message is printed when the JIT encounters an instruction it can't
1675 translate. The actual behavior -- delivery of SIGILL to the
1676 application -- is unchanged.
1678 * The maximum amount of memory that Valgrind can use on 64 bit targets
1679 has been increased from 32GB to 64GB. This should make it possible
1680 to run applications on Memcheck that natively require up to about 35GB.
1682 * ==================== FIXED BUGS ====================
1684 The following bugs have been fixed or resolved. Note that "n-i-bz"
1685 stands for "not in bugzilla" -- that is, a bug that was reported to us
1686 but never got a bugzilla entry. We encourage you to file bugs in
1687 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1688 than mailing the developers (or mailing lists) directly -- bugs that
1689 are not entered into bugzilla tend to get forgotten about or ignored.
1691 To see details of a given bug, visit
1692 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1693 where XXXXXX is the bug number as listed below.
1695 123837 system call: 4th argument is optional, depending on cmd
1696 135425 memcheck should tell you where Freed blocks were Mallocd
1697 164485 VG_N_SEGNAMES and VG_N_SEGMENTS are (still) too small
1698 207815 Adds some of the drm ioctls to syswrap-linux.c
1699 251569 vex amd64->IR: 0xF 0x1 0xF9 0xBF 0x90 0xD0 0x3 0x0 (RDTSCP)
1700 252955 Impossible to compile with ccache
1701 253519 Memcheck reports auxv pointer accesses as invalid reads.
1702 263034 Crash when loading some PPC64 binaries
1703 269599 Increase deepest backtrace
1704 274695 s390x: Support "compare to/from logical" instructions (z196)
1705 275800 s390x: Autodetect cache info (part 2)
1706 280271 Valgrind reports possible memory leaks on still-reachable std::string
1707 284540 Memcheck shouldn't count suppressions matching still-reachable [..]
1708 289578 Backtraces with ARM unwind tables (stack scan flags)
1709 296311 Wrong stack traces due to -fomit-frame-pointer (x86)
1710 304832 ppc32: build failure
1711 305431 Use find_buildid shdr fallback for separate .debug files
1712 305728 Add support for AVX2 instructions
1713 305948 ppc64: code generation for ShlD64 / ShrD64 asserts
1714 306035 s390x: Fix IR generation for LAAG and friends
1715 306054 s390x: Condition code computation for convert-to-int/logical
1716 306098 s390x: alternate opcode form for convert to/from fixed
1717 306587 Fix cache line detection from auxiliary vector for PPC.
1718 306783 Mips unhandled syscall : 4025 / 4079 / 4182
1719 307038 DWARF2 CFI reader: unhandled DW_OP_ opcode 0x8 (DW_OP_const1u et al)
1720 307082 HG false positive: pthread_cond_destroy: destruction of unknown CV
1721 307101 sys_capget second argument can be NULL
1722 307103 sys_openat: If pathname is absolute, then dirfd is ignored.
1723 307106 amd64->IR: f0 0f c0 02 (lock xadd byte)
1724 307113 s390x: DFP support
1725 307141 valgrind does't work in mips-linux system
1726 307155 filter_gdb should filter out syscall-template.S T_PSEUDO
1727 307285 x86_amd64 feature test for avx in test suite is wrong
1728 307290 memcheck overlap testcase needs memcpy version filter
1729 307463 Please add "&limit=0" to the "all open bugs" link
1730 307465 --show-possibly-lost=no should reduce the error count / exit code
1731 307557 Leaks on Mac OS X 10.7.5 libraries at ImageLoader::recursiveInit[..]
1732 307729 pkgconfig support broken valgrind.pc
1733 307828 Memcheck false errors SSE optimized wcscpy, wcscmp, wcsrchr, wcschr
1734 307955 Building valgrind 3.7.0-r4 fails in Gentoo AMD64 when using clang
1735 308089 Unhandled syscall on ppc64: prctl
1736 308135 PPC32 MPC8xx has 16 bytes cache size
1737 308321 testsuite memcheck filter interferes with gdb_filter
1739 308341 vgdb should report process exit (or fatal signal)
1740 308427 s390 memcheck reports tsearch cjump/cmove depends on uninit
1741 308495 Remove build dependency on installed Xen headers
1742 308573 Internal error on 64-bit instruction executed in 32-bit mode
1744 308627 pmovmskb validity bit propagation is imprecise
1745 308644 vgdb command for having the info for the track-fds option
1746 308711 give more info about aspacemgr and arenas in out_of_memory
1747 308717 ARM: implement fixed-point VCVT.F64.[SU]32
1748 308718 ARM implement SMLALBB family of instructions
1749 308886 Missing support for PTRACE_SET/GETREGSET
1750 308930 syscall name_to_handle_at (303 on amd64) not handled
1751 309229 V-bit tester does not report number of tests generated
1752 309323 print unrecognized instuction on MIPS
1753 309425 Provide a --sigill-diagnostics flag to suppress illegal [..]
1754 309427 SSE optimized stpncpy trigger uninitialised value [..] errors
1755 309430 Self hosting ppc64 encounters a vassert error on operand type
1756 309600 valgrind is a bit confused about 0-sized sections
1757 309823 Generate errors for still reachable blocks
1758 309921 PCMPISTRI validity bit propagation is imprecise
1759 309922 none/tests/ppc64/test_dfp5 sometimes fails
1760 310169 The Iop_CmpORD class of Iops is not supported by the vbit checker.
1761 310424 --read-var-info does not properly describe static variables
1762 310792 search additional path for debug symbols
1763 310931 s390x: Message-security assist (MSA) instruction extension [..]
1764 311100 PPC DFP implementation of the integer operands is inconsistent [..]
1765 311318 ARM: "128-bit constant is not implemented" error message
1766 311407 ssse3 bcopy (actually converted memcpy) causes invalid read [..]
1767 311690 V crashes because it redirects branches inside of a redirected function
1768 311880 x86_64: make regtest hangs at shell_valid1
1769 311922 WARNING: unhandled syscall: 170
1771 312171 ppc: insn selection for DFP
1772 312571 Rounding mode call wrong for the DFP Iops [..]
1773 312620 Change to Iop_D32toD64 [..] for s390 DFP support broke ppc [..]
1774 312913 Dangling pointers error should also report the alloc stack trace
1775 312980 Building on Mountain Lion generates some compiler warnings
1776 313267 Adding MIPS64/Linux port to Valgrind
1779 313811 Buffer overflow in assert_fail
1780 314099 coverity pointed out error in VEX guest_ppc_toIR.c insn_suffix
1781 314269 ppc: dead code in insn selection
1782 314718 ARM: implement integer divide instruction (sdiv and udiv)
1783 315345 cl-format.xml and callgrind/dump.c don't agree on using cfl= or cfi=
1784 315441 sendmsg syscall should ignore unset msghdr msg_flags
1785 315534 msgrcv inside a thread causes valgrind to hang (block)
1786 315545 Assertion '(UChar*)sec->tt[tteNo].tcptr <= (UChar*)hcode' failed
1787 315689 disInstr(thumb): unhandled instruction: 0xF852 0x0E10 (LDRT)
1788 315738 disInstr(arm): unhandled instruction: 0xEEBE0BEE (vcvt.s32.f64)
1789 315959 valgrind man page has bogus SGCHECK (and no BBV) OPTIONS section
1790 316144 valgrind.1 manpage contains unknown ??? strings [..]
1791 316145 callgrind command line options in manpage reference (unknown) [..]
1792 316145 callgrind command line options in manpage reference [..]
1793 316181 drd: Fixed a 4x slowdown for certain applications
1794 316503 Valgrind does not support SSE4 "movntdqa" instruction
1795 316535 Use of |signed int| instead of |size_t| in valgrind messages
1796 316696 fluidanimate program of parsec 2.1 stuck
1797 316761 syscall open_by_handle_at (304 on amd64, 342 on x86) not handled
1798 317091 Use -Wl,-Ttext-segment when static linking if possible [..]
1799 317186 "Impossible happens" when occurs VCVT instruction on ARM
1800 317318 Support for Threading Building Blocks "scalable_malloc"
1801 317444 amd64->IR: 0xC4 0x41 0x2C 0xC2 0xD2 0x8 (vcmpeq_uqps)
1802 317461 Fix BMI assembler configure check and avx2/bmi/fma vgtest prereqs
1803 317463 bmi testcase IR SANITY CHECK FAILURE
1804 317506 memcheck/tests/vbit-test fails with unknown opcode after [..]
1805 318050 libmpiwrap fails to compile with out-of-source build
1806 318203 setsockopt handling needs to handle SOL_SOCKET/SO_ATTACH_FILTER
1807 318643 annotate_trace_memory tests infinite loop on arm and ppc [..]
1808 318773 amd64->IR: 0xF3 0x48 0x0F 0xBC 0xC2 0xC3 0x66 0x0F
1809 318929 Crash with: disInstr(thumb): 0xF321 0x0001 (ssat16)
1810 318932 Add missing PPC64 and PPC32 system call support
1811 319235 --db-attach=yes is broken with Yama (ptrace scoping) enabled
1812 319395 Crash with unhandled instruction on STRT (Thumb) instructions
1813 319494 VEX Makefile-gcc standalone build update after r2702
1814 319505 [MIPSEL] Crash: unhandled UNRAY operator.
1815 319858 disInstr(thumb): unhandled instruction on instruction STRBT
1816 319932 disInstr(thumb): unhandled instruction on instruction STRHT
1817 320057 Problems when we try to mmap more than 12 memory pages on MIPS32
1818 320063 Memory from PTRACE_GET_THREAD_AREA is reported uninitialised
1819 320083 disInstr(thumb): unhandled instruction on instruction LDRBT
1820 320116 bind on AF_BLUETOOTH produces warnings because of sockaddr_rc padding
1821 320131 WARNING: unhandled syscall: 369 on ARM (prlimit64)
1822 320211 Stack buffer overflow in ./coregrind/m_main.c with huge TMPDIR
1823 320661 vgModuleLocal_read_elf_debug_info(): "Assertion '!di->soname'
1824 320895 add fanotify support (patch included)
1825 320998 vex amd64->IR pcmpestri and pcmpestrm SSE4.2 instruction
1826 321065 Valgrind updates for Xen 4.3
1827 321148 Unhandled instruction: PLI (Thumb 1, 2, 3)
1828 321363 Unhandled instruction: SSAX (ARM + Thumb)
1829 321364 Unhandled instruction: SXTAB16 (ARM + Thumb)
1830 321466 Unhandled instruction: SHASX (ARM + Thumb)
1831 321467 Unhandled instruction: SHSAX (ARM + Thumb)
1832 321468 Unhandled instruction: SHSUB16 (ARM + Thumb)
1833 321619 Unhandled instruction: SHSUB8 (ARM + Thumb)
1834 321620 Unhandled instruction: UASX (ARM + Thumb)
1835 321621 Unhandled instruction: USAX (ARM + Thumb)
1836 321692 Unhandled instruction: UQADD16 (ARM + Thumb)
1837 321693 Unhandled instruction: LDRSBT (Thumb)
1838 321694 Unhandled instruction: UQASX (ARM + Thumb)
1839 321696 Unhandled instruction: UQSAX (Thumb + ARM)
1840 321697 Unhandled instruction: UHASX (ARM + Thumb)
1841 321703 Unhandled instruction: UHSAX (ARM + Thumb)
1842 321704 Unhandled instruction: REVSH (ARM + Thumb)
1843 321730 Add cg_diff and cg_merge man pages
1844 321738 Add vgdb and valgrind-listener man pages
1846 321891 Unhandled instruction: LDRHT (Thumb)
1847 321960 pthread_create() then alloca() causing invalid stack write errors
1848 321969 ppc32 and ppc64 don't support [lf]setxattr
1849 322254 Show threadname together with tid if set by application
1850 322294 Add initial support for IBM Power ISA 2.07
1851 322368 Assertion failure in wqthread_hijack under OS X 10.8
1852 322563 vex mips->IR: 0x70 0x83 0xF0 0x3A
1853 322807 VALGRIND_PRINTF_BACKTRACE writes callstack to xml and text to stderr
1854 322851 0bXXX binary literal syntax is not standard
1855 323035 Unhandled instruction: LDRSHT(Thumb)
1856 323036 Unhandled instruction: SMMLS (ARM and Thumb)
1857 323116 The memcheck/tests/ppc64/power_ISA2_05.c fails to build [..]
1858 323175 Unhandled instruction: SMLALD (ARM + Thumb)
1859 323177 Unhandled instruction: SMLSLD (ARM + Thumb)
1860 323432 Calling pthread_cond_destroy() or pthread_mutex_destroy() [..]
1861 323437 Phase 2 support for IBM Power ISA 2.07
1862 323713 Support mmxext (integer sse) subset on i386 (athlon)
1863 323803 Transactional memory instructions are not supported for Power
1864 323893 SSE3 not available on amd cpus in valgrind
1865 323905 Probable false positive from Valgrind/drd on close()
1866 323912 valgrind.h header isn't compatible for mingw64
1867 324047 Valgrind doesn't support [LDR,ST]{S}[B,H]T ARM instructions
1868 324149 helgrind: When pthread_cond_timedwait returns ETIMEDOUT [..]
1869 324181 mmap does not handle MAP_32BIT
1870 324227 memcheck false positive leak when a thread calls exit+block [..]
1871 324421 Support for fanotify API on ARM architecture
1872 324514 gdbserver monitor cmd output behaviour consistency [..]
1873 324518 ppc64: Emulation of dcbt instructions does not handle [..]
1874 324546 none/tests/ppc32 test_isa_2_07_part2 requests -m64
1875 324582 When access is made to freed memory, report both allocation [..]
1876 324594 Fix overflow computation for Power ISA 2.06 insns: mulldo/mulldo.
1877 324765 ppc64: illegal instruction when executing none/tests/ppc64/jm-misc
1878 324816 Incorrect VEX implementation for xscvspdp/xvcvspdp for SNaN inputs
1879 324834 Unhandled instructions in Microsoft C run-time for x86_64
1880 324894 Phase 3 support for IBM Power ISA 2.07
1881 326091 drd: Avoid false race reports from optimized strlen() impls
1882 326113 valgrind libvex hwcaps error on AMD64
1883 n-i-bz Some wrong command line options could be ignored
1884 n-i-bz patch to allow fair-sched on android
1885 n-i-bz report error for vgdb snapshot requested before execution
1886 n-i-bz same as 303624 (fixed in 3.8.0), but for x86 android
1888 (3.9.0: 31 October 2013, vex r2796, valgrind r13708)
1892 Release 3.8.1 (19 September 2012)
1893 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1894 3.8.1 is a bug fix release. It fixes some assertion failures in 3.8.0
1895 that occur moderately frequently in real use cases, adds support for
1896 some missing instructions on ARM, and fixes a deadlock condition on
1897 MacOSX. If you package or deliver 3.8.0 for others to use, you might
1898 want to consider upgrading to 3.8.1 instead.
1900 The following bugs have been fixed or resolved. Note that "n-i-bz"
1901 stands for "not in bugzilla" -- that is, a bug that was reported to us
1902 but never got a bugzilla entry. We encourage you to file bugs in
1903 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
1904 than mailing the developers (or mailing lists) directly -- bugs that
1905 are not entered into bugzilla tend to get forgotten about or ignored.
1907 To see details of a given bug, visit
1908 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
1909 where XXXXXX is the bug number as listed below.
1912 289584 Unhandled instruction: 0xF 0x29 0xE5 (MOVAPS)
1913 295808 amd64->IR: 0xF3 0xF 0xBC 0xC0 (TZCNT)
1914 298281 wcslen causes false(?) uninitialised value warnings
1915 301281 valgrind hangs on OS X when the process calls system()
1916 304035 disInstr(arm): unhandled instruction 0xE1023053
1917 304867 implement MOVBE instruction in x86 mode
1918 304980 Assertion 'lo <= hi' failed in vgModuleLocal_find_rx_mapping
1919 305042 amd64: implement 0F 7F encoding of movq between two registers
1920 305199 ARM: implement QDADD and QDSUB
1921 305321 amd64->IR: 0xF 0xD 0xC (prefetchw)
1922 305513 killed by fatal signal: SIGSEGV
1923 305690 DRD reporting invalid semaphore when sem_trywait fails
1924 305926 Invalid alignment checks for some AVX instructions
1925 306297 disInstr(thumb): unhandled instruction 0xE883 0x000C
1926 306310 3.8.0 release tarball missing some files
1927 306612 RHEL 6 glibc-2.X default suppressions need /lib*/libc-*patterns
1928 306664 vex amd64->IR: 0x66 0xF 0x3A 0x62 0xD1 0x46 0x66 0xF
1929 n-i-bz shmat of a segment > 4Gb does not work
1930 n-i-bz simulate_control_c script wrong USR1 signal number on mips
1931 n-i-bz vgdb ptrace calls wrong on mips [...]
1932 n-i-bz Fixes for more MPI false positives
1933 n-i-bz exp-sgcheck's memcpy causes programs to segfault
1934 n-i-bz OSX build w/ clang: asserts at startup
1935 n-i-bz Incorrect undef'dness prop for Iop_DPBtoBCD and Iop_BCDtoDPB
1936 n-i-bz fix a couple of union tag-vs-field mixups
1937 n-i-bz OSX: use __NR_poll_nocancel rather than __NR_poll
1939 The following bugs were fixed in 3.8.0 but not listed in this NEWS
1942 254088 Valgrind should know about UD2 instruction
1945 304754 NEWS blows TeX's little mind
1947 (3.8.1: 19 September 2012, vex r2537, valgrind r12996)
1951 Release 3.8.0 (10 August 2012)
1952 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1953 3.8.0 is a feature release with many improvements and the usual
1954 collection of bug fixes.
1956 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
1957 PPC64/Linux, S390X/Linux, MIPS/Linux, ARM/Android, X86/Android,
1958 X86/MacOSX 10.6/10.7 and AMD64/MacOSX 10.6/10.7. Support for recent
1959 distros and toolchain components (glibc 2.16, gcc 4.7) has been added.
1960 There is initial support for MacOSX 10.8, but it is not usable for
1961 serious work at present.
1963 * ================== PLATFORM CHANGES =================
1965 * Support for MIPS32 platforms running Linux. Valgrind has been
1966 tested on MIPS32 and MIPS32r2 platforms running different Debian
1967 Squeeze and MeeGo distributions. Both little-endian and big-endian
1968 cores are supported. The tools Memcheck, Massif and Lackey have
1969 been tested and are known to work. See README.mips for more details.
1971 * Preliminary support for Android running on x86.
1973 * Preliminary (as-yet largely unusable) support for MacOSX 10.8.
1975 * Support for Intel AVX instructions and for AES instructions. This
1976 support is available only for 64 bit code.
1978 * Support for POWER Decimal Floating Point instructions.
1980 * ==================== TOOL CHANGES ====================
1982 * Non-libc malloc implementations are now supported. This is useful
1983 for tools that replace malloc (Memcheck, Massif, DRD, Helgrind).
1984 Using the new option --soname-synonyms, such tools can be informed
1985 that the malloc implementation is either linked statically into the
1986 executable, or is present in some other shared library different
1987 from libc.so. This makes it possible to process statically linked
1988 programs, and programs using other malloc libraries, for example
1989 TCMalloc or JEMalloc.
1991 * For tools that provide their own replacement for malloc et al, the
1992 option --redzone-size=<number> allows users to specify the size of
1993 the padding blocks (redzones) added before and after each client
1994 allocated block. Smaller redzones decrease the memory needed by
1995 Valgrind. Bigger redzones increase the chance to detect blocks
1996 overrun or underrun. Prior to this change, the redzone size was
1997 hardwired to 16 bytes in Memcheck.
2001 - The leak_check GDB server monitor command now can
2002 control the maximum nr of loss records to output.
2004 - Reduction of memory use for applications allocating
2005 many blocks and/or having many partially defined bytes.
2007 - Addition of GDB server monitor command 'block_list' that lists
2008 the addresses/sizes of the blocks of a leak search loss record.
2010 - Addition of GDB server monitor command 'who_points_at' that lists
2011 the locations pointing at a block.
2013 - If a redzone size > 0 is given, VALGRIND_MALLOCLIKE_BLOCK now will
2014 detect an invalid access of these redzones, by marking them
2015 noaccess. Similarly, if a redzone size is given for a memory
2016 pool, VALGRIND_MEMPOOL_ALLOC will mark the redzones no access.
2017 This still allows to find some bugs if the user has forgotten to
2018 mark the pool superblock noaccess.
2020 - Performance of memory leak check has been improved, especially in
2021 cases where there are many leaked blocks and/or many suppression
2022 rules used to suppress leak reports.
2024 - Reduced noise (false positive) level on MacOSX 10.6/10.7, due to
2025 more precise analysis, which is important for LLVM/Clang
2026 generated code. This is at the cost of somewhat reduced
2027 performance. Note there is no change to analysis precision or
2028 costs on Linux targets.
2032 - Added even more facilities that can help finding the cause of a data
2033 race, namely the command-line option --ptrace-addr and the macro
2034 DRD_STOP_TRACING_VAR(x). More information can be found in the manual.
2036 - Fixed a subtle bug that could cause false positive data race reports.
2038 * ==================== OTHER CHANGES ====================
2040 * The C++ demangler has been updated so as to work well with C++
2041 compiled by up to at least g++ 4.6.
2043 * Tool developers can make replacement/wrapping more flexible thanks
2044 to the new option --soname-synonyms. This was reported above, but
2045 in fact is very general and applies to all function
2046 replacement/wrapping, not just to malloc-family functions.
2048 * Round-robin scheduling of threads can be selected, using the new
2049 option --fair-sched= yes. Prior to this change, the pipe-based
2050 thread serialisation mechanism (which is still the default) could
2051 give very unfair scheduling. --fair-sched=yes improves
2052 responsiveness of interactive multithreaded applications, and
2053 improves repeatability of results from the thread checkers Helgrind
2056 * For tool developers: support to run Valgrind on Valgrind has been
2057 improved. We can now routinely Valgrind on Helgrind or Memcheck.
2059 * gdbserver now shows the float shadow registers as integer
2060 rather than float values, as the shadow values are mostly
2061 used as bit patterns.
2063 * Increased limit for the --num-callers command line flag to 500.
2065 * Performance improvements for error matching when there are many
2066 suppression records in use.
2068 * Improved support for DWARF4 debugging information (bug 284184).
2070 * Initial support for DWZ compressed Dwarf debug info.
2072 * Improved control over the IR optimiser's handling of the tradeoff
2073 between performance and precision of exceptions. Specifically,
2074 --vex-iropt-precise-memory-exns has been removed and replaced by
2075 --vex-iropt-register-updates, with extended functionality. This
2076 allows the Valgrind gdbserver to always show up to date register
2079 * Modest performance gains through the use of translation chaining for
2082 * ==================== FIXED BUGS ====================
2084 The following bugs have been fixed or resolved. Note that "n-i-bz"
2085 stands for "not in bugzilla" -- that is, a bug that was reported to us
2086 but never got a bugzilla entry. We encourage you to file bugs in
2087 bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
2088 than mailing the developers (or mailing lists) directly -- bugs that
2089 are not entered into bugzilla tend to get forgotten about or ignored.
2091 To see details of a given bug, visit
2092 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2093 where XXXXXX is the bug number as listed below.
2095 197914 Building valgrind from svn now requires automake-1.10
2096 203877 increase to 16Mb maximum allowed alignment for memalign et al
2097 219156 Handle statically linked malloc or other malloc lib (e.g. tcmalloc)
2098 247386 make perf does not run all performance tests
2099 270006 Valgrind scheduler unfair
2100 270777 Adding MIPS/Linux port to Valgrind
2101 270796 s390x: Removed broken support for the TS insn
2102 271438 Fix configure for proper SSE4.2 detection
2103 273114 s390x: Support TR, TRE, TROO, TROT, TRTO, and TRTT instructions
2104 273475 Add support for AVX instructions
2105 274078 improved configure logic for mpicc
2106 276993 fix mremap 'no thrash checks'
2107 278313 Fedora 15/x64: err read debug info with --read-var-info=yes flag
2108 281482 memcheck incorrect byte allocation count in realloc() for silly argument
2109 282230 group allocator for small fixed size, use it for MC_Chunk/SEc vbit
2110 283413 Fix wrong sanity check
2111 283671 Robustize alignment computation in LibVEX_Alloc
2112 283961 Adding support for some HCI IOCTLs
2113 284124 parse_type_DIE: confused by: DWARF 4
2114 284864 == 273475 (Add support for AVX instructions)
2115 285219 Too-restrictive constraints for Thumb2 "SP plus/minus register"
2116 285662 (MacOSX): Memcheck needs to replace memcpy/memmove
2117 285725 == 273475 (Add support for AVX instructions)
2118 286261 add wrapper for linux I2C_RDWR ioctl
2119 286270 vgpreload is not friendly to 64->32 bit execs, gives ld.so warnings
2120 286374 Running cachegrind with --branch-sim=yes on 64-bit PowerPC program fails
2121 286384 configure fails "checking for a supported version of gcc"
2122 286497 == 273475 (Add support for AVX instructions)
2123 286596 == 273475 (Add support for AVX instructions)
2124 286917 disInstr(arm): unhandled instruction: QADD (also QSUB)
2125 287175 ARM: scalar VFP fixed-point VCVT instructions not handled
2126 287260 Incorrect conditional jump or move depends on uninitialised value(s)
2127 287301 vex amd64->IR: 0x66 0xF 0x38 0x41 0xC0 0xB8 0x0 0x0 (PHMINPOSUW)
2128 287307 == 273475 (Add support for AVX instructions)
2129 287858 VG_(strerror): unknown error
2130 288298 (MacOSX) unhandled syscall shm_unlink
2131 288995 == 273475 (Add support for AVX instructions)
2132 289470 Loading of large Mach-O thin binaries fails.
2133 289656 == 273475 (Add support for AVX instructions)
2134 289699 vgdb connection in relay mode erroneously closed due to buffer overrun
2135 289823 == 293754 (PCMPxSTRx not implemented for 16-bit characters)
2136 289839 s390x: Provide support for unicode conversion instructions
2137 289939 monitor cmd 'leak_check' with details about leaked or reachable blocks
2138 290006 memcheck doesn't mark %xmm as initialized after "pcmpeqw %xmm %xmm"
2139 290655 Add support for AESKEYGENASSIST instruction
2140 290719 valgrind-3.7.0 fails with automake-1.11.2 due to"pkglibdir" usage
2141 290974 vgdb must align pages to VKI_SHMLBA (16KB) on ARM
2142 291253 ES register not initialised in valgrind simulation
2143 291568 Fix 3DNOW-related crashes with baseline x86_64 CPU (w patch)
2144 291865 s390x: Support the "Compare Double and Swap" family of instructions
2145 292300 == 273475 (Add support for AVX instructions)
2146 292430 unrecognized instruction in __intel_get_new_mem_ops_cpuid
2147 292493 == 273475 (Add support for AVX instructions)
2148 292626 Missing fcntl F_SETOWN_EX and F_GETOWN_EX support
2149 292627 Missing support for some SCSI ioctls
2150 292628 none/tests/x86/bug125959-x86.c triggers undefined behavior
2151 292841 == 273475 (Add support for AVX instructions)
2152 292993 implement the getcpu syscall on amd64-linux
2153 292995 Implement the “cross memory attach” syscalls introduced in Linux 3.2
2154 293088 Add some VEX sanity checks for ppc64 unhandled instructions
2155 293751 == 290655 (Add support for AESKEYGENASSIST instruction)
2156 293754 PCMPxSTRx not implemented for 16-bit characters
2157 293755 == 293754 (No tests for PCMPxSTRx on 16-bit characters)
2158 293808 CLFLUSH not supported by latest VEX for amd64
2159 294047 valgrind does not correctly emulate prlimit64(..., RLIMIT_NOFILE, ...)
2160 294048 MPSADBW instruction not implemented
2161 294055 regtest none/tests/shell fails when locale is not set to C
2162 294185 INT 0x44 (and others) not supported on x86 guest, but used by Jikes RVM
2163 294190 --vgdb-error=xxx can be out of sync with errors shown to the user
2164 294191 amd64: fnsave/frstor and 0x66 size prefixes on FP instructions
2165 294260 disInstr_AMD64: disInstr miscalculated next %rip
2166 294523 --partial-loads-ok=yes causes false negatives
2167 294617 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A
2168 294736 vex amd64->IR: 0x48 0xF 0xD7 0xD6 0x48 0x83
2169 294812 patch allowing to run (on x86 at least) helgrind/drd on tool.
2170 295089 can not annotate source for both helgrind and drd
2171 295221 POWER Processor decimal floating point instruction support missing
2172 295427 building for i386 with clang on darwin11 requires "-new_linker linker"
2173 295428 coregrind/m_main.c has incorrect x86 assembly for darwin
2174 295590 Helgrind: Assertion 'cvi->nWaiters > 0' failed
2175 295617 ARM - Add some missing syscalls
2176 295799 Missing \n with get_vbits in gdbserver when line is % 80 [...]
2177 296229 Linux user input device ioctls missing wrappers
2178 296318 ELF Debug info improvements (more than one rx/rw mapping)
2179 296422 Add translation chaining support
2180 296457 vex amd64->IR: 0x66 0xF 0x3A 0xDF 0xD1 0x1 0xE8 0x6A (dup of AES)
2181 296792 valgrind 3.7.0: add SIOCSHWTSTAMP (0x89B0) ioctl wrapper
2182 296983 Fix build issues on x86_64/ppc64 without 32-bit toolchains
2183 297078 gdbserver signal handling problems [..]
2184 297147 drd false positives on newly allocated memory
2185 297329 disallow decoding of IBM Power DFP insns on some machines
2186 297497 POWER Processor decimal floating point instruction support missing
2187 297701 Another alias for strncasecmp_l in libc-2.13.so
2188 297911 'invalid write' not reported when using APIs for custom mem allocators.
2189 297976 s390x: revisit EX implementation
2190 297991 Valgrind interferes with mmap()+ftell()
2191 297992 Support systems missing WIFCONTINUED (e.g. pre-2.6.10 Linux)
2192 297993 Fix compilation of valgrind with gcc -g3.
2193 298080 POWER Processor DFP support missing, part 3
2194 298227 == 273475 (Add support for AVX instructions)
2195 298335 == 273475 (Add support for AVX instructions)
2196 298354 Unhandled ARM Thumb instruction 0xEB0D 0x0585 (streq)
2197 298394 s390x: Don't bail out on an unknown machine model. [..]
2198 298421 accept4() syscall (366) support is missing for ARM
2199 298718 vex amd64->IR: 0xF 0xB1 0xCB 0x9C 0x8F 0x45
2200 298732 valgrind installation problem in ubuntu with kernel version 3.x
2201 298862 POWER Processor DFP instruction support missing, part 4
2202 298864 DWARF reader mis-parses DW_FORM_ref_addr
2203 298943 massif asserts with --pages-as-heap=yes when brk is changing [..]
2204 299053 Support DWARF4 DW_AT_high_pc constant form
2205 299104 == 273475 (Add support for AVX instructions)
2206 299316 Helgrind: hg_main.c:628 (map_threads_lookup): Assertion 'thr' failed.
2207 299629 dup3() syscall (358) support is missing for ARM
2208 299694 POWER Processor DFP instruction support missing, part 5
2209 299756 Ignore --free-fill for MEMPOOL_FREE and FREELIKE client requests
2210 299803 == 273475 (Add support for AVX instructions)
2211 299804 == 273475 (Add support for AVX instructions)
2212 299805 == 273475 (Add support for AVX instructions)
2213 300140 ARM - Missing (T1) SMMUL
2214 300195 == 296318 (ELF Debug info improvements (more than one rx/rw mapping))
2215 300389 Assertion `are_valid_hwcaps(VexArchAMD64, [..])' failed.
2216 300414 FCOM and FCOMP unimplemented for amd64 guest
2217 301204 infinite loop in canonicaliseSymtab with ifunc symbol
2218 301229 == 203877 (increase to 16Mb maximum allowed alignment for memalign etc)
2219 301265 add x86 support to Android build
2220 301984 configure script doesn't detect certain versions of clang
2221 302205 Fix compiler warnings for POWER VEX code and POWER test cases
2222 302287 Unhandled movbe instruction on Atom processors
2223 302370 PPC: fnmadd, fnmsub, fnmadds, fnmsubs insns always negate the result
2224 302536 Fix for the POWER Valgrind regression test: memcheck-ISA2.0.
2225 302578 Unrecognized isntruction 0xc5 0x32 0xc2 0xca 0x09 vcmpngess
2226 302656 == 273475 (Add support for AVX instructions)
2227 302709 valgrind for ARM needs extra tls support for android emulator [..]
2228 302827 add wrapper for CDROM_GET_CAPABILITY
2229 302901 Valgrind crashes with dwz optimized debuginfo
2230 302918 Enable testing of the vmaddfp and vnsubfp instructions in the testsuite
2231 303116 Add support for the POWER instruction popcntb
2232 303127 Power test suite fixes for frsqrte, vrefp, and vrsqrtefp instructions.
2233 303250 Assertion `instrs_in->arr_used <= 10000' failed w/ OpenSSL code
2234 303466 == 273475 (Add support for AVX instructions)
2235 303624 segmentation fault on Android 4.1 (e.g. on Galaxy Nexus OMAP)
2236 303963 strstr() function produces wrong results under valgrind callgrind
2237 304054 CALL_FN_xx macros need to enforce stack alignment
2238 304561 tee system call not supported
2239 715750 (MacOSX): Incorrect invalid-address errors near 0xFFFFxxxx (mozbug#)
2240 n-i-bz Add missing gdbserver xml files for shadow registers for ppc32
2241 n-i-bz Bypass gcc4.4/4.5 code gen bugs causing out of memory or asserts
2242 n-i-bz Fix assert in gdbserver for watchpoints watching the same address
2243 n-i-bz Fix false positive in sys_clone on amd64 when optional args [..]
2244 n-i-bz s390x: Shadow registers can now be examined using vgdb
2246 (3.8.0-TEST3: 9 August 2012, vex r2465, valgrind r12865)
2247 (3.8.0: 10 August 2012, vex r2465, valgrind r12866)
2251 Release 3.7.0 (5 November 2011)
2252 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2253 3.7.0 is a feature release with many significant improvements and the
2254 usual collection of bug fixes.
2256 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2257 PPC64/Linux, S390X/Linux, ARM/Android, X86/Darwin and AMD64/Darwin.
2258 Support for recent distros and toolchain components (glibc 2.14, gcc
2259 4.6, MacOSX 10.7) has been added.
2261 * ================== PLATFORM CHANGES =================
2263 * Support for IBM z/Architecture (s390x) running Linux. Valgrind can
2264 analyse 64-bit programs running on z/Architecture. Most user space
2265 instructions up to and including z10 are supported. Valgrind has
2266 been tested extensively on z9, z10, and z196 machines running SLES
2267 10/11, RedHat 5/6m, and Fedora. The Memcheck and Massif tools are
2268 known to work well. Callgrind, Helgrind, and DRD work reasonably
2269 well on z9 and later models. See README.s390 for more details.
2271 * Preliminary support for MacOSX 10.7 and XCode 4. Both 32- and
2272 64-bit processes are supported. Some complex threaded applications
2273 (Firefox) are observed to hang when run as 32 bit applications,
2274 whereas 64-bit versions run OK. The cause is unknown. Memcheck
2275 will likely report some false errors. In general, expect some rough
2276 spots. This release also supports MacOSX 10.6, but drops support
2279 * Preliminary support for Android (on ARM). Valgrind can now run
2280 large applications (eg, Firefox) on (eg) a Samsung Nexus S. See
2281 README.android for more details, plus instructions on how to get
2284 * Support for the IBM Power ISA 2.06 (Power7 instructions)
2286 * General correctness and performance improvements for ARM/Linux, and,
2287 by extension, ARM/Android.
2289 * Further solidification of support for SSE 4.2 in 64-bit mode. AVX
2290 instruction set support is under development but is not available in
2293 * Support for AIX5 has been removed.
2295 * ==================== TOOL CHANGES ====================
2297 * Memcheck: some incremental changes:
2299 - reduction of memory use in some circumstances
2301 - improved handling of freed memory, which in some circumstances
2302 can cause detection of use-after-free that would previously have
2305 - fix of a longstanding bug that could cause false negatives (missed
2306 errors) in programs doing vector saturated narrowing instructions.
2308 * Helgrind: performance improvements and major memory use reductions,
2309 particularly for large, long running applications which perform many
2310 synchronisation (lock, unlock, etc) events. Plus many smaller
2313 - display of locksets for both threads involved in a race
2315 - general improvements in formatting/clarity of error messages
2317 - addition of facilities and documentation regarding annotation
2318 of thread safe reference counted C++ classes
2320 - new flag --check-stack-refs=no|yes [yes], to disable race checking
2321 on thread stacks (a performance hack)
2323 - new flag --free-is-write=no|yes [no], to enable detection of races
2324 where one thread accesses heap memory but another one frees it,
2325 without any coordinating synchronisation event
2327 * DRD: enabled XML output; added support for delayed thread deletion
2328 in order to detect races that occur close to the end of a thread
2329 (--join-list-vol); fixed a memory leak triggered by repeated client
2330 memory allocatation and deallocation; improved Darwin support.
2332 * exp-ptrcheck: this tool has been renamed to exp-sgcheck
2334 * exp-sgcheck: this tool has been reduced in scope so as to improve
2335 performance and remove checking that Memcheck does better.
2336 Specifically, the ability to check for overruns for stack and global
2337 arrays is unchanged, but the ability to check for overruns of heap
2338 blocks has been removed. The tool has accordingly been renamed to
2339 exp-sgcheck ("Stack and Global Array Checking").
2341 * ==================== OTHER CHANGES ====================
2343 * GDB server: Valgrind now has an embedded GDB server. That means it
2344 is possible to control a Valgrind run from GDB, doing all the usual
2345 things that GDB can do (single stepping, breakpoints, examining
2346 data, etc). Tool-specific functionality is also available. For
2347 example, it is possible to query the definedness state of variables
2348 or memory from within GDB when running Memcheck; arbitrarily large
2349 memory watchpoints are supported, etc. To use the GDB server, start
2350 Valgrind with the flag --vgdb-error=0 and follow the on-screen
2353 * Improved support for unfriendly self-modifying code: a new option
2354 --smc-check=all-non-file is available. This adds the relevant
2355 consistency checks only to code that originates in non-file-backed
2356 mappings. In effect this confines the consistency checking only to
2357 code that is or might be JIT generated, and avoids checks on code
2358 that must have been compiled ahead of time. This significantly
2359 improves performance on applications that generate code at run time.
2361 * It is now possible to build a working Valgrind using Clang-2.9 on
2364 * new client requests VALGRIND_{DISABLE,ENABLE}_ERROR_REPORTING.
2365 These enable and disable error reporting on a per-thread, and
2366 nestable, basis. This is useful for hiding errors in particularly
2367 troublesome pieces of code. The MPI wrapper library (libmpiwrap.c)
2368 now uses this facility.
2370 * Added the --mod-funcname option to cg_diff.
2372 * ==================== FIXED BUGS ====================
2374 The following bugs have been fixed or resolved. Note that "n-i-bz"
2375 stands for "not in bugzilla" -- that is, a bug that was reported to us
2376 but never got a bugzilla entry. We encourage you to file bugs in
2377 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2378 mailing the developers (or mailing lists) directly -- bugs that are
2379 not entered into bugzilla tend to get forgotten about or ignored.
2381 To see details of a given bug, visit
2382 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2383 where XXXXXX is the bug number as listed below.
2385 79311 malloc silly arg warning does not give stack trace
2386 210935 port valgrind.h (not valgrind) to win32 to support client requests
2387 214223 valgrind SIGSEGV on startup gcc 4.4.1 ppc32 (G4) Ubuntu 9.10
2388 243404 Port to zSeries
2389 243935 Helgrind: incorrect handling of ANNOTATE_HAPPENS_BEFORE()/AFTER()
2390 247223 non-x86: Suppress warning: 'regparm' attribute directive ignored
2391 250101 huge "free" memory usage due to m_mallocfree.c fragmentation
2392 253206 Some fixes for the faultstatus testcase
2393 255223 capget testcase fails when running as root
2394 256703 xlc_dbl_u32.c testcase broken
2395 256726 Helgrind tests have broken inline asm
2396 259977 == 214223 (Valgrind segfaults doing __builtin_longjmp)
2397 264800 testcase compile failure on zseries
2398 265762 make public VEX headers compilable by G++ 3.x
2399 265771 assertion in jumps.c (r11523) fails with glibc-2.3
2400 266753 configure script does not give the user the option to not use QtCore
2401 266931 gen_insn_test.pl is broken
2402 266961 ld-linux.so.2 i?86-linux strlen issues
2403 266990 setns instruction causes false positive
2404 267020 Make directory for temporary files configurable at run-time.
2405 267342 == 267997 (segmentation fault on Mac OS 10.6)
2406 267383 Assertion 'vgPlain_strlen(dir) + vgPlain_strlen(file) + 1 < 256' failed
2407 267413 Assertion 'DRD_(g_threadinfo)[tid].synchr_nesting >= 1' failed.
2408 267488 regtest: darwin support for 64-bit build
2409 267552 SIGSEGV (misaligned_stack_error) with DRD, but not with other tools
2410 267630 Add support for IBM Power ISA 2.06 -- stage 1
2411 267769 == 267997 (Darwin: memcheck triggers segmentation fault)
2412 267819 Add client request for informing the core about reallocation
2413 267925 laog data structure quadratic for a single sequence of lock
2414 267968 drd: (vgDrd_thread_set_joinable): Assertion '0 <= (int)tid ..' failed
2415 267997 MacOSX: 64-bit V segfaults on launch when built with Xcode 4.0.1
2416 268513 missed optimizations in fold_Expr
2417 268619 s390x: fpr - gpr transfer facility
2418 268620 s390x: reconsider "long displacement" requirement
2419 268621 s390x: improve IR generation for XC
2420 268715 s390x: FLOGR is not universally available
2421 268792 == 267997 (valgrind seg faults on startup when compiled with Xcode 4)
2422 268930 s390x: MHY is not universally available
2423 269078 arm->IR: unhandled instruction SUB (SP minus immediate/register)
2424 269079 Support ptrace system call on ARM
2425 269144 missing "Bad option" error message
2426 269209 conditional load and store facility (z196)
2427 269354 Shift by zero on x86 can incorrectly clobber CC_NDEP
2428 269641 == 267997 (valgrind segfaults immediately (segmentation fault))
2429 269736 s390x: minor code generation tweaks
2430 269778 == 272986 (valgrind.h: swap roles of VALGRIND_DO_CLIENT_REQUEST() ..)
2431 269863 s390x: remove unused function parameters
2432 269864 s390x: tweak s390_emit_load_cc
2433 269884 == 250101 (overhead for huge blocks exhausts space too soon)
2434 270082 s390x: Make sure to point the PSW address to the next address on SIGILL
2435 270115 s390x: rewrite some testcases
2436 270309 == 267997 (valgrind crash on startup)
2437 270320 add support for Linux FIOQSIZE ioctl() call
2438 270326 segfault while trying to sanitize the environment passed to execle
2439 270794 IBM POWER7 support patch causes regression in none/tests
2440 270851 IBM POWER7 fcfidus instruction causes memcheck to fail
2441 270856 IBM POWER7 xsnmaddadp instruction causes memcheck to fail on 32bit app
2442 270925 hyper-optimized strspn() in /lib64/libc-2.13.so needs fix
2443 270959 s390x: invalid use of R0 as base register
2444 271042 VSX configure check fails when it should not
2445 271043 Valgrind build fails with assembler error on ppc64 with binutils 2.21
2446 271259 s390x: fix code confusion
2447 271337 == 267997 (Valgrind segfaults on MacOS X)
2448 271385 s390x: Implement Ist_MBE
2449 271501 s390x: misc cleanups
2450 271504 s390x: promote likely and unlikely
2451 271579 ppc: using wrong enum type
2452 271615 unhandled instruction "popcnt" (arch=amd10h)
2453 271730 Fix bug when checking ioctls: duplicate check
2454 271776 s390x: provide STFLE instruction support
2455 271779 s390x: provide clock instructions like STCK
2456 271799 Darwin: ioctls without an arg report a memory error
2457 271820 arm: fix type confusion
2458 271917 pthread_cond_timedwait failure leads to not-locked false positive
2459 272067 s390x: fix DISP20 macro
2460 272615 A typo in debug output in mc_leakcheck.c
2461 272661 callgrind_annotate chokes when run from paths containing regex chars
2462 272893 amd64->IR: 0x66 0xF 0x38 0x2B 0xC1 0x66 0xF 0x7F == (closed as dup)
2463 272955 Unhandled syscall error for pwrite64 on ppc64 arch
2464 272967 make documentation build-system more robust
2465 272986 Fix gcc-4.6 warnings with valgrind.h
2466 273318 amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38 (missing PCMPxSTRx case)
2467 273318 unhandled PCMPxSTRx case: vex amd64->IR: 0x66 0xF 0x3A 0x61 0xC1 0x38
2468 273431 valgrind segfaults in evalCfiExpr (debuginfo.c:2039)
2469 273465 Callgrind: jumps.c:164 (new_jcc): Assertion '(0 <= jmp) && ...'
2470 273536 Build error: multiple definition of `vgDrd_pthread_cond_initializer'
2471 273640 ppc64-linux: unhandled syscalls setresuid(164) and setresgid(169)
2472 273729 == 283000 (Illegal opcode for SSE2 "roundsd" instruction)
2473 273778 exp-ptrcheck: unhandled sysno == 259
2474 274089 exp-ptrcheck: unhandled sysno == 208
2475 274378 s390x: Various dispatcher tweaks
2476 274447 WARNING: unhandled syscall: 340
2477 274776 amd64->IR: 0x66 0xF 0x38 0x2B 0xC5 0x66
2478 274784 == 267997 (valgrind ls -l results in Segmentation Fault)
2479 274926 valgrind does not build against linux-3
2480 275148 configure FAIL with glibc-2.14
2481 275151 Fedora 15 / glibc-2.14 'make regtest' FAIL
2482 275168 Make Valgrind work for MacOSX 10.7 Lion
2483 275212 == 275284 (lots of false positives from __memcpy_ssse3_back et al)
2484 275278 valgrind does not build on Linux kernel 3.0.* due to silly
2485 275284 Valgrind memcpy/memmove redirection stopped working in glibc 2.14/x86_64
2486 275308 Fix implementation for ppc64 fres instruc
2487 275339 s390x: fix testcase compile warnings
2488 275517 s390x: Provide support for CKSM instruction
2489 275710 s390x: get rid of redundant address mode calculation
2490 275815 == 247894 (Valgrind doesn't know about Linux readahead(2) syscall)
2491 275852 == 250101 (valgrind uses all swap space and is killed)
2492 276784 Add support for IBM Power ISA 2.06 -- stage 3
2493 276987 gdbsrv: fix tests following recent commits
2494 277045 Valgrind crashes with unhandled DW_OP_ opcode 0x2a
2495 277199 The test_isa_2_06_part1.c in none/tests/ppc64 should be a symlink
2496 277471 Unhandled syscall: 340
2497 277610 valgrind crashes in VG_(lseek)(core_fd, phdrs[idx].p_offset, ...)
2498 277653 ARM: support Thumb2 PLD instruction
2499 277663 ARM: NEON float VMUL by scalar incorrect
2500 277689 ARM: tests for VSTn with register post-index are broken
2501 277694 ARM: BLX LR instruction broken in ARM mode
2502 277780 ARM: VMOV.F32 (immediate) instruction is broken
2503 278057 fuse filesystem syscall deadlocks
2504 278078 Unimplemented syscall 280 on ppc32
2505 278349 F_GETPIPE_SZ and F_SETPIPE_SZ Linux fcntl commands
2506 278454 VALGRIND_STACK_DEREGISTER has wrong output type
2507 278502 == 275284 (Valgrind confuses memcpy() and memmove())
2508 278892 gdbsrv: factorize gdb version handling, fix doc and typos
2509 279027 Support for MVCL and CLCL instruction
2510 279027 s390x: Provide support for CLCL and MVCL instructions
2511 279062 Remove a redundant check in the insn selector for ppc.
2512 279071 JDK creates PTEST with redundant REX.W prefix
2513 279212 gdbsrv: add monitor cmd v.info scheduler.
2514 279378 exp-ptrcheck: the 'impossible' happened on mkfifo call
2515 279698 memcheck discards valid-bits for packuswb
2516 279795 memcheck reports uninitialised values for mincore on amd64
2517 279994 Add support for IBM Power ISA 2.06 -- stage 3
2518 280083 mempolicy syscall check errors
2519 280290 vex amd64->IR: 0x66 0xF 0x38 0x28 0xC1 0x66 0xF 0x6F
2520 280710 s390x: config files for nightly builds
2521 280757 /tmp dir still used by valgrind even if TMPDIR is specified
2522 280965 Valgrind breaks fcntl locks when program does mmap
2523 281138 WARNING: unhandled syscall: 340
2524 281241 == 275168 (valgrind useless on Macos 10.7.1 Lion)
2525 281304 == 275168 (Darwin: dyld "cannot load inserted library")
2526 281305 == 275168 (unhandled syscall: unix:357 on Darwin 11.1)
2527 281468 s390x: handle do_clone and gcc clones in call traces
2528 281488 ARM: VFP register corruption
2529 281828 == 275284 (false memmove warning: "Source and destination overlap")
2530 281883 s390x: Fix system call wrapper for "clone".
2531 282105 generalise 'reclaimSuperBlock' to also reclaim splittable superblock
2532 282112 Unhandled instruction bytes: 0xDE 0xD9 0x9B 0xDF (fcompp)
2533 282238 SLES10: make check fails
2534 282979 strcasestr needs replacement with recent(>=2.12) glibc
2535 283000 vex amd64->IR: 0x66 0xF 0x3A 0xA 0xC0 0x9 0xF3 0xF
2536 283243 Regression in ppc64 memcheck tests
2537 283325 == 267997 (Darwin: V segfaults on startup when built with Xcode 4.0)
2538 283427 re-connect epoll_pwait syscall on ARM linux
2539 283600 gdbsrv: android: port vgdb.c
2540 283709 none/tests/faultstatus needs to account for page size
2541 284305 filter_gdb needs enhancement to work on ppc64
2542 284384 clang 3.1 -Wunused-value warnings in valgrind.h, memcheck.h
2543 284472 Thumb2 ROR.W encoding T2 not implemented
2544 284621 XML-escape process command line in XML output
2545 n-i-bz cachegrind/callgrind: handle CPUID information for Core iX Intel CPUs
2546 that have non-power-of-2 sizes (also AMDs)
2547 n-i-bz don't be spooked by libraries mashed by elfhack
2548 n-i-bz don't be spooked by libxul.so linked with gold
2549 n-i-bz improved checking for VALGRIND_CHECK_MEM_IS_DEFINED
2551 (3.7.0-TEST1: 27 October 2011, vex r2228, valgrind r12245)
2552 (3.7.0.RC1: 1 November 2011, vex r2231, valgrind r12257)
2553 (3.7.0: 5 November 2011, vex r2231, valgrind r12258)
2557 Release 3.6.1 (16 February 2011)
2558 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2559 3.6.1 is a bug fix release. It adds support for some SSE4
2560 instructions that were omitted in 3.6.0 due to lack of time. Initial
2561 support for glibc-2.13 has been added. A number of bugs causing
2562 crashing or assertion failures have been fixed.
2564 The following bugs have been fixed or resolved. Note that "n-i-bz"
2565 stands for "not in bugzilla" -- that is, a bug that was reported to us
2566 but never got a bugzilla entry. We encourage you to file bugs in
2567 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2568 mailing the developers (or mailing lists) directly -- bugs that are
2569 not entered into bugzilla tend to get forgotten about or ignored.
2571 To see details of a given bug, visit
2572 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2573 where XXXXXX is the bug number as listed below.
2575 188572 Valgrind on Mac should suppress setenv() mem leak
2576 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 (proper FX{SAVE,RSTOR} support)
2577 210481 vex amd64->IR: Assertion `sz == 2 || sz == 4' failed (REX.W POPQ)
2578 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
2579 250038 ppc64: Altivec LVSR and LVSL instructions fail their regtest
2580 254420 memory pool tracking broken
2581 254957 Test code failing to compile due to changes in memcheck.h
2582 255009 helgrind/drd: crash on chmod with invalid parameter
2583 255130 readdwarf3.c parse_type_DIE confused by GNAT Ada types
2584 255355 helgrind/drd: crash on threaded programs doing fork
2586 255418 (SSE4.x) rint call compiled with ICC
2587 255822 --gen-suppressions can create invalid files: "too many callers [...]"
2588 255888 closing valgrindoutput tag outputted to log-stream on error
2589 255963 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x9 0xDB 0x0 (ROUNDPD)
2590 255966 Slowness when using mempool annotations
2591 256387 vex x86->IR: 0xD4 0xA 0x2 0x7 (AAD and AAM)
2592 256600 super-optimized strcasecmp() false positive
2593 256669 vex amd64->IR: Unhandled LOOPNEL insn on amd64
2594 256968 (SSE4.x) vex amd64->IR: 0x66 0xF 0x38 0x10 0xD3 0x66 (BLENDVPx)
2595 257011 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0xE 0xFD 0xA0 (PBLENDW)
2596 257063 (SSE4.x) vex amd64->IR: 0x66 0xF 0x3A 0x8 0xC0 0x0 (ROUNDPS)
2597 257276 Missing case in memcheck --track-origins=yes
2598 258870 (SSE4.x) Add support for EXTRACTPS SSE 4.1 instruction
2599 261966 (SSE4.x) support for CRC32B and CRC32Q is lacking (also CRC32{W,L})
2600 262985 VEX regression in valgrind 3.6.0 in handling PowerPC VMX
2601 262995 (SSE4.x) crash when trying to valgrind gcc-snapshot (PCMPxSTRx $0)
2602 263099 callgrind_annotate counts Ir improperly [...]
2603 263877 undefined coprocessor instruction on ARMv7
2604 265964 configure FAIL with glibc-2.13
2605 n-i-bz Fix compile error w/ icc-12.x in guest_arm_toIR.c
2606 n-i-bz Docs: fix bogus descriptions for VALGRIND_CREATE_BLOCK et al
2607 n-i-bz Massif: don't assert on shmat() with --pages-as-heap=yes
2608 n-i-bz Bug fixes and major speedups for the exp-DHAT space profiler
2609 n-i-bz DRD: disable --free-is-write due to implementation difficulties
2611 (3.6.1: 16 February 2011, vex r2103, valgrind r11561).
2615 Release 3.6.0 (21 October 2010)
2616 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2617 3.6.0 is a feature release with many significant improvements and the
2618 usual collection of bug fixes.
2620 This release supports X86/Linux, AMD64/Linux, ARM/Linux, PPC32/Linux,
2621 PPC64/Linux, X86/Darwin and AMD64/Darwin. Support for recent distros
2622 and toolchain components (glibc 2.12, gcc 4.5, OSX 10.6) has been added.
2624 -------------------------
2626 Here are some highlights. Details are shown further down:
2628 * Support for ARM/Linux.
2630 * Support for recent Linux distros: Ubuntu 10.10 and Fedora 14.
2632 * Support for Mac OS X 10.6, both 32- and 64-bit executables.
2634 * Support for the SSE4.2 instruction set.
2636 * Enhancements to the Callgrind profiler, including the ability to
2637 handle CPUs with three levels of cache.
2639 * A new experimental heap profiler, DHAT.
2641 * A huge number of bug fixes and small enhancements.
2643 -------------------------
2645 Here are details of the above changes, together with descriptions of
2646 many other changes, and a list of fixed bugs.
2648 * ================== PLATFORM CHANGES =================
2650 * Support for ARM/Linux. Valgrind now runs on ARMv7 capable CPUs
2651 running Linux. It is known to work on Ubuntu 10.04, Ubuntu 10.10,
2652 and Maemo 5, so you can run Valgrind on your Nokia N900 if you want.
2654 This requires a CPU capable of running the ARMv7-A instruction set
2655 (Cortex A5, A8 and A9). Valgrind provides fairly complete coverage
2656 of the user space instruction set, including ARM and Thumb integer
2657 code, VFPv3, NEON and V6 media instructions. The Memcheck,
2658 Cachegrind and Massif tools work properly; other tools work to
2661 * Support for recent Linux distros (Ubuntu 10.10 and Fedora 14), along
2662 with support for recent releases of the underlying toolchain
2663 components, notably gcc-4.5 and glibc-2.12.
2665 * Support for Mac OS X 10.6, both 32- and 64-bit executables. 64-bit
2666 support also works much better on OS X 10.5, and is as solid as
2669 * Support for the SSE4.2 instruction set. SSE4.2 is supported in
2670 64-bit mode. In 32-bit mode, support is only available up to and
2671 including SSSE3. Some exceptions: SSE4.2 AES instructions are not
2672 supported in 64-bit mode, and 32-bit mode does in fact support the
2673 bare minimum SSE4 instructions to needed to run programs on Mac OS X
2674 10.6 on 32-bit targets.
2676 * Support for IBM POWER6 cpus has been improved. The Power ISA up to
2677 and including version 2.05 is supported.
2679 * ==================== TOOL CHANGES ====================
2681 * Cachegrind has a new processing script, cg_diff, which finds the
2682 difference between two profiles. It's very useful for evaluating
2683 the performance effects of a change in a program.
2685 Related to this change, the meaning of cg_annotate's (rarely-used)
2686 --threshold option has changed; this is unlikely to affect many
2687 people, if you do use it please see the user manual for details.
2689 * Callgrind now can do branch prediction simulation, similar to
2690 Cachegrind. In addition, it optionally can count the number of
2691 executed global bus events. Both can be used for a better
2692 approximation of a "Cycle Estimation" as derived event (you need to
2693 update the event formula in KCachegrind yourself).
2695 * Cachegrind and Callgrind now refer to the LL (last-level) cache
2696 rather than the L2 cache. This is to accommodate machines with
2697 three levels of caches -- if Cachegrind/Callgrind auto-detects the
2698 cache configuration of such a machine it will run the simulation as
2699 if the L2 cache isn't present. This means the results are less
2700 likely to match the true result for the machine, but
2701 Cachegrind/Callgrind's results are already only approximate, and
2702 should not be considered authoritative. The results are still
2703 useful for giving a general idea about a program's locality.
2705 * Massif has a new option, --pages-as-heap, which is disabled by
2706 default. When enabled, instead of tracking allocations at the level
2707 of heap blocks (as allocated with malloc/new/new[]), it instead
2708 tracks memory allocations at the level of memory pages (as mapped by
2709 mmap, brk, etc). Each mapped page is treated as its own block.
2710 Interpreting the page-level output is harder than the heap-level
2711 output, but this option is useful if you want to account for every
2712 byte of memory used by a program.
2714 * DRD has two new command-line options: --free-is-write and
2715 --trace-alloc. The former allows to detect reading from already freed
2716 memory, and the latter allows tracing of all memory allocations and
2719 * DRD has several new annotations. Custom barrier implementations can
2720 now be annotated, as well as benign races on static variables.
2722 * DRD's happens before / happens after annotations have been made more
2723 powerful, so that they can now also be used to annotate e.g. a smart
2724 pointer implementation.
2726 * Helgrind's annotation set has also been drastically improved, so as
2727 to provide to users a general set of annotations to describe locks,
2728 semaphores, barriers and condition variables. Annotations to
2729 describe thread-safe reference counted heap objects have also been
2732 * Memcheck has a new command-line option, --show-possibly-lost, which
2733 is enabled by default. When disabled, the leak detector will not
2734 show possibly-lost blocks.
2736 * A new experimental heap profiler, DHAT (Dynamic Heap Analysis Tool),
2737 has been added. DHAT keeps track of allocated heap blocks, and also
2738 inspects every memory reference to see which block (if any) is being
2739 accessed. This gives a lot of insight into block lifetimes,
2740 utilisation, turnover, liveness, and the location of hot and cold
2741 fields. You can use DHAT to do hot-field profiling.
2743 * ==================== OTHER CHANGES ====================
2745 * Improved support for unfriendly self-modifying code: the extra
2746 overhead incurred by --smc-check=all has been reduced by
2747 approximately a factor of 5 as compared with 3.5.0.
2749 * Ability to show directory names for source files in error messages.
2750 This is combined with a flexible mechanism for specifying which
2751 parts of the paths should be shown. This is enabled by the new flag
2754 * A new flag, --require-text-symbol, which will stop the run if a
2755 specified symbol is not found it a given shared object when it is
2756 loaded into the process. This makes advanced working with function
2757 intercepting and wrapping safer and more reliable.
2759 * Improved support for the Valkyrie GUI, version 2.0.0. GUI output
2760 and control of Valgrind is now available for the tools Memcheck and
2761 Helgrind. XML output from Valgrind is available for Memcheck,
2762 Helgrind and exp-Ptrcheck.
2764 * More reliable stack unwinding on amd64-linux, particularly in the
2765 presence of function wrappers, and with gcc-4.5 compiled code.
2767 * Modest scalability (performance improvements) for massive
2768 long-running applications, particularly for those with huge amounts
2771 * Support for analyzing programs running under Wine with has been
2772 improved. The header files <valgrind/valgrind.h>,
2773 <valgrind/memcheck.h> and <valgrind/drd.h> can now be used in
2774 Windows-programs compiled with MinGW or one of the Microsoft Visual
2777 * A rare but serious error in the 64-bit x86 CPU simulation was fixed.
2778 The 32-bit simulator was not affected. This did not occur often,
2779 but when it did would usually crash the program under test.
2782 * A large number of bugs were fixed. These are shown below.
2784 * A number of bugs were investigated, and were candidates for fixing,
2785 but are not fixed in 3.6.0, due to lack of developer time. They may
2786 get fixed in later releases. They are:
2788 194402 vex amd64->IR: 0x48 0xF 0xAE 0x4 0x24 0x49 (FXSAVE64)
2789 212419 false positive "lock order violated" (A+B vs A)
2790 213685 Undefined value propagates past dependency breaking instruction
2791 216837 Incorrect instrumentation of NSOperationQueue on Darwin
2792 237920 valgrind segfault on fork failure
2793 242137 support for code compiled by LLVM-2.8
2794 242423 Another unknown Intel cache config value
2795 243232 Inconsistent Lock Orderings report with trylock
2796 243483 ppc: callgrind triggers VEX assertion failure
2797 243935 Helgrind: implementation of ANNOTATE_HAPPENS_BEFORE() is wrong
2798 244677 Helgrind crash hg_main.c:616 (map_threads_lookup): Assertion
2800 246152 callgrind internal error after pthread_cancel on 32 Bit Linux
2801 249435 Analyzing wine programs with callgrind triggers a crash
2802 250038 ppc64: Altivec lvsr and lvsl instructions fail their regtest
2803 250065 Handling large allocations
2804 250101 huge "free" memory usage due to m_mallocfree.c
2805 "superblocks fragmentation"
2806 251569 vex amd64->IR: 0xF 0x1 0xF9 0x8B 0x4C 0x24 (RDTSCP)
2807 252091 Callgrind on ARM does not detect function returns correctly
2808 252600 [PATCH] Allow lhs to be a pointer for shl/shr
2809 254420 memory pool tracking broken
2810 n-i-bz support for adding symbols for JIT generated code
2813 The following bugs have been fixed or resolved. Note that "n-i-bz"
2814 stands for "not in bugzilla" -- that is, a bug that was reported to us
2815 but never got a bugzilla entry. We encourage you to file bugs in
2816 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
2817 mailing the developers (or mailing lists) directly -- bugs that are
2818 not entered into bugzilla tend to get forgotten about or ignored.
2820 To see details of a given bug, visit
2821 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
2822 where XXXXXX is the bug number as listed below.
2824 135264 dcbzl instruction missing
2826 153699 Valgrind should report unaligned reads with movdqa
2828 190429 Valgrind reports lost of errors in ld.so
2829 with x86_64 2.9.90 glibc
2830 197266 valgrind appears to choke on the xmms instruction
2832 197988 Crash when demangling very large symbol names
2833 202315 unhandled syscall: 332 (inotify_init1)
2834 203256 Add page-level profiling to Massif
2835 205093 dsymutil=yes needs quotes, locking (partial fix)
2836 205241 Snow Leopard 10.6 support (partial fix)
2837 206600 Leak checker fails to upgrade indirect blocks when their
2838 parent becomes reachable
2839 210935 port valgrind.h (not valgrind) to win32 so apps run under
2840 wine can make client requests
2841 211410 vex amd64->IR: 0x15 0xFF 0xFF 0x0 0x0 0x89
2842 within Linux ip-stack checksum functions
2843 212335 unhandled instruction bytes: 0xF3 0xF 0xBD 0xC0
2845 213685 Undefined value propagates past dependency breaking instruction
2847 215914 Valgrind inserts bogus empty environment variable
2849 219538 adjtimex syscall wrapper wrong in readonly adjtime mode
2850 222545 shmat fails under valgind on some arm targets
2851 222560 ARM NEON support
2854 232509 Docs build fails with formatting inside <title></title> elements
2856 235642 [PATCH] syswrap-linux.c: support evdev EVIOCG* ioctls
2857 236546 vex x86->IR: 0x66 0xF 0x3A 0xA
2858 237202 vex amd64->IR: 0xF3 0xF 0xB8 0xC0 0x49 0x3B
2859 237371 better support for VALGRIND_MALLOCLIKE_BLOCK
2860 237485 symlink (syscall 57) is not supported on Mac OS
2861 237723 sysno == 101 exp-ptrcheck: the 'impossible' happened:
2863 238208 is_just_below_ESP doesn't take into account red-zone
2864 238345 valgrind passes wrong $0 when executing a shell script
2865 238679 mq_timedreceive syscall doesn't flag the reception buffer
2867 238696 fcntl command F_DUPFD_CLOEXEC not supported
2868 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
2869 238713 unhandled instruction bytes: 0x66 0xF 0x29 0xC6
2870 238745 3.5.0 Make fails on PPC Altivec opcodes, though configure
2872 239992 vex amd64->IR: 0x48 0xF 0xC4 0xC1 0x0 0x48
2878 242606 unhandled syscall: setegid (in Ptrcheck)
2879 242814 Helgrind "Impossible has happened" during
2880 QApplication::initInstance();
2881 243064 Valgrind attempting to read debug information from iso
2882 243270 Make stack unwinding in Valgrind wrappers more reliable
2883 243884 exp-ptrcheck: the 'impossible happened: unhandled syscall
2884 sysno = 277 (mq_open)
2885 244009 exp-ptrcheck unknown syscalls in analyzing lighttpd
2886 244493 ARM VFP d16-d31 registers support
2887 244670 add support for audit_session_self syscall on Mac OS 10.6
2888 244921 The xml report of helgrind tool is not well format
2889 244923 In the xml report file, the <preamble> not escape the
2890 xml char, eg '<','&','>'
2891 245535 print full path names in plain text reports
2892 245925 x86-64 red zone handling problem
2893 246258 Valgrind not catching integer underruns + new [] s
2894 246311 reg/reg cmpxchg doesn't work on amd64
2895 246549 unhandled syscall unix:277 while testing 32-bit Darwin app
2896 246888 Improve Makefile.vex.am
2897 247510 [OS X 10.6] Memcheck reports unaddressable bytes passed
2898 to [f]chmod_extended
2899 247526 IBM POWER6 (ISA 2.05) support is incomplete
2900 247561 Some leak testcases fails due to reachable addresses in
2902 247875 sizeofIRType to handle Ity_I128
2903 247894 [PATCH] unhandled syscall sys_readahead
2904 247980 Doesn't honor CFLAGS passed to configure
2905 248373 darwin10.supp is empty in the trunk
2906 248822 Linux FIBMAP ioctl has int parameter instead of long
2907 248893 [PATCH] make readdwarf.c big endianess safe to enable
2908 unwinding on big endian systems
2909 249224 Syscall 336 not supported (SYS_proc_info)
2911 249775 Incorrect scheme for detecting NEON capabilities of host CPU
2912 249943 jni JVM init fails when using valgrind
2913 249991 Valgrind incorrectly declares AESKEYGENASSIST support
2915 249996 linux/arm: unhandled syscall: 181 (__NR_pwrite64)
2916 250799 frexp$fenv_access_off function generates SIGILL
2917 250998 vex x86->IR: unhandled instruction bytes: 0x66 0x66 0x66 0x2E
2918 251251 support pclmulqdq insn
2919 251362 valgrind: ARM: attach to debugger either fails or provokes
2921 251674 Unhandled syscall 294
2924 254257 Add support for debugfiles found by build-id
2925 254550 [PATCH] Implement DW_ATE_UTF (DWARF4)
2926 254646 Wrapped functions cause stack misalignment on OS X
2927 (and possibly Linux)
2928 254556 ARM: valgrinding anything fails with SIGSEGV for 0xFFFF0FA0
2930 (3.6.0: 21 October 2010, vex r2068, valgrind r11471).
2934 Release 3.5.0 (19 August 2009)
2935 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2936 3.5.0 is a feature release with many significant improvements and the
2937 usual collection of bug fixes. The main improvement is that Valgrind
2938 now works on Mac OS X.
2940 This release supports X86/Linux, AMD64/Linux, PPC32/Linux, PPC64/Linux
2941 and X86/Darwin. Support for recent distros and toolchain components
2942 (glibc 2.10, gcc 4.5) has been added.
2944 -------------------------
2946 Here is a short summary of the changes. Details are shown further
2949 * Support for Mac OS X (10.5.x).
2951 * Improvements and simplifications to Memcheck's leak checker.
2953 * Clarification and simplifications in various aspects of Valgrind's
2956 * XML output for Helgrind and Ptrcheck.
2958 * Performance and stability improvements for Helgrind and DRD.
2960 * Genuinely atomic support for x86/amd64/ppc atomic instructions.
2962 * A new experimental tool, BBV, useful for computer architecture
2965 * Improved Wine support, including ability to read Windows PDB
2968 -------------------------
2970 Here are details of the above changes, followed by descriptions of
2971 many other minor changes, and a list of fixed bugs.
2974 * Valgrind now runs on Mac OS X. (Note that Mac OS X is sometimes
2975 called "Darwin" because that is the name of the OS core, which is the
2976 level that Valgrind works at.)
2980 - It requires OS 10.5.x (Leopard). Porting to 10.4.x is not planned
2981 because it would require work and 10.4 is only becoming less common.
2983 - 32-bit programs on x86 and AMD64 (a.k.a x86-64) machines are supported
2984 fairly well. For 10.5.x, 32-bit programs are the default even on
2985 64-bit machines, so it handles most current programs.
2987 - 64-bit programs on x86 and AMD64 (a.k.a x86-64) machines are not
2988 officially supported, but simple programs at least will probably work.
2989 However, start-up is slow.
2991 - PowerPC machines are not supported.
2993 Things that don't work:
2995 - The Ptrcheck tool.
2997 - Objective-C garbage collection.
3001 - If you have Rogue Amoeba's "Instant Hijack" program installed,
3002 Valgrind will fail with a SIGTRAP at start-up. See
3003 https://bugs.kde.org/show_bug.cgi?id=193917 for details and a
3008 - You will likely find --dsymutil=yes a useful option, as error
3009 messages may be imprecise without it.
3011 - Mac OS X support is new and therefore will be less robust than the
3012 Linux support. Please report any bugs you find.
3014 - Threaded programs may run more slowly than on Linux.
3016 Many thanks to Greg Parker for developing this port over several years.
3019 * Memcheck's leak checker has been improved.
3021 - The results for --leak-check=summary now match the summary results
3022 for --leak-check=full. Previously they could differ because
3023 --leak-check=summary counted "indirectly lost" blocks and
3024 "suppressed" blocks as "definitely lost".
3026 - Blocks that are only reachable via at least one interior-pointer,
3027 but are directly pointed to by a start-pointer, were previously
3028 marked as "still reachable". They are now correctly marked as
3031 - The default value for the --leak-resolution option has been
3032 changed from "low" to "high". In general, this means that more
3033 leak reports will be produced, but each leak report will describe
3034 fewer leaked blocks.
3036 - With --leak-check=full, "definitely lost" and "possibly lost"
3037 leaks are now considered as proper errors, ie. they are counted
3038 for the "ERROR SUMMARY" and affect the behaviour of
3039 --error-exitcode. These leaks are not counted as errors if
3040 --leak-check=summary is specified, however.
3042 - Documentation for the leak checker has been improved.
3045 * Various aspects of Valgrind's text output have changed.
3047 - Valgrind's start-up message has changed. It is shorter but also
3048 includes the command being run, which makes it easier to use
3049 --trace-children=yes. An example:
3051 - Valgrind's shut-down messages have also changed. This is most
3052 noticeable with Memcheck, where the leak summary now occurs before
3053 the error summary. This change was necessary to allow leaks to be
3054 counted as proper errors (see the description of the leak checker
3055 changes above for more details). This was also necessary to fix a
3056 longstanding bug in which uses of suppressions against leaks were
3057 not "counted", leading to difficulties in maintaining suppression
3058 files (see https://bugs.kde.org/show_bug.cgi?id=186790).
3060 - Behavior of -v has changed. In previous versions, -v printed out
3061 a mixture of marginally-user-useful information, and tool/core
3062 statistics. The statistics printing has now been moved to its own
3063 flag, --stats=yes. This means -v is less verbose and more likely
3064 to convey useful end-user information.
3066 - The format of some (non-XML) stack trace entries has changed a
3067 little. Previously there were six possible forms:
3069 0x80483BF: really (a.c:20)
3070 0x80483BF: really (in /foo/a.out)
3072 0x80483BF: (within /foo/a.out)
3073 0x80483BF: ??? (a.c:20)
3076 The third and fourth of these forms have been made more consistent
3077 with the others. The six possible forms are now:
3079 0x80483BF: really (a.c:20)
3080 0x80483BF: really (in /foo/a.out)
3081 0x80483BF: really (in ???)
3082 0x80483BF: ??? (in /foo/a.out)
3083 0x80483BF: ??? (a.c:20)
3086 Stack traces produced when --xml=yes is specified are different
3090 * Helgrind and Ptrcheck now support XML output, so they can be used
3091 from GUI tools. Also, the XML output mechanism has been
3094 - The XML format has been overhauled and generalised, so it is more
3095 suitable for error reporting tools in general. The Memcheck
3096 specific aspects of it have been removed. The new format, which
3097 is an evolution of the old format, is described in
3098 docs/internals/xml-output-protocol4.txt.
3100 - Memcheck has been updated to use the new format.
3102 - Helgrind and Ptrcheck are now able to emit output in this format.
3104 - The XML output mechanism has been overhauled. XML is now output
3105 to its own file descriptor, which means that:
3107 * Valgrind can output text and XML independently.
3109 * The longstanding problem of XML output being corrupted by
3110 unexpected un-tagged text messages is solved.
3112 As before, the destination for text output is specified using
3113 --log-file=, --log-fd= or --log-socket=.
3115 As before, XML output for a tool is enabled using --xml=yes.
3117 Because there's a new XML output channel, the XML output
3118 destination is now specified by --xml-file=, --xml-fd= or
3121 Initial feedback has shown this causes some confusion. To
3122 clarify, the two envisaged usage scenarios are:
3124 (1) Normal text output. In this case, do not specify --xml=yes
3125 nor any of --xml-file=, --xml-fd= or --xml-socket=.
3127 (2) XML output. In this case, specify --xml=yes, and one of
3128 --xml-file=, --xml-fd= or --xml-socket= to select the XML
3129 destination, one of --log-file=, --log-fd= or --log-socket=
3130 to select the destination for any remaining text messages,
3131 and, importantly, -q.
3133 -q makes Valgrind completely silent on the text channel,
3134 except in the case of critical failures, such as Valgrind
3135 itself segfaulting, or failing to read debugging information.
3136 Hence, in this scenario, it suffices to check whether or not
3137 any output appeared on the text channel. If yes, then it is
3138 likely to be a critical error which should be brought to the
3139 attention of the user. If no (the text channel produced no
3140 output) then it can be assumed that the run was successful.
3142 This allows GUIs to make the critical distinction they need to
3143 make (did the run fail or not?) without having to search or
3144 filter the text output channel in any way.
3146 It is also recommended to use --child-silent-after-fork=yes in
3150 * Improvements and changes in Helgrind:
3152 - XML output, as described above
3154 - Checks for consistent association between pthread condition
3155 variables and their associated mutexes are now performed.
3157 - pthread_spinlock functions are supported.
3159 - Modest performance improvements.
3161 - Initial (skeletal) support for describing the behaviour of
3162 non-POSIX synchronisation objects through ThreadSanitizer
3163 compatible ANNOTATE_* macros.
3165 - More controllable tradeoffs between performance and the level of
3166 detail of "previous" accesses in a race. There are now three
3169 * --history-level=full. This is the default, and was also the
3170 default in 3.4.x. It shows both stacks involved in a race, but
3171 requires a lot of memory and can be very slow in programs that
3172 do many inter-thread synchronisation events.
3174 * --history-level=none. This only shows the later stack involved
3175 in a race. This can be much faster than --history-level=full,
3176 but makes it much more difficult to find the other access
3177 involved in the race.
3179 The new intermediate setting is
3181 * --history-level=approx
3183 For the earlier (other) access, two stacks are presented. The
3184 earlier access is guaranteed to be somewhere in between the two
3185 program points denoted by those stacks. This is not as useful
3186 as showing the exact stack for the previous access (as per
3187 --history-level=full), but it is better than nothing, and it's
3188 almost as fast as --history-level=none.
3191 * New features and improvements in DRD:
3193 - The error messages printed by DRD are now easier to interpret.
3194 Instead of using two different numbers to identify each thread
3195 (Valgrind thread ID and DRD thread ID), DRD does now identify
3196 threads via a single number (the DRD thread ID). Furthermore
3197 "first observed at" information is now printed for all error
3198 messages related to synchronization objects.
3200 - Added support for named semaphores (sem_open() and sem_close()).
3202 - Race conditions between pthread_barrier_wait() and
3203 pthread_barrier_destroy() calls are now reported.
3205 - Added support for custom allocators through the macros
3206 VALGRIND_MALLOCLIKE_BLOCK() VALGRIND_FREELIKE_BLOCK() (defined in
3207 in <valgrind/valgrind.h>). An alternative for these two macros is
3208 the new client request VG_USERREQ__DRD_CLEAN_MEMORY (defined in
3211 - Added support for annotating non-POSIX synchronization objects
3212 through several new ANNOTATE_*() macros.
3214 - OpenMP: added support for the OpenMP runtime (libgomp) included
3215 with gcc versions 4.3.0 and 4.4.0.
3219 - Added two new command-line options (--first-race-only and
3220 --segment-merging-interval).
3223 * Genuinely atomic support for x86/amd64/ppc atomic instructions
3225 Valgrind will now preserve (memory-access) atomicity of LOCK-
3226 prefixed x86/amd64 instructions, and any others implying a global
3227 bus lock. Ditto for PowerPC l{w,d}arx/st{w,d}cx. instructions.
3229 This means that Valgrinded processes will "play nicely" in
3230 situations where communication with other processes, or the kernel,
3231 is done through shared memory and coordinated with such atomic
3232 instructions. Prior to this change, such arrangements usually
3233 resulted in hangs, races or other synchronisation failures, because
3234 Valgrind did not honour atomicity of such instructions.
3237 * A new experimental tool, BBV, has been added. BBV generates basic
3238 block vectors for use with the SimPoint analysis tool, which allows
3239 a program's overall behaviour to be approximated by running only a
3240 fraction of it. This is useful for computer architecture
3241 researchers. You can run BBV by specifying --tool=exp-bbv (the
3242 "exp-" prefix is short for "experimental"). BBV was written by
3246 * Modestly improved support for running Windows applications under
3247 Wine. In particular, initial support for reading Windows .PDB debug
3248 information has been added.
3251 * A new Memcheck client request VALGRIND_COUNT_LEAK_BLOCKS has been
3252 added. It is similar to VALGRIND_COUNT_LEAKS but counts blocks
3256 * The Valgrind client requests VALGRIND_PRINTF and
3257 VALGRIND_PRINTF_BACKTRACE have been changed slightly. Previously,
3258 the string was always printed immediately on its own line. Now, the
3259 string will be added to a buffer but not printed until a newline is
3260 encountered, or other Valgrind output is printed (note that for
3261 VALGRIND_PRINTF_BACKTRACE, the back-trace itself is considered
3262 "other Valgrind output"). This allows you to use multiple
3263 VALGRIND_PRINTF calls to build up a single output line, and also to
3264 print multiple output lines with a single request (by embedding
3265 multiple newlines in the string).
3268 * The graphs drawn by Massif's ms_print program have changed slightly:
3270 - The half-height chars '.' and ',' are no longer drawn, because
3271 they are confusing. The --y option can be used if the default
3272 y-resolution is not high enough.
3274 - Horizontal lines are now drawn after the top of a snapshot if
3275 there is a gap until the next snapshot. This makes it clear that
3276 the memory usage has not dropped to zero between snapshots.
3279 * Something that happened in 3.4.0, but wasn't clearly announced: the
3280 option --read-var-info=yes can be used by some tools (Memcheck,
3281 Helgrind and DRD). When enabled, it causes Valgrind to read DWARF3
3282 variable type and location information. This makes those tools
3283 start up more slowly and increases memory consumption, but
3284 descriptions of data addresses in error messages become more
3288 * exp-Omega, an experimental instantaneous leak-detecting tool, was
3289 disabled in 3.4.0 due to a lack of interest and maintenance,
3290 although the source code was still in the distribution. The source
3291 code has now been removed from the distribution. For anyone
3292 interested, the removal occurred in SVN revision r10247.
3295 * Some changes have been made to the build system.
3297 - VEX/ is now integrated properly into the build system. This means
3298 that dependency tracking within VEX/ now works properly, "make
3299 install" will work without requiring "make" before it, and
3300 parallel builds (ie. 'make -j') now work (previously a
3301 .NOTPARALLEL directive was used to serialize builds, ie. 'make -j'
3302 was effectively ignored).
3304 - The --with-vex configure option has been removed. It was of
3305 little use and removing it simplified the build system.
3307 - The location of some install files has changed. This should not
3308 affect most users. Those who might be affected:
3310 * For people who use Valgrind with MPI programs, the installed
3311 libmpiwrap.so library has moved from
3312 $(INSTALL)/<platform>/libmpiwrap.so to
3313 $(INSTALL)/libmpiwrap-<platform>.so.
3315 * For people who distribute standalone Valgrind tools, the
3316 installed libraries such as $(INSTALL)/<platform>/libcoregrind.a
3317 have moved to $(INSTALL)/libcoregrind-<platform>.a.
3319 These changes simplify the build system.
3321 - Previously, all the distributed suppression (*.supp) files were
3322 installed. Now, only default.supp is installed. This should not
3323 affect users as the other installed suppression files were not
3324 read; the fact that they were installed was a mistake.
3327 * KNOWN LIMITATIONS:
3329 - Memcheck is unusable with the Intel compiler suite version 11.1,
3330 when it generates code for SSE2-and-above capable targets. This
3331 is because of icc's use of highly optimised inlined strlen
3332 implementations. It causes Memcheck to report huge numbers of
3333 false errors even in simple programs. Helgrind and DRD may also
3336 Versions 11.0 and earlier may be OK, but this has not been
3340 The following bugs have been fixed or resolved. Note that "n-i-bz"
3341 stands for "not in bugzilla" -- that is, a bug that was reported to us
3342 but never got a bugzilla entry. We encourage you to file bugs in
3343 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3344 mailing the developers (or mailing lists) directly -- bugs that are
3345 not entered into bugzilla tend to get forgotten about or ignored.
3347 To see details of a given bug, visit
3348 https://bugs.kde.org/show_bug.cgi?id=XXXXXX
3349 where XXXXXX is the bug number as listed below.
3351 84303 How about a LockCheck tool?
3352 91633 dereference of null ptr in vgPlain_st_basetype
3353 97452 Valgrind doesn't report any pthreads problems
3354 100628 leak-check gets assertion failure when using
3355 VALGRIND_MALLOCLIKE_BLOCK on malloc()ed memory
3356 108528 NPTL pthread cleanup handlers not called
3357 110126 Valgrind 2.4.1 configure.in tramples CFLAGS
3358 110128 mallinfo is not implemented...
3359 110770 VEX: Generated files not always updated when making valgrind
3360 111102 Memcheck: problems with large (memory footprint) applications
3361 115673 Vex's decoder should never assert
3362 117564 False positive: Syscall param clone(child_tidptr) contains
3363 uninitialised byte(s)
3364 119404 executing ssh from inside valgrind fails
3365 133679 Callgrind does not write path names to sources with dwarf debug
3367 135847 configure.in problem with non gnu compilers (and possible fix)
3368 136154 threads.c:273 (vgCallgrind_post_signal): Assertion
3369 '*(vgCallgrind_current_fn_stack.top) == 0' failed.
3370 136230 memcheck reports "possibly lost", should be "still reachable"
3371 137073 NULL arg to MALLOCLIKE_BLOCK causes crash
3372 137904 Valgrind reports a memory leak when using POSIX threads,
3374 139076 valgrind VT_GETSTATE error
3375 142228 complaint of elf_dynamic_do_rela in trivial usage
3376 145347 spurious warning with USBDEVFS_REAPURB
3377 148441 (wine) can't find memory leak in Wine, win32 binary
3379 148742 Leak-check fails assert on exit
3380 149878 add (proper) check for calloc integer overflow
3381 150606 Call graph is broken when using callgrind control
3382 152393 leak errors produce an exit code of 0. I need some way to
3383 cause leak errors to result in a nonzero exit code.
3384 157154 documentation (leak-resolution doc speaks about num-callers
3385 def=4) + what is a loss record
3386 159501 incorrect handling of ALSA ioctls
3387 162020 Valgrinding an empty/zero-byte file crashes valgrind
3388 162482 ppc: Valgrind crashes while reading stabs information
3389 162718 x86: avoid segment selector 0 in sys_set_thread_area()
3390 163253 (wine) canonicaliseSymtab forgot some fields in DiSym
3391 163560 VEX/test_main.c is missing from valgrind-3.3.1
3392 164353 malloc_usable_size() doesn't return a usable size
3393 165468 Inconsistent formatting in memcheck manual -- please fix
3394 169505 main.c:286 (endOfInstr):
3395 Assertion 'ii->cost_offset == *cost_offset' failed
3396 177206 Generate default.supp during compile instead of configure
3397 177209 Configure valt_load_address based on arch+os
3398 177305 eventfd / syscall 323 patch lost
3399 179731 Tests fail to build because of inlining of non-local asm labels
3400 181394 helgrind: libhb_core.c:3762 (msm_write): Assertion
3401 'ordxx == POrd_EQ || ordxx == POrd_LT' failed.
3402 181594 Bogus warning for empty text segment
3403 181707 dwarf doesn't require enumerations to have name
3404 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3405 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3406 Assertion '!already_present' failed.
3407 185359 exp-ptrcheck: unhandled syscall getresuid()
3408 185794 "WARNING: unhandled syscall: 285" (fallocate) on x86_64
3409 185816 Valgrind is unable to handle debug info for files with split
3410 debug info that are prelinked afterwards
3411 185980 [darwin] unhandled syscall: sem_open
3412 186238 bbToIR_AMD64: disInstr miscalculated next %rip
3413 186507 exp-ptrcheck unhandled syscalls prctl, etc.
3414 186790 Suppression pattern used for leaks are not reported
3415 186796 Symbols with length>200 in suppression files are ignored
3416 187048 drd: mutex PTHREAD_PROCESS_SHARED attribute missinterpretation
3417 187416 exp-ptrcheck: support for __NR_{setregid,setreuid,setresuid}
3418 188038 helgrind: hg_main.c:926: mk_SHVAL_fail: the 'impossible' happened
3419 188046 bashisms in the configure script
3420 188127 amd64->IR: unhandled instruction bytes: 0xF0 0xF 0xB0 0xA
3421 188161 memcheck: --track-origins=yes asserts "mc_machine.c:672
3422 (get_otrack_shadow_offset_wrk): the 'impossible' happened."
3423 188248 helgrind: pthread_cleanup_push, pthread_rwlock_unlock,
3424 assertion fail "!lock->heldBy"
3425 188427 Add support for epoll_create1 (with patch)
3426 188530 Support for SIOCGSTAMPNS
3427 188560 Include valgrind.spec in the tarball
3428 188572 Valgrind on Mac should suppress setenv() mem leak
3429 189054 Valgrind fails to build because of duplicate non-local asm labels
3430 189737 vex amd64->IR: unhandled instruction bytes: 0xAC
3431 189762 epoll_create syscall not handled (--tool=exp-ptrcheck)
3432 189763 drd assertion failure: s_threadinfo[tid].is_recording
3433 190219 unhandled syscall: 328 (x86-linux)
3434 190391 dup of 181394; see above
3435 190429 Valgrind reports lots of errors in ld.so with x86_64 2.9.90 glibc
3436 190820 No debug information on powerpc-linux
3437 191095 PATCH: Improve usbdevfs ioctl handling
3438 191182 memcheck: VALGRIND_LEAK_CHECK quadratic when big nr of chunks
3440 191189 --xml=yes should obey --gen-suppressions=all
3441 191192 syslog() needs a suppression on macosx
3442 191271 DARWIN: WARNING: unhandled syscall: 33554697 a.k.a.: 265
3443 191761 getrlimit on MacOSX
3444 191992 multiple --fn-skip only works sometimes; dependent on order
3445 192634 V. reports "aspacem sync_check_mapping_callback:
3446 segment mismatch" on Darwin
3447 192954 __extension__ missing on 2 client requests
3448 194429 Crash at start-up with glibc-2.10.1 and linux-2.6.29
3449 194474 "INSTALL" file has different build instructions than "README"
3450 194671 Unhandled syscall (sem_wait?) from mac valgrind
3451 195069 memcheck: reports leak (memory still reachable) for
3453 195169 drd: (vgDrd_barrier_post_wait):
3454 Assertion 'r->sg[p->post_iteration]' failed.
3455 195268 valgrind --log-file doesn't accept ~/...
3456 195838 VEX abort: LibVEX_N_SPILL_BYTES too small for CPUID boilerplate
3457 195860 WARNING: unhandled syscall: unix:223
3458 196528 need a error suppression for pthread_rwlock_init under os x?
3459 197227 Support aio_* syscalls on Darwin
3460 197456 valgrind should reject --suppressions=(directory)
3461 197512 DWARF2 CFI reader: unhandled CFI instruction 0:10
3462 197591 unhandled syscall 27 (mincore)
3463 197793 Merge DCAS branch to the trunk == 85756, 142103
3464 197794 Avoid duplicate filenames in Vex
3465 197898 make check fails on current SVN
3466 197901 make check fails also under exp-ptrcheck in current SVN
3467 197929 Make --leak-resolution=high the default
3468 197930 Reduce spacing between leak reports
3469 197933 Print command line of client at start-up, and shorten preamble
3470 197966 unhandled syscall 205 (x86-linux, --tool=exp-ptrcheck)
3471 198395 add BBV to the distribution as an experimental tool
3472 198624 Missing syscalls on Darwin: 82, 167, 281, 347
3473 198649 callgrind_annotate doesn't cumulate counters
3474 199338 callgrind_annotate sorting/thresholds are broken for all but Ir
3475 199977 Valgrind complains about an unrecognized instruction in the
3476 atomic_incs test program
3477 200029 valgrind isn't able to read Fedora 12 debuginfo
3478 200760 darwin unhandled syscall: unix:284
3479 200827 DRD doesn't work on Mac OS X
3480 200990 VG_(read_millisecond_timer)() does not work correctly
3481 201016 Valgrind does not support pthread_kill() on Mac OS
3482 201169 Document --read-var-info
3483 201323 Pre-3.5.0 performance sanity checking
3484 201384 Review user manual for the 3.5.0 release
3485 201585 mfpvr not implemented on ppc
3486 201708 tests failing because x86 direction flag is left set
3487 201757 Valgrind doesn't handle any recent sys_futex additions
3488 204377 64-bit valgrind can not start a shell script
3489 (with #!/path/to/shell) if the shell is a 32-bit executable
3490 n-i-bz drd: fixed assertion failure triggered by mutex reinitialization.
3491 n-i-bz drd: fixed a bug that caused incorrect messages to be printed
3492 about memory allocation events with memory access tracing enabled
3493 n-i-bz drd: fixed a memory leak triggered by vector clock deallocation
3495 (3.5.0: 19 Aug 2009, vex r1913, valgrind r10846).
3499 Release 3.4.1 (28 February 2009)
3500 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3501 3.4.1 is a bug-fix release that fixes some regressions and assertion
3502 failures in debug info reading in 3.4.0, most notably incorrect stack
3503 traces on amd64-linux on older (glibc-2.3 based) systems. Various
3504 other debug info problems are also fixed. A number of bugs in the
3505 exp-ptrcheck tool introduced in 3.4.0 have been fixed.
3507 In view of the fact that 3.4.0 contains user-visible regressions
3508 relative to 3.3.x, upgrading to 3.4.1 is recommended. Packagers are
3509 encouraged to ship 3.4.1 in preference to 3.4.0.
3511 The fixed bugs are as follows. Note that "n-i-bz" stands for "not in
3512 bugzilla" -- that is, a bug that was reported to us but never got a
3513 bugzilla entry. We encourage you to file bugs in bugzilla
3514 (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than mailing the
3515 developers (or mailing lists) directly -- bugs that are not entered
3516 into bugzilla tend to get forgotten about or ignored.
3518 n-i-bz Fix various bugs reading icc-11 generated debug info
3519 n-i-bz Fix various bugs reading gcc-4.4 generated debug info
3520 n-i-bz Preliminary support for glibc-2.10 / Fedora 11
3521 n-i-bz Cachegrind and Callgrind: handle non-power-of-two cache sizes,
3522 so as to support (eg) 24k Atom D1 and Core2 with 3/6/12MB L2.
3523 179618 exp-ptrcheck crashed / exit prematurely
3524 179624 helgrind: false positive races with pthread_create and
3525 recv/open/close/read
3526 134207 pkg-config output contains @VG_PLATFORM@
3527 176926 floating point exception at valgrind startup with PPC 440EPX
3528 181594 Bogus warning for empty text segment
3529 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant rex prefixes)
3530 181707 Dwarf3 doesn't require enumerations to have name
3531 185038 exp-ptrcheck: "unhandled syscall: 285" (fallocate) on x86_64
3532 185050 exp-ptrcheck: sg_main.c:727 (add_block_to_GlobalTree):
3533 Assertion '!already_present' failed.
3534 185359 exp-ptrcheck unhandled syscall getresuid()
3536 (3.4.1.RC1: 24 Feb 2008, vex r1884, valgrind r9253).
3537 (3.4.1: 28 Feb 2008, vex r1884, valgrind r9293).
3541 Release 3.4.0 (2 January 2009)
3542 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3543 3.4.0 is a feature release with many significant improvements and the
3544 usual collection of bug fixes. This release supports X86/Linux,
3545 AMD64/Linux, PPC32/Linux and PPC64/Linux. Support for recent distros
3546 (using gcc 4.4, glibc 2.8 and 2.9) has been added.
3548 3.4.0 brings some significant tool improvements. Memcheck can now
3549 report the origin of uninitialised values, the thread checkers
3550 Helgrind and DRD are much improved, and we have a new experimental
3551 tool, exp-Ptrcheck, which is able to detect overruns of stack and
3552 global arrays. In detail:
3554 * Memcheck is now able to track the origin of uninitialised values.
3555 When it reports an uninitialised value error, it will try to show
3556 the origin of the value, as either a heap or stack allocation.
3557 Origin tracking is expensive and so is not enabled by default. To
3558 use it, specify --track-origins=yes. Memcheck's speed will be
3559 essentially halved, and memory usage will be significantly
3560 increased. Nevertheless it can drastically reduce the effort
3561 required to identify the root cause of uninitialised value errors,
3562 and so is often a programmer productivity win, despite running more
3565 * A version (1.4.0) of the Valkyrie GUI, that works with Memcheck in
3566 3.4.0, will be released shortly.
3568 * Helgrind's race detection algorithm has been completely redesigned
3569 and reimplemented, to address usability and scalability concerns:
3571 - The new algorithm has a lower false-error rate: it is much less
3572 likely to report races that do not really exist.
3574 - Helgrind will display full call stacks for both accesses involved
3575 in a race. This makes it easier to identify the root causes of
3578 - Limitations on the size of program that can run have been removed.
3580 - Performance has been modestly improved, although that is very
3583 - Direct support for Qt4 threading has been added.
3585 - pthread_barriers are now directly supported.
3587 - Helgrind works well on all supported Linux targets.
3589 * The DRD thread debugging tool has seen major improvements:
3591 - Greatly improved performance and significantly reduced memory
3594 - Support for several major threading libraries (Boost.Thread, Qt4,
3595 glib, OpenMP) has been added.
3597 - Support for atomic instructions, POSIX semaphores, barriers and
3598 reader-writer locks has been added.
3600 - Works now on PowerPC CPUs too.
3602 - Added support for printing thread stack usage at thread exit time.
3604 - Added support for debugging lock contention.
3606 - Added a manual for Drd.
3608 * A new experimental tool, exp-Ptrcheck, has been added. Ptrcheck
3609 checks for misuses of pointers. In that sense it is a bit like
3610 Memcheck. However, Ptrcheck can do things Memcheck can't: it can
3611 detect overruns of stack and global arrays, it can detect
3612 arbitrarily far out-of-bounds accesses to heap blocks, and it can
3613 detect accesses heap blocks that have been freed a very long time
3614 ago (millions of blocks in the past).
3616 Ptrcheck currently works only on x86-linux and amd64-linux. To use
3617 it, use --tool=exp-ptrcheck. A simple manual is provided, as part
3618 of the main Valgrind documentation. As this is an experimental
3619 tool, we would be particularly interested in hearing about your
3620 experiences with it.
3622 * exp-Omega, an experimental instantaneous leak-detecting tool, is no
3623 longer built by default, although the code remains in the repository
3624 and the tarball. This is due to three factors: a perceived lack of
3625 users, a lack of maintenance, and concerns that it may not be
3626 possible to achieve reliable operation using the existing design.
3628 * As usual, support for the latest Linux distros and toolchain
3629 components has been added. It should work well on Fedora Core 10,
3630 OpenSUSE 11.1 and Ubuntu 8.10. gcc-4.4 (in its current pre-release
3631 state) is supported, as is glibc-2.9. The C++ demangler has been
3632 updated so as to work well with C++ compiled by even the most recent
3635 * You can now use frame-level wildcards in suppressions. This was a
3636 frequently-requested enhancement. A line "..." in a suppression now
3637 matches zero or more frames. This makes it easier to write
3638 suppressions which are precise yet insensitive to changes in
3641 * 3.4.0 adds support on x86/amd64 for the SSSE3 instruction set.
3643 * Very basic support for IBM Power6 has been added (64-bit processes only).
3645 * Valgrind is now cross-compilable. For example, it is possible to
3646 cross compile Valgrind on an x86/amd64-linux host, so that it runs
3647 on a ppc32/64-linux target.
3649 * You can set the main thread's stack size at startup using the
3650 new --main-stacksize= flag (subject of course to ulimit settings).
3651 This is useful for running apps that need a lot of stack space.
3653 * The limitation that you can't use --trace-children=yes together
3654 with --db-attach=yes has been removed.
3656 * The following bugs have been fixed. Note that "n-i-bz" stands for
3657 "not in bugzilla" -- that is, a bug that was reported to us but
3658 never got a bugzilla entry. We encourage you to file bugs in
3659 bugzilla (http://bugs.kde.org/enter_valgrind_bug.cgi) rather than
3660 mailing the developers (or mailing lists) directly.
3662 n-i-bz Make return types for some client requests 64-bit clean
3663 n-i-bz glibc 2.9 support
3664 n-i-bz ignore unsafe .valgrindrc's (CVE-2008-4865)
3665 n-i-bz MPI_Init(0,0) is valid but libmpiwrap.c segfaults
3666 n-i-bz Building in an env without gdb gives bogus gdb attach
3667 92456 Tracing the origin of uninitialised memory
3668 106497 Valgrind does not demangle some C++ template symbols
3670 151612 Suppression with "..." (frame-level wildcards in .supp files)
3671 156404 Unable to start oocalc under memcheck on openSUSE 10.3 (64-bit)
3672 159285 unhandled syscall:25 (stime, on x86-linux)
3673 159452 unhandled ioctl 0x8B01 on "valgrind iwconfig"
3674 160954 ppc build of valgrind crashes with illegal instruction (isel)
3675 160956 mallinfo implementation, w/ patch
3676 162092 Valgrind fails to start gnome-system-monitor
3677 162819 malloc_free_fill test doesn't pass on glibc2.8 x86
3678 163794 assertion failure with "--track-origins=yes"
3679 163933 sigcontext.err and .trapno must be set together
3680 163955 remove constraint !(--db-attach=yes && --trace-children=yes)
3681 164476 Missing kernel module loading system calls
3682 164669 SVN regression: mmap() drops posix file locks
3683 166581 Callgrind output corruption when program forks
3684 167288 Patch file for missing system calls on Cell BE
3685 168943 unsupported scas instruction pentium
3686 171645 Unrecognised instruction (MOVSD, non-binutils encoding)
3687 172417 x86->IR: 0x82 ...
3688 172563 amd64->IR: 0xD9 0xF5 - fprem1
3689 173099 .lds linker script generation error
3690 173177 [x86_64] syscalls: 125/126/179 (capget/capset/quotactl)
3691 173751 amd64->IR: 0x48 0xF 0x6F 0x45 (even more redundant prefixes)
3693 174908 --log-file value not expanded correctly for core file
3694 175044 Add lookup_dcookie for amd64
3695 175150 x86->IR: 0xF2 0xF 0x11 0xC1 (movss non-binutils encoding)
3697 Developer-visible changes:
3699 * Valgrind's debug-info reading machinery has been majorly overhauled.
3700 It can now correctly establish the addresses for ELF data symbols,
3701 which is something that has never worked properly before now.
3703 Also, Valgrind can now read DWARF3 type and location information for
3704 stack and global variables. This makes it possible to use the
3705 framework to build tools that rely on knowing the type and locations
3706 of stack and global variables, for example exp-Ptrcheck.
3708 Reading of such information is disabled by default, because most
3709 tools don't need it, and because it is expensive in space and time.
3710 However, you can force Valgrind to read it, using the
3711 --read-var-info=yes flag. Memcheck, Helgrind and DRD are able to
3712 make use of such information, if present, to provide source-level
3713 descriptions of data addresses in the error messages they create.
3715 (3.4.0.RC1: 24 Dec 2008, vex r1878, valgrind r8882).
3716 (3.4.0: 3 Jan 2009, vex r1878, valgrind r8899).